Regioinvent: a regionalized version of ecoinvent integrating detailed trade data

Regioinvent: a regionalized version of ecoinvent integrating detailed trade data | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Regioinvent: a regionalized version of ecoinvent integrating detailed trade data Maxime Agez, Guillaume Majeau-Bettez This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8159063/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 7 You are reading this latest preprint version Abstract Life cycle inventory databases currently offer limited country-level coverage and often rely on outdated supply chain representations in their market processes. In this research, we combine ecoinvent with the BACI database—which provides detailed international trade statistics—to regionalize ecoinvent and model average supply chains based on up-to-date data. To achieve this, we automatically duplicate regional processes and adapt them to specific national contexts by modifying three key inputs in the ecoinvent database: electricity, heat, and municipal solid waste treatment. We also construct national consumption markets based on international trade and domestic production data to capture country-specific supply chain characteristics. These markets are fully relinked throughout the database, and elementary flows are spatialized. The resulting regioinvent adaptation of ecoinvent comprises 4,031 products regionalized across 225 countries, yielding 669,571 newly created processes. We assess the effect of regionalization by comparing each regionalized process with its corresponding original ecoinvent process. On average, regionalization introduces differences of 10.8% for climate change, 13.9% for human health, and 11.6% for ecosystem quality. Spatialization of elementary flows is found to exert a stronger influence on regionalized impact categories than inventory regionalization alone. Future developments of regioinvent will focus on addressing its current limitations, particularly the estimation of domestic production data and the contextualization of transportation within national consumption markets. Despite these limitations, we recommend that LCA practitioners use regioinvent, as it is expected to provide more accurate results than the traditional ecoinvent database. “industrial ecology” “life cycle assessment (LCA)” “regionalization” “tool” “international trade” “ecoinvent” Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. INTRODUCTION Regionalization is a core aspect of conducting any Life Cycle Assessment (LCA) study. In practice, it involves copying an existing process from an LCA database and adapting its inputs to the relevant geographical context (Baitz et al. 2013). However, the term regionalization itself is broad and can refer to several distinct aspects. Patouillard et al. ( 2018 ) distinguished between three components. The regionalization of the inventory, pertaining to the modification of the technosphere inputs/outputs to match the required geographical context (e.g., from "electricity - RER" to "electricity - BE"). The spatialization of the inventory, which consists in specifying a location for the extraction/release of elementary flows (e.g., from "Water" to "Water, BE"). The impact regionalization which is the characterization of a spatialized inventory with regionalized characterization factors (e.g., the characterization factor of "Water, BE" according to the AWARE 2.0 method is 3.56 m³ world-eq., while it is 38.2 m³ world-eq for the global average (Seitfudem et al. 2025 )). Typically, LCA practitioners handle the regionalization and spatialization of inventories, whereas LCIA model developers are responsible for impact regionalization (Patouillard et al. 2019 ). These three dimensions of regionalization have been shown to significantly influence LCA results (Mutel and Hellweg 2009), (Yang and Heijungs 2017), (Hung et al. 2021 ), (Schenker, Oberschelp, and Pfister 2022), (Adrianto, Pfister, and Hellweg 2022), highlighting the necessity for regionalization within LCA studies. There are multiple ways to apply regionalization and spatialization to an LCA inventory. n its simplest form, regionalizing the inventory consists of changing the origin of a commodity or service—for example, replacing a European electricity grid mix with a Belgian one. In more advanced applications, additional data are collected to adjust not only the origin of a commodity but also the quantities and types of its inputs and outputs, or even the production technologies used (Patouillard et al. 2018 ). Examples include the regionalized modeling of wind turbines in Denmark (Sacchi et al. 2019 ) and of milk production in the United States (Henderson et al. 2023 ). For spatialization, the most basic approach is to specify the region or country where an emission or resource extraction occurs. This can be done by creating new elementary flows (e.g., “Water, CA-QC”) or by assigning the location during calculation, thereby avoiding the creation of thousands of additional flows (Sacchi et al. 2025 ). More advanced approaches make use of geographic information systems (GIS) (Mutel 2012 ), to pinpoint emissions or extractions at much finer spatial resolution. Li et al. ( 2021 ) review 105 LCA studies that rely on GIS-based spatialization. Ongoing research also seeks to integrate both regionalization and spatialization within a unified computational framework (Mutel and Hellweg 2023). The application of regionalization in LCA is often restricted to the foreground system—that is, to the data directly collected and modeled by the practitioner. The background database (e.g., ecoinvent) is almost always left unchanged. However, LCA databases themselves offer limited geographical resolution. For example, ecoinvent v3.10.1 (cut-off) covers on average only about 0.85 countries per process, meaning that fewer than one country-specific process is available per activity. Like other LCA databases, ecoinvent therefore relies heavily on broad regional representations such as Europe (RER), Rest of the World (RoW), or Global (GLO). Moreover, each new ecoinvent release updates only a small subset of process models and markets. As a result, most market datasets remain several years—or even decades—out of date (Steubing et al. 2016 ). Given that the global economy evolves continually, the market shares used in ecoinvent are unlikely to reflect current average supply chains. This issue is especially pronounced for energy-related markets (e.g., crude oil), which are highly sensitive to temporal changes. These limitations, shared by all LCA databases, could in principle be addressed, since the simplest forms of regionalization and spatialization of the inventory can be automated and applied across an entire background database. Alaux et al. (2024), for example, developed an algorithm that systematically screens the ecoinvent database and reconnects regional inputs of country-specific processes to national inputs whenever possible. This improves the geographical representativeness of ecoinvent processes to some extent, but it neither corrects the outdated or misrepresented market structures found in LCA databases nor provides spatialization of elementary flows. Sacchi ( 2018 ) by contrast, adapted the consequential version of ecoinvent by generating consumption markets based on trade data and even explored how these markets would respond to marginal changes in product demand. However, this approach was applied only to two products and was not generalized to the full database. Sanyé Mengual et al. ( 2023 ) extended regionalization further by applying it to 164 selected commodities in ecoinvent and Agrifootprint, focusing on adapting energy inputs, waste treatment, feedstock diets, and consumption markets. Yet, their work also does not cover the full set of database commodities, and while the resulting data are open-access, the code used to generate the regionalized dataset is not open-source. Peng & Pfister (2024) released code to generate a regionalized version of ecoinvent using EXIOBASE, a multi-regional input–output database. Their implementation also created consumption markets based on EXIOBASE trade data, covering 44 countries for each commodity and adapting the corresponding ecoinvent activities, while also spatializing water-related elementary flows. Unlike Sacchi ( 2018 ) and Sanyé Mengual et al. ( 2023 ), their approach was applied to the entire ecoinvent database. Although this represents a major step toward fully automating the regionalization of ecoinvent, several limitations remain. Because the approach relies on EXIOBASE, much of the detailed trade information available at the commodity level is lost, as the model is restricted to the 163 industries and 44 countries covered by EXIOBASE. This leads to well-known aggregation issues in input–output analyses (Lenzen 2011 ), where commodities within the same sector cannot be distinguished. For instance, EXIOBASE cannot differentiate between individual chemical products, meaning that all trade data for substances such as propanol or benzene collapse into the sector-level average for “Chemicals.” The reliance on EXIOBASE also automatically excludes all countries not represented among its 44 regions. Another limitation arises from the monetary nature of EXIOBASE trade flows. The authors derived monetary trade shares to build consumption markets and applied these shares to ecoinvent markets, which are expressed in physical units. Since commodity prices vary substantially across countries, this procedure inevitably introduces inconsistencies. Their implementation also has practical constraints, requiring more than 350 GB of RAM—an amount far beyond what most LCA practitioners can access. Finally, as with previous studies generating consumption markets, their approach relies on proxy regions whenever no country-specific production dataset exists. In other words, the market description introduces a regional process (e.g., RER) whenever the national dataset (e.g., AT) is missing. 2. AIM AND SCOPE This article aims at creating --- in a fully automated, reproducible, highly maintainable and iteratively improvable data fusion approach --- a fully regionalized version of the ecoinvent LCA database at the country level. This entails the automatic creation of thousands of regionalized and spatialized inventories of national production processes of commodities and of national consumption markets. This research is looking to rely on the best detailed available data regarding trade and regionalized characterization factors to achieve the most fine-grained possible regionalization of ecoinvent through an automated process. As a result, in order to automatically apply heuristics across the database, this research will adopt the simplified approach to regionalization/spatialization, which means that only the origin of inputs/outputs will be adapted in national production processes (not the quantities) and that we will create new elementary flows for spatialization, which also means no GIS will be involved. This research thus does not aim at generating new processes through additional data collection, but rather through integration of already existing sources of data. In this article, the resulting “regioinvent” adaptation of the ecoinvent database described requires a significant amount of computation resources. However, another goal of this research is to ensure that any LCA practitioner be able to use the fruits of this research. This implies that any typical laptop must be able to generate the resulting "regioinvent" adaptation of ecoinvent within a few hours and that regioinvent must be usable in mainstream LCA software. In the article, we will thus also study the effect of certain strategies to reduce the size of the regioinvent database on the results, in order to reduce required computation resources and ensure the operationalization of the research. Through the comparison of ecoinvent with the resulting regioinvent database, the effect of the regionalization of inventories and of the spatialization of inventories will be estimated on a broad sample. 3. METHODS 3.1 Data sources To perform regionalization as implemented in regioinvent, four components are required: life cycle inventories, trade data, production volumes, and a reference list of elementary flows to be spatialized. In this study, we focus on the ecoinvent database (cutoff v3.10.1), as it provides the most comprehensive coverage of commodities and geographies among existing LCA databases. Additionally, substantial efforts by the ecoinvent team have resulted in detailed geographical differentiation for energy-related processes, such as electricity and heat production, which this research leverages. Nevertheless, the methodology developed here for automated regionalization can be applied to other LCA databases as well. We selected the cutoff system model because it is the most widely used, but regioinvent is also compatible with the APOS (Allocation at Point of Substitution) version. Applying regioinvent to the consequential version of ecoinvent, however, would require the use of marginal trade data, whereas the current implementation relies on average trade data. Similarly, applying regioinvent to prospective versions of ecoinvent—such as those generated through the premise package (Sacchi et al. 2022 )—would require prospective trade data to capture the evolving dynamics of international trade. The trade data used in regioinvent are sourced from the BACI (Base pour l’Analyse du Commerce International) database, which itself is based on the United Nations’ UN COMTRADE dataset. BACI provides detailed records of imports and exports for thousands of commodities traded between virtually all countries worldwide. For domestic production—that is, commodities that do not cross borders and therefore do not appear in BACI—we rely on additional data sources. FAOSTAT is used to obtain production volumes for agricultural commodities. Production volumes for mineral extraction and base metal production are drawn from the British Geological Survey (BGS) and the United States Geological Survey (USGS). For all remaining commodities, we use EXIOBASE to estimate the share of domestic production consumed domestically versus exported. For the list of elementary flows to be spatialized, we rely on three regionalized LCIA methods: IMPACT World+ (IW+), ReCiPe, and Environmental Footprint (EF). These methods are already implemented in mainstream LCA software and are among the most widely used LCIA frameworks. Compared with earlier studies that introduced trade data into ecoinvent, (Peng and Pfister 2024) solely relied on EXIOBASE as a source of data, (Sacchi 2018 ) used the UN COMTRADE, FAOSTAT and USGS databases and (Sanyé Mengual et al. 2023 ) relied on Eurostat data. 3.2 Overall methodology Figure 3 − 1 illustrates the algorithm followed by regioinvent to generate a regionalized version of the ecoinvent database. The colors in the figure correspond to the different components of the methodology: extraction and processing of trade data (yellow), creation of national production processes (orange), creation of national consumption markets and technology mixes (blue), relinking (purple), and spatialization of elementary flows (red). Each of these steps is detailed in the subsequent subsections of the methodology. 3.3 Trade data extraction and treatment (in yellow) Aim of the section : The objective is to determine, for each ecoinvent commodity, the distribution of producing countries within the global production market and within each national consumption market. In other words: Which countries produce commodity A globally? For the commodity A consumed in country B , what share originates from domestic production versus imports? And, if imported, from which countries and in what proportions? The ecoinvent cutoff v3.10.1 database includes 4,031 commodities (products or services) represented through 23,523 processes. Of these commodities, 1,982 are traded internationally, while the remaining 2,049 are not—for example, “building, hall,” “diesel, burned in machinery,” or “metal working, average for aluminium product.” The trade data extraction described in this section concerns only the 1,982 internationally traded commodities. International import and export data for these commodities were extracted from the BACI database using a concordance between ecoinvent product names and the Harmonized System (HS) classification employed by BACI. This concordance was initially generated with the assistance of ChatGPT-4o to accelerate the process, and each mapping was subsequently checked and validated manually by the authors. The final mapping is provided in the Supplementary Information. Trade data were extracted for the five most recent available years (2018–2022) from the HS17 202401b version of BACI and averaged arithmetically over this period. Using a five-year average mitigates the influence of atypical years and reduces the risk of missing data points. We used BACI’s physical data (tonnes of traded products), thereby avoiding inconsistencies that would arise from applying monetary trade ratios to ecoinvent’s physically based flows. The extracted and formatted dataset is available on Zenodo and can be reused in other projects. The BACI database does not distinguish between exports and re-exports. Re-exporting occurs when a country imports a commodity and subsequently exports it without any transformation. As a result, BACI may identify countries as major exporters of goods they do not produce. For example, the Netherlands appears as the 7th largest exporter of bananas, even though bananas do not grow in its climate. Directly using raw export data would therefore introduce substantial inconsistencies. To address this issue, regioinvent relies on net exports rather than raw export values. Net exports are calculated as exports minus imports. When a country imports large quantities of a commodity for the purpose of re-exporting it, its net export value will become zero or negative. Such a condition signals that the country is not a genuine producer or exporter of the commodity but primarily a re-exporter. For all commodities with identified re-exporting behavior, we correct the corresponding import data by redistributing the import volume to the top five global net exporters of that commodity. The selection of the five largest net exporters is arbitrary; however, because the imported quantities are simply reallocated, the global balance between supply and consumption remains conserved. Note that Sanyé Mengual et al. ( 2023 ) proposed a more refined method to correct for re-exports. Instead of reallocating the entire export value, their approach accounts for the share that may originate from domestic production using the national import ratio. For example, if France has a negative net export balance for fridges and imports 25% of the fridges sold domestically, their method assumes that 25% of France’s fridge exports originate from domestic production, while the remaining 75% stem from re-exports. In regioinvent, by contrast, we implicitly assume that 100% of such exports are due to re-exports. The BACI database provides information only on commodities that cross national borders. As a result, commodities that are produced and consumed domestically do not appear in BACI. To the best of the authors’ knowledge, no open-access database offers comprehensive production volume data at a sufficiently detailed level (i.e., comparable to HS6 resolution). We therefore rely on multiple data sources. When total domestic production data are available—as is the case for agricultural commodities (via FAOSTAT) and for mineral extraction and metal production (via BGS and USGS)—we estimate domestically consumed production by subtracting net exports from the reported domestic production volume. When domestic production volumes are not available, we estimate domestic production and consumption using the EXIOBASE database. For each commodity and each EXIOBASE country or region, we compute the ratio of exports to domestic consumption. This ratio is then applied to the BACI net export values to estimate the domestic consumption of domestically produced goods in physical units, following the equation presented below. $$\:{D}_{x,y}={E}_{x,y}(\frac{1}{{R}_{X,Y}}-1)$$ Where: D is the amount (in tonnes) of domestically produced commodity \(\:x\) that is consumed within country \(\:y\) E is the amount (in tonnes) of net exports of commodity \(\:x\) of the country \(\:y\) R is the ratio of domestically consumed vs exported of the sector \(\:X\) to which belongs commodity \(\:x\) in region \(\:Y\) to which belongs country \(\:y\) To illustrate this approach, consider the example of benzene in South Korea. We do not know the country’s benzene production volume directly. However, BACI reports that South Korea exports 630,000 t of benzene, and EXIOBASE indicates that, on average, South Korea exports 54% of its domestic production of chemicals. We can therefore estimate the national benzene production by dividing the net exports by this export share, yielding an estimated domestically consumed production of approximately 536,000 t of benzene. 3.4 Creation of national production processes (in orange) Aim of the section : The goal of this section is to create national production processes, that is, regionalized copies of the corresponding ecoinvent processes. After identifying the commodities to be regionalized, the next step is to determine for which countries each commodity should be regionalized. Peng & Pfister (2024) chose to regionalize every activity for all 44 EXIOBASE countries. However, this produces a large number of irrelevant datasets—such as a “production of banana” process for Canada, which is implausible given the country’s climate. A more pragmatic approach is to determine the relevant countries commodity by commodity, based on the major global producers. For internationally traded commodities, we therefore rely on the production data (i.e., domestic production plus net exports, as determined in the previous section) to identify which countries should be included in the regionalization. Countries are added in descending order of their production volumes until their cumulative production reaches 99% of the global total for that commodity. All remaining countries are aggregated into a Rest-of-the-World (RoW) region. For example, if the global production of commodity A is distributed as follows—China (55%), United States (24.5%), Australia (12.5%), Chile (7.5%), Guatemala (0.3%), and Spain (0.2%)—then regionalized production processes would be created for China, the United States, Australia, and Chile. Guatemala and Spain would be grouped into a RoW process. For commodities that are not traded internationally, this country-selection method cannot be applied. In these cases, we generate regionalized versions of each process for all countries covered by the BACI database (225 countries). For each selected commodity–country combination, we create a regionalized production process by copying the ecoinvent process corresponding to the region in which that country is located (e.g., “RER” for Germany, “RME” for Lebanon). No assumptions are made regarding technological equivalency between countries. Thus, if a country does not have its own dataset in ecoinvent, we use the broader regional dataset rather than selecting another country with potentially similar technology (e.g., we do not model Germany based on France, but on the RER region). To adapt this regional process to the specific country, three key types of inputs are systematically replaced: electricity, heat, and municipal solid waste treatment. These inputs are substituted with their country-specific counterparts in each regionalized process. They were selected because they appear in virtually all ecoinvent processes and are already provided with detailed geographical differentiation in ecoinvent. When multiple production technologies exist for the same commodity, we generate a regionalized copy for each technology in every relevant country. For example, 1-butanol in ecoinvent can be produced either via hydroformylation of propylene or via a Fischer–Tropsch process. Regardless of which technology is used in a given country in reality, regioinvent creates a regionalized version of each technology for each country. 3.5 Creation of national consumption markets and technology mixes (in blue) Aim of the section : The objective is to create national consumption market processes, i.e., processes that represent the average origin of a commodity purchased within a given country. As with the creation of national production processes, there is no need to create consumption markets for countries that do not meaningfully consume a given commodity. Countries included in the regionalization are therefore selected based on their consumption values, sorted in descending order, until their cumulative consumption reaches 99% of global consumption for that commodity. Regarding commodities produced through multiple technologies, no available dataset provides trade information disaggregated by technology. In the absence of such detail, we assign technologies according to their global market shares in ecoinvent. For example, in the case of 1-butanol, the hydroformylation of propylene accounts for 94.5% of the global market in ecoinvent, while the Fischer–Tropsch route accounts for 5.5%. Consequently, every national consumption market for 1-butanol adopts this 94.5/5.5 technology split. Transportation inputs are copied directly from the corresponding global market process in ecoinvent. For commodities that are not traded internationally, we create “technology mix” processes to represent the distribution of production technologies. These mixes simply reuse the technology shares already present in ecoinvent. 3.6 Relinking (in purple) We then relink all created processes—both national production processes and national consumption markets—to one another within regioinvent. This ensures, for example, that the newly created Algerian brass production process uses Algerian consumption markets for zinc and copper, rather than relying on global markets. In addition, we relink all created processes to the original ecoinvent datasets in order to improve the regionalization of the three key inputs central to our approach (electricity, heat, and waste). As a result, for instance, the Austrian electricity-from-coal process now draws on the Austrian consumption market for coal instead of using the European market. 3.7 Spatialization of elementary flows (in red) For spatialization, regioinvent applies a straightforward rule: each elementary flow is assigned the geography of the process that emits or extracts it. In other words, if 1-butanol is produced in Belgium, all associated water use and water emissions are assumed to occur in Belgium. This assumption is already widely used in LCA practice, including in SimaPro and, previously, in openLCA. To determine which elementary flows should be spatialized, we compiled the spatialized flows from three regionalized LCIA methods—IMPACT World+, ReCiPe, and Environmental Footprint (EF)—at both midpoint and endpoint/damage levels. Consequently, the following categories of elementary flows were spatialized: water, land, acidification, eutrophication, particulate matter formation, and photochemical ozone formation. Spatialization was applied to both regioinvent processes and the original ecoinvent processes. 3.8 Operationalization of regioinvent The version of regioinvent used in this article consists of 669,571 newly created processes and 110,559 newly spatialized elementary flows, in addition to the 23,523 original processes and 4,362 elementary flows in ecoinvent 3.10.1 (cutoff). Generating this full database requires substantial computational resources—approximately 10 hours of processing time and 128 GB of RAM. Once generated, running a simple cradle-to-gate LCA on an ecoinvent process requires a machine with at least 64 GB of RAM and takes roughly 30 minutes on a 2.5 GHz CPU. This full version, while appropriate for research purposes, is not practical for everyday LCA use in unit-process format. When converted to system processes, computational requirements decrease significantly and become manageable on typical laptops, but this conversion also entails a loss of information. To enable the use of regioinvent in unit-process form, we therefore introduce several rules and parameters to reduce the size of the database, thereby lowering both the computational burden of generating it and the resources needed to run LCAs with it. First, we reduce the number of non-internationally traded commodities to be regionalized. Instead of regionalizing all 2,049 such commodities, we focus only on those that are most relevant. To identify which commodities should be included, we perform an LCA of every process in ecoinvent 3.10.1 cutoff (excluding markets, i.e., 14,975 processes) and analyze the contribution of each of the 2,049 commodities across all unit processes and across the two areas of protection in the IMPACT World+ method: Human Health and Ecosystem Quality. If, in any of the 14,975 processes and for either area of protection, a commodity contributes more than 0.3% to the total impact, it is classified as relevant. If a commodity never exceeds this 0.3% contribution threshold, it is deemed irrelevant and is no longer regionalized. The 0.3% threshold is arbitrary: a 1% threshold resulted in too few relevant commodities, while a 0.1% threshold produced too many. At 0.3%, only 67 of the 2,049 commodities qualify as relevant and are therefore regionalized. This reduction brings the database size down to 228,013 processes, making it feasible to generate and run calculations on a typical machine with 16 GB of RAM. We also introduce a cut-off threshold for determining which countries are included in the regionalization of the 1,982 internationally traded commodities. This corresponds to the 99% cumulative coverage threshold described in sections 3.4 & 3.5. Users may choose to lower this threshold, thereby reducing the number of countries for which each commodity is regionalized. For example, selecting a 90% cut-off reduces the database size to 95,871 processes, while a 75% cut-off results in a database of 51,327 processes. With a 75% threshold, an LCA performed with regioinvent in brightway2 takes approximately 30 seconds for the first run and less than 10 seconds for subsequent runs on a standard laptop with 16 GB RAM and a 4-core 3.3 GHz CPU. At the time of writing, regioinvent operates exclusively within the brightway framework. Importers for SimaPro and openLCA are planned, but it remains uncertain to what extent these software platforms will be able to handle datasets of this size. 3.9 Data availability Regioinvent is a completely open-source and free project that anyone can install and use, provided they have access to an ecoinvent license. The version used in this article is the v1.3.0 (Agez 2025a) 1 and operates with BACI HS17 202401b data (Gaulier and Zignago 2010) that was already extracted and formatted for use with regioinvent (Agez 2025b). We relied on production volumes from FAOSTAT (FAOSTAT 2024) and from BGS/USGS which were previously extracted in (Bucciarelli, Hache, and Mignon 2025).The v3.9.5 of EXIOBASE (Stadler et al. 2021) is used to determine export/production ratios. Regioinvent currently only supports the regionalization of the versions 3.9/1 and 3.10/1 cut-off of ecoinvent (Wernet et al. 2016). It works alongside the brightway2 LCA software (Mutel 2017) and relies on the wurst Python package 2 , to speed up the generation of the database. A specific fully regionalized version of IMPACT World+ v2.1 (Agez et al. 2024) is used for characterization in this research, but fully regionalized versions of ReCiPe 2016 v1.03 (H) and EF3.1 are also available within the regioinvent data package. 4. RESULTS AND DISCUSSION 4.1 Database description In this results section, we first illustrate the level of interpretability that regioinvent offers to LCA practitioners. For this purpose, we use “nitrous dioxide” as an example, as it provides a clear and relevant system for comparison. Following this illustrative example, the remainder of the results section examines: (i) the effects of regioinvent on the database as a whole, (ii) the relative influence of spatializing elementary flows versus regionalizing technosphere flows, and (iii) the impact of the chosen cut-off threshold on the results. A key contribution of regioinvent is the creation of consumption markets for each commodity—that is, markets representing the distribution of the countries of origin for a commodity purchased within a given country. Figure 4 − 1 displays the consumption market for nitrous dioxide in Morocco. It shows that Morocco primarily imports nitrous dioxide from Spain (28%), followed by China (13%), Sweden (11%), and other countries. All production processes referenced within the national consumption markets of regioinvent are themselves regionalized. Figure 4 − 2 illustrates such a regionalized production process for Spain. As shown, the electricity input is regionalized by substituting the corresponding country-specific process from the original ecoinvent database. In addition, existing consumption markets for other commodities—such as liquid oxygen or nitrous oxide—are used to further adapt the process to the Spanish context. Non-internationally traded commodities are also regionalized for Spain. For instance, the “chemical factory” input is regionalized, which means it ultimately relies on Spanish cement and other Spain-specific upstream processes. Relevant elementary flows are likewise spatialized to Spain, including direct emissions of nitrogen oxides and water-related flows. Note that numerical values have been hidden in the figure for licensing reasons; however, the underlying values remain identical to those in the original ecoinvent process. The original ecoinvent database is also modified so that its processes rely on the corresponding regioinvent processes whenever these are available and relevant. For example, Fig. 4 − 3 shows that nuclear electricity produced in Spain now draws on the Spanish technology mix for decarbonized water, as well as on the Spanish consumption market for liquid nitrogen. In addition, its water extractions are spatialized to Spain. 4.2 Results analysis for nitrous dioxide production processes To illustrate how regioinvent affects LCA results, the next two figures present outcomes for two impact categories and their main contributors using the IMPACT World + v2.1 LCIA method. The comparison includes the 22 regionalized production processes created in regioinvent for nitrous dioxide, along with the two original (i.e., non-regionalized and non-spatialized) ecoinvent production processes. Results are normalized to the RoW production process of ecoinvent, which serves as the reference (left side of the graph). For the climate change indicator (Fig. 4 – 4 ), the regioinvent processes range from 52% (Germany) to 95% (China) of the reference ecoinvent value. The climate change impact of nitrous dioxide production is primarily driven by nitrous oxide production (blue) and oxygen production (orange). Their contributions vary across all regionalized processes because each country relies on different consumption markets for these inputs. Further down the supply chain, most greenhouse gas emissions stem from the electricity used in air separation to produce oxygen—an input that is also required for the production of nitrous oxide. Germany shows the lowest climate change impact because its nitrous oxide supply is predominantly domestic and, in turn, relies on oxygen that is mostly imported from France, where electricity has a relatively low carbon intensity. At the opposite end, China relies entirely on domestic nitrous oxide and oxygen production, which means that Chinese electricity—typically more carbon intensive—is used for air separation, driving the higher impact. An interesting observation is that none of the 22 regionalized production processes exceeds the climate change impact of the original ecoinvent RoW process. This is largely because the RoW liquid oxygen dataset in ecoinvent relies heavily on the RAS electricity market group. The RAS market mixes electricity based on total electricity production shares in Asia rather than on production shares specific to liquid oxygen, resulting in an inflated impact for the RoW dataset. For the water scarcity indicator (Fig. 4 –5), the influence of regioinvent is more pronounced—an expected outcome given that water scarcity is a regionalized indicator. Results range from 25% (Finland) to 147% (Iran) of the reference ecoinvent value. Most water consumption occurs during the production of oxygen, either directly for the nitrous dioxide process (orange) or indirectly through the production of nitrous oxide (blue). The process itself also consumes substantial amounts of water directly (purple). In Finland, the oxygen supply ultimately comes from Finland and Sweden, both of which have very low water scarcity characterization factors (2.2 and 2.4 m³ world-eq). In contrast, for Iran, oxygen is produced domestically, where water scarcity is high (70 m³ world-eq), explaining the large differences observed. A notable aspect of this figure is that the two ecoinvent processes show nearly identical results. This occurs because ecoinvent does not, by default, spatialize water flows. Even if water flows were spatialized—as implemented in SimaPro, for example—the resulting datasets would apply RoW and RER characterization factors (e.g., from the AWARE method), which are very similar (39.5 and 40.5 m³ world-eq). Therefore, spatialization without the creation of national production processes and consumption markets would not substantially improve the accuracy of water scarcity assessments for nitrous dioxide production. Overall, these results show that regioinvent, through its systemic regionalization approach, enables a much deeper interpretation of LCA outcomes and provides a more accurate representation of average supply chains. 4.3 Whole database results analysis 4.3.1 The effect of regioinvent Regioinvent adds 669,571 new processes, covering 4,031 commodities across 225 countries. What is the effect of these additions on LCA results? To assess this, we compute impacts for all regioinvent production processes of internationally traded commodities (62,423 processes) and compare them with the corresponding original ecoinvent production processes—specifically, those for which a GLO or RoW dataset is available to serve as a reference. For each product/technology combination, we evaluate the spread of the regioinvent results relative to the ecoinvent reference value. To do so we calculate the relative standard deviation of each regioinvent process with respect to its corresponding ecoinvent reference. Outliers are removed using the interquartile range (IQR) criterion, retaining only observations within \(\:\left[{Q}_{1}-1.5\cdot\:\text{I}\text{Q}\text{R},\text{\hspace{0.17em}}{Q}_{3}+1.5\cdot\:\text{I}\text{Q}\text{R}\right]\) , where \(\:{Q}_{1}\) and \(\:{Q}_{3}\) are the first and third quartiles. A simple arithmetic mean of the remaining values is then computed to obtain an overall relative standard deviation for each impact category. This value reflects the typical spread in results introduced by regioinvent. To separate the influence of regionalization of technosphere flows from the influence of spatialization of elementary flows, we also generate a version of regioinvent without spatialization. Applying the same relative standard deviation calculation and subtracting the resulting values from those of the fully spatialized version allows us to quantify the distinct contributions of inventory regionalization and inventory spatialization across a large sample of data. Table 4 − 1 presents these differences for 34 IMPACT World + indicators; the full table (53 indicators) is provided in the Supplementary Information. On average, regioinvent introduces a spread of 10.8% for the climate change (GWP100) indicator, 13.9% for total human health, and 11.6% for total ecosystem quality. The latter two correspond to the aggregated damage indicators of IMPACT World+. These values therefore represent the average effect of applying regioinvent. Some indicators show very limited sensitivity to regioinvent—notably fisheries impacts, mineral resource use, and short-term toxicity and ecotoxicity indicators. Others are strongly affected, particularly those that are inherently regionalized: freshwater and terrestrial acidification, freshwater eutrophication, particulate matter formation, and water-related indicators. The results also show that spatialization of elementary flows generally has a larger influence on the outcomes than the regionalization of technosphere flows. Exceptions include land occupation, biodiversity, water scarcity (AWARE 2.0), and water availability (human health). These exceptions can be explained by the fact that regionalizing the inventory modifies electricity mixes throughout entire supply chains. This in turn alters the use of hydroelectricity, leading to substantial differences in associated water and land consumption. For the water availability (human health) indicator in particular, spatialization even reduces the spread. This is likely because non-spatialized water flows receive a positive characterization factor (i.e., they contribute to impact), whereas spatialized water flows for most developed countries receive a zero characterization factor in this category. These countries are assumed to be capable of mitigating water shortages through technological or economic means, such that water availability does not significantly affect human health. Finally, ionizing radiation results are also strongly affected. This is due both to the small absolute magnitude of the indicator and to the regionalization of electricity mixes along supply chains, which may result in greater reliance on nuclear energy for some countries. It is important to note that the spatialization effects reported in Table 4 − 1 are in addition to trade-based regionalization; they do not represent the effect of applying spatialization directly to the original ecoinvent database. Sector-level results (based on ISIC classifications) are provided in the Supplementary Information. ISIC sectors with fewer than 10 corresponding ecoinvent processes were excluded. These sectoral results can help identify where regionalization efforts should be prioritized. For example, processes related to crop and animal production are much more strongly affected by regionalization than processes related to repair and maintenance or waste treatment. Unfortunately, a direct comparison with previous studies that regionalized ecoinvent is not possible. Peng and Pfister (2024), despite spatializing water flows, applied only the global AWARE characterization factor (except in their wheat case study), and their workflow cannot be replicated here due to its high computational requirements. Sacchi ( 2018 ) did not spatialized elementary flows. For Sanyé Mengual et al. ( 2023 ), it is unclear whether elementary flows were spatialized, and their database is not available in brightway, preventing replication and comparison. Table 4-1 Median differences of minimum and maximum of regioinvent processes compared to a reference ecoinvent process for 34 IMPACT World + v2.1 indicators. Trade regionalization Spatialization Total regionalization Climate change 10.8% - 10.8% Fisheries impact 0.0% - 0.0% Fossil and nuclear energy use 8.1% - 8.1% Freshwater acidification 13.7% 68.7% 82.4% Freshwater ecotoxicity, long term 10.1% - 10.1% Freshwater ecotoxicity, short term 5.3% - 5.3% Freshwater eutrophication 8.2% 25.6% 33.7% Human toxicity cancer, long term 9.9% - 9.9% Human toxicity cancer, short term 3.4% - 3.4% Human toxicity non-cancer, long term 9.2% - 9.2% Human toxicity non-cancer, short term 5.2% - 5.2% Ionizing radiations, ecosystem quality 55.1% - 55.1% Ionizing radiations, human health 58.4% - 58.4% Land occupation, biodiversity 12.2% 8.4% 20.6% Land transformation, biodiversity 6.0% 11.5% 17.5% Marine acidification 11.2% - 11.2% Marine ecotoxicity, long term 9.6% - 9.6% Marine ecotoxicity, short term 4.1% - 4.1% Marine eutrophication 8.8% 12.7% 21.5% Mineral resources use 18.6% - 18.6% Ozone layer depletion 8.3% - 8.3% Particulate matter formation 13.1% 27.1% 40.2% Photochemical ozone formation, ecosystem quality 10.9% - 10.9% Photochemical ozone formation, human health 11.1% - 11.1% Terrestrial acidification 13.0% 24.1% 37.1% Terrestrial ecotoxicity, long term 9.6% - 9.6% Terrestrial ecotoxicity, short term 4.1% - 4.1% Thermally polluted water 35.3% - 35.3% Water availability, freshwater ecosystem 20.0% - 20.0% Water availability, human health 20.0% -2.9% 17.1% Water availability, terrestrial ecosystem 14.5% 19.6% 34.1% Water scarcity 20.0% 8.7% 28.7% Total human health 9.0% 4.8% 13.8% Total ecosystem quality 7.2% 4.4% 11.6% 4.3.2 The effect of the cutoff of regioinvent As mentioned in section 3.8, the regioinvent database generated for this article is too large for most current machines to handle efficiently. We therefore introduced several simplifications to enable its use on typical computers. These simplifications introduce an aggregation bias at the country level. To assess the magnitude of this bias, we evaluate four different “versions” of regioinvent: Full version: the version used throughout the article, in which all eligible processes are regionalized and the country coverage cut-off is set at 99% regio99: a version with the same 99% cut-off, but in which only the most relevant non-internationally traded commodities are regionalized. regio90: a version in which the country coverage cut-off is reduced to 90%. regio75: a version in which the cut-off is reduced to 75%. We generated these versions and calculated impacts for all processes in each. We then compared results only for the processes that appear in all versions, yielding a sample of 15,154 common processes. For each of these processes, we calculated the difference between the full version—which we assume provides the most accurate results—and each of the three reduced versions. Reducing the regionalization scope from the full version to regio99 results in approximately a 0.5% difference in the two areas of protection (Human Health and Ecosystem Quality). Moving to regio90 introduces a 1.5–2% difference, and reducing further to regio75 leads to a 2–3% difference. These results indicate that using smaller versions of regioinvent does not substantially compromise overall accuracy. However, some midpoint categories exhibit substantially larger deviations—most notably freshwater eutrophication and water availability, with differences of roughly 30% and 50%, respectively. The Supplementary Information provides detailed comparisons for all IMPACT World + indicators. 4.3.3 The case of regionalized water processes Spatialization of water flows is already implemented in mainstream LCA software such as SimaPro. However, the way this spatialization is currently carried out can introduce inconsistencies because the geographies of water extraction and water release do not always match. These inconsistencies reduce the robustness of the assessments. For example, in ecoinvent the production of apples in Chile uses the process “market for irrigation, RoW.” When spatialization is applied, this process extracts water from RoW (characterization factor: 39.5 m³ world-eq). This irrigation water is then used within the Chilean apple production process, where it is released as Chilean water (88.1 m³ world-eq). Under these conditions, the production of apples in Chile appears environmentally beneficial, which is clearly an artefact of inconsistent spatialization. In regioinvent, these inconsistencies are almost entirely eliminated. Regioinvent generates thousands of nationally contextualized processes and systematically relinks them throughout the database. As a result, Chilean apple production now relies on a Chilean irrigation process that extracts Chilean water, rather than water from a Rest-of-the-World aggregate. This ensures that mismatches in water geographies do not occur. One exception remains: RoW national production processes in regioinvent can still exhibit inconsistencies. This is because an “apple production RoW” process represents a specific set of countries that does not perfectly match the set of countries represented by an “irrigation RoW” process. In such cases, the geography of an input (e.g., irrigation) may still differ from the geography represented by the process itself. 4.4 Limitations and outlook While regioinvent represents a significant advancement in the representation of supply chains within LCA, its current implementation still has several limitations, which also represent clear avenues for future improvement. Although the uncertainty introduced by each limitation is difficult to quantify, all contribute some degree of uncertainty to the results produced by regioinvent. Nevertheless, we believe that the models generated by regioinvent still improve overall accuracy compared with the original ecoinvent processes, as they incorporate substantial additional information that more finely resolves supply chain structures. As specified in section 3.3, our approach relies on a combination of actual production volume data and estimates derived by applying export/domestic consumption ratios from EXIOBASE to BACI net-export values. This estimation method is highly dependent on the EXIOBASE ratios and is therefore subject to aggregation bias—both in terms of commodities (EXIOBASE includes only ~ 200 commodity groups) and geography (49 regions). Moreover, EXIOBASE ratios are expressed in monetary units (euros), which we then apply to physical quantities (tonnes). This creates inconsistencies because commodity prices vary significantly depending on origin. These issues mirror the limitations identified in Peng and Pfister (2024) although in our case they affect only domestic production estimate. Most importantly, this estimation approach fails when countries produce a commodity exclusively for domestic consumption and do not export it. For example, if China produces benzene but consumes it entirely domestically, the method would incorrectly infer that China produces no benzene, simply because it exports none. Correcting this issue requires supplementing the workflow with additional production volume data. However, integrating comprehensive production statistics for thousands of commodities across all countries represents a substantial effort and remains an open challenge. Whenever multiple technologies exist for producing a given commodity, we distribute these technologies uniformly across all national consumption markets, using the global technology shares provided in ecoinvent. However, this approach introduces limitations: outdated technologies used in certain regions are inadvertently attributed to countries that predominantly rely on more advanced technologies. For example, due to this approximation, European countries are modeled in regioinvent as producing a portion of their butanol via Fischer–Tropsch synthesis, even though this technology is largely obsolete in Europe. Addressing this limitation would require extensive expert knowledge for the many commodities in ecoinvent that have multiple production technologies. Another major area for improvement concerns the contextualization of transportation. In its current form, regioinvent simply copies the transport distances and modes from the global market dataset of each commodity in ecoinvent. This leads to unrealistic transport patterns—for example, sea tanker transport between Switzerland and Germany. Ideally, transport distances and modes should be adapted for each national consumption market based on the actual origin-destination pairs, as done at the European scale in (Sanyé Mengual et al. 2023 ). The automated approach of (Fry et al. 2024 ) could thus be adapted and integrated in regioinvent in the future to correct these inconsistencies. Centralized trade databases such as BACI provide only national-level information and do not distinguish between sub-national regions (e.g., states or provinces). However, finer-scale data are often available from national statistics agencies—for instance, Statistics Canada provides international merchandise trade information for each Canadian province. Incorporating such sub-national data into regioinvent would refine regionalization for large countries with substantial internal variation in production technologies, such as Canada, the United States or China. Finally, BACI provides annual time-series data, enabling the tracking of year-to-year fluctuations in trade flows. These temporal dynamics could be incorporated into regioinvent, either through regular annual updates of the trade data or through less frequent updates that include interannual variability as an explicit source of uncertainty within the datasets. 5. RECOMMENDATIONS & CONCLUSION For LCA professionals, we recommend using regioinvent rather than relying solely on ecoinvent, as regioinvent provides a more accurate representation of supply chains and consistently spatializes elementary flows across multiple impact categories. When modeling with regioinvent, users should distinguish between the three types of processes it provides: national production processes, national consumption markets, and global production markets (the latter are provided as default options). If the country of production is known, the corresponding national production process should be used. If the country of production is unknown but the country of purchase is known, the national consumption market should be selected, as it represents the average origin of the commodity sold in that country. If neither production nor purchase country is known, users can rely on the global production markets provided in regioinvent. For policy-makers, regioinvent allows rapid generation of country-specific supply chains, effectively creating a tailored version of ecoinvent. This is especially valuable for developing countries, where dedicated datasets and studies are often scarce. Using regioinvent’s supply chains can therefore enhance the modeling and evaluation of policies. For LCA database providers, regioinvent offers an opportunity to strengthen the completeness and realism of their databases. It enables more accurate modeling of supply chains through explicit representation of international trade. Because BACI trade data are updated annually, consumption markets can also be refreshed automatically each year. In addition, our work highlights an increasing need for accessible production volume data to improve market modeling. For LCA software providers, there is a growing need to rethink how large-scale LCA data are handled. Many current LCA tools cannot accommodate regioinvent because of its size. While mainstream LCA databases are currently limited to roughly 25,000 processes, future datasets will inevitably become much larger. If this expansion is not driven by database providers, it will come from practitioners themselves, as LCA increasingly moves toward big-data applications. Most existing LCA software is not yet equipped for this future. Regioinvent is a regionalized version of ecoinvent in which supply chains are modeled using international trade data from the BACI database and elementary flows are spatialized. The resulting database contains 669,571 processes and 110,559 spatialized elementary flows. Comparing regioinvent with the corresponding ecoinvent processes shows that regionalization introduces average differences of 10.8%, 13.9%, and 11.6% for climate change, total human health, and total ecosystem quality indicators, respectively. This research also quantifies, for the first time, the specific contribution of spatialization relative to inventory regionalization, on a large sample of data. Regioinvent builds on recent efforts to regionalize the ecoinvent database and provides a substantial improvement over current practice. Nevertheless, important work remains—such as consolidating domestic production data and improving the regionalization of commodity distributions. Furthermore, significant updates will be necessary from LCA software developers to ensure that large-scale datasets like regioinvent can be handled efficiently. Declarations Conflict of Interest Statement: The authors declare no conflict of interest. Author Contribution M. Agez conceptualized and coded regioinvent, and wrote the main manuscript text. G. Majeau-Bettez contributed to the ideation of some aspects. All authors reviewed the manuscript. Acknowledgement The authors would like to thank Titouan Greffe for his assistance in collecting production volume data for minerals and metals and Matthieu Souttre for code testing and the interesting discussions. Data Availability The data that support the findings of this study are openly available in Zenodo at [https://doi.org/10.5281/zenodo.15474318](https:/doi.org/10.5281/zenodo.15474318) , “Trade data from BACI to-be-used with regioinvent” v4. The code and all other required data are available on Github at [https://github.com/CIRAIG/Regioinvent](https:/github.com/CIRAIG/Regioinvent) , v1.3.0. Supporting Information Supporting information is linked to this article on the JIE website: Supporting Information S1: This supporting information provides the data plotted in Figure 4-4 and Figure 4-5, the full results for Table 4-1 and the table for the effect of cutoff on regioinvent. References Adrianto, Lugas Raka, Stephan Pfister, and Stefanie Hellweg. 2022. “Regionalized Life Cycle Inventories of Global Sulfidic Copper Tailings.” Environmental Science & Technology 56(7):4553–64. doi:10.1021/acs.est.1c01786. Agez, Maxime. 2025a. “Regioinvent v1.3.0.” Agez, Maxime. 2025b. Trade data from BACI to-be-used with regioinvent. : Version 4. doi:https://doi.org/10.5281/zenodo.11583814. Agez, Maxime, Elliot Muller, Cécile Bulle, Laura Debarre, Georg Seitfudem, Ivan Viveros Santos, and Lisa Duval. 2024. IMPACT World+ / a globally regionalized method for life cycle impact assessment : Version 2.1. doi:https://doi.org/10.5281/zenodo.14041258. Alaux, Nicolas, Marcella Ruschi Mendes Saade, and Alexander Passer. 2024. “Inventory Regionalization of Background Data: Influence on Building Life Cycle Assessment and Carbon Reduction Strategies.” Journal of Cleaner Production 459:142434. doi:10.1016/j.jclepro.2024.142434. Baitz, Martin, Stefan Albrecht, Eloise Brauner, Clare Broadbent, Guy Castellan, Pierre Conrath, James Fava, Matthias Finkbeiner, Matthias Fischer, Pere Fullana I Palmer, Stephan Krinke, Christian Leroy, Oliver Loebel, Phil McKeown, Ivo Mersiowsky, Bernhard Möginger, Marcus Pfaadt, Gerald Rebitzer, Elmar Rother, Klaus Ruhland, Aafko Schanssema, and Ladji Tikana. 2013. “LCA’s Theory and Practice: Like Ebony and Ivory Living in Perfect Harmony?” The International Journal of Life Cycle Assessment 18(1):5–13. doi:10.1007/s11367-012-0476-x. Bucciarelli, Pauline, Emmanuel Hache, and Valérie Mignon. 2025. “Evaluating Criticality of Strategic Metals: Are the Herfindahl–Hirschman Index and Usual Concentration Thresholds Still Relevant?” Energy Economics 143:108208. doi:10.1016/j.eneco.2025.108208. FAOSTAT. 2024. “FAOSTAT.” https://www.fao.org/faostat. Fry, Jacob, Keiichiro Kanemoto, Alastair Fraser, and Keisuke Nansai. 2024. “Global Freight Transport Emissions Responsibility.” Environmental Science & Technology 58(43):19231–42. doi:10.1021/acs.est.4c05658. Gaulier, Guillaume, and Soledad Zignago. 2010. “BACI: International Trade Database at the Product-Level (the 1994-2007 Version).” SSRN Electronic Journal . doi:10.2139/ssrn.1994500. Henderson, Andrew D., Anne Asselin-Balençon, Martin C. Heller, Jasmina Burek, Daesoo Kim, Lindsay Lessard, Manuele Margni, Rosie Saad, Marty D. Matlock, Greg Thoma, Ying Wang, and Olivier Jolliet. 2023. “Spatialized Life Cycle Assessment of Fluid Milk Production and Consumption in the United States.” Sustainability 15(3):1890. doi:10.3390/su15031890. Hung, Christine Roxanne, Steve Völler, Maxime Agez, Guillaume Majeau-Bettez, and Anders Hammer Strømman. 2021. “Regionalized Climate Footprints of Battery Electric Vehicles in Europe.” Journal of Cleaner Production 322:129052. doi:10.1016/j.jclepro.2021.129052. Lenzen, Manfred. 2011. “AGGREGATION VERSUS DISAGGREGATION IN INPUT–OUTPUT ANALYSIS OF THE ENVIRONMENT.” Economic Systems Research 23(1):73–89. doi:10.1080/09535314.2010.548793. Li, Junjie, Yajun Tian, Yueling Zhang, and Kechang Xie. 2021. “Spatializing Environmental Footprint by Integrating Geographic Information System into Life Cycle Assessment: A Review and Practice Recommendations.” Journal of Cleaner Production 323:129113. doi:10.1016/j.jclepro.2021.129113. Mutel, Chris. 2017. “Brightway: An Open Source Framework for Life Cycle Assessment.” The Journal of Open Source Software 2(12):236. doi:10.21105/joss.00236. Mutel, Chris, and Stefanie Hellweg. 2023. “Matrix-Based Methods for Regionalized Life Cycle Assessment.” Mutel, Christopher. 2012. “Framework and Tools for Regionalization in Life Cycle Assessment.” University of Iowa. Mutel, Christopher L., and Stefanie Hellweg. 2009. “Regionalized Life Cycle Assessment: Computational Methodology and Application to Inventory Databases.” Environmental Science & Technology 43(15):5797–5803. doi:10.1021/es803002j. Patouillard, Laure, Cécile Bulle, Cécile Querleu, Dominique Maxime, Philippe Osset, and Manuele Margni. 2018. “Critical Review and Practical Recommendations to Integrate the Spatial Dimension into Life Cycle Assessment.” Journal of Cleaner Production 177:398–412. doi:10.1016/j.jclepro.2017.12.192. Patouillard, Laure, Pierre Collet, Pascal Lesage, Pablo Tirado Seco, Cécile Bulle, and Manuele Margni. 2019. “Prioritizing Regionalization Efforts in Life Cycle Assessment through Global Sensitivity Analysis: A Sector Meta-Analysis Based on Ecoinvent V3.” The International Journal of Life Cycle Assessment 24(12):2238–54. doi:10.1007/s11367-019-01635-5. Peng, Sidi, and Stephan Pfister. 2024. “Regionalizing the Supply Chain in Process Life Cycle Inventory with Multiregional Input–Output Data: An Implementation for Ecoinvent with EXIOBASE.” Journal of Industrial Ecology 28(4):680–94. doi:10.1111/jiec.13491. Sacchi, R., T. Terlouw, K. Siala, A. Dirnaichner, C. Bauer, B. Cox, C. Mutel, V. Daioglou, and G. Luderer. 2022. “PRospective EnvironMental Impact asSEment (Premise): A Streamlined Approach to Producing Databases for Prospective Life Cycle Assessment Using Integrated Assessment Models.” Renewable and Sustainable Energy Reviews 160:112311. doi:10.1016/j.rser.2022.112311. Sacchi, Romain. 2018. “A Trade-Based Method for Modelling Supply Markets in Consequential LCA Exemplified with Portland Cement and Bananas.” The International Journal of Life Cycle Assessment 23(10):1966–80. doi:10.1007/s11367-017-1423-7. Sacchi, Romain, Romain Besseau, Paula Pérez-López, and Isabelle Blanc. 2019. “Exploring Technologically, Temporally and Geographically-Sensitive Life Cycle Inventories for Wind Turbines: A Parameterized Model for Denmark.” Renewable Energy 132:1238–50. doi:10.1016/j.renene.2018.09.020. Sacchi, Romain, Alvaro Hahn Menacho, Georg Seitfudem, Maxime Agez, Joanna Schlesinger-Martinat, Anish Koyamparambath, Jair Santillan Saldivar, Philippe Loubet, and Christian Bauer. 2025. “Contextual LCIA without the Overhead: An Exchange-Based Framework for Flexible Impact Assessment.” The International Journal of Life Cycle Assessment . doi:10.1007/s11367-025-02551-7. Sanyé Mengual, E., J. Foschi, V. Orza, T. Sinkko, P. Wierzgala, and S. Sala. 2023. Consumption Footprint: Methodological Overview . Joint Research Centre - European Comission. https://dx.doi.org/10.2760/413081. Schenker, Vanessa, Christopher Oberschelp, and Stephan Pfister. 2022. “Regionalized Life Cycle Assessment of Present and Future Lithium Production for Li-Ion Batteries.” Resources, Conservation and Recycling 187:106611. doi:10.1016/j.resconrec.2022.106611. Seitfudem, Georg, Markus Berger, Hannes Müller Schmied, and Anne‐Marie Boulay. 2025. “The Updated and Improved Method for Water Scarcity Impact Assessment in LCA, AWARE2.0.” Journal of Industrial Ecology 29(3):891–907. doi:10.1111/jiec.70023. Stadler, Konstantin, Richard Wood, Tatyana Bulavskaya, Carl-Johan Södersten, Moana Simas, Sarah Schmidt, Arkaitz Usubiaga, Jose Acosta-Fernandez, Jeroen Kuene, Martin Bruckner, Stefan Giljum, Stephan Lutter, Stefano Merciai, Jannick Schmidt, Michaela Theurl, Christoph Plutzar, Thomas Kastner, Nina Eisenmenger, Karl-Heinz Erb, Arjan Koning, and Arnold Tukker. 2021. EXIOBASE 3 - v3.8.2. https://doi.org/10.5281/zenodo.5589597 : Version v3.8.2. doi:https://doi.org/10.5281/zenodo.5589597. Steubing, Bernhard, Gregor Wernet, Jürgen Reinhard, Christian Bauer, and Emilia Moreno-Ruiz. 2016. “The Ecoinvent Database Version 3 (Part II): Analyzing LCA Results and Comparison to Version 2.” The International Journal of Life Cycle Assessment 21(9):1269–81. doi:10.1007/s11367-016-1109-6. Wernet, Gregor, Christian Bauer, Bernhard Steubing, Jürgen Reinhard, Emilia Moreno-Ruiz, and Bo Weidema. 2016. “The Ecoinvent Database Version 3 (Part I): Overview and Methodology.” The International Journal of Life Cycle Assessment 21(9):1218–30. doi:10.1007/s11367-016-1087-8. Yang, Yi, and Reinout Heijungs. 2017. “A Generalized Computational Structure for Regional Life-Cycle Assessment.” The International Journal of Life Cycle Assessment 22(2):213–21. doi:10.1007/s11367-016-1155-0. Footnotes The Github page can be found at https://github.com/CIRAIG/Regioinvent The Github page of the wurst package: https://github.com/polca/wurst Additional Declarations No competing interests reported. Supplementary Files Supplementaryinformationv5.xlsx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 01 Feb, 2026 Reviews received at journal 11 Jan, 2026 Reviewers agreed at journal 02 Dec, 2025 Reviewers invited by journal 25 Nov, 2025 Editor assigned by journal 22 Nov, 2025 Submission checks completed at journal 22 Nov, 2025 First submitted to journal 19 Nov, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8159063","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":550530048,"identity":"f88494c7-990e-4b19-8c0a-39f5b8817560","order_by":0,"name":"Maxime Agez","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+ElEQVRIiWNgGAWjYJCCA1Ca8QEDgwWQZm4gWguzAQODBAMDGyNhLTDAJkGUFt32M4aHblQwyMu7tz+rLqiQSFw7v7GB4UcNbi1mZ9ISDuecYTDceOZA2u0ZZyQStx1jbGDsOYZHy4HkA4dz2xgYN85IOHabtw2ihZmBDY+W8w8bDuf+Y7DfOP9hWzFCyz88Wm6AbGlgSJwvwczGDNfC2IZPyzOgX45JJG/gSWOWBvrFeNuxxIaDvX34HJZj/DmnxsZ2fvvxh58LKmxktx0+fPDBj2+4tUCBBIPBAWBcAlmODQyIyMUP5BsgWuyJUj0KRsEoGAUjCgAAwBVZhthrUsoAAAAASUVORK5CYII=","orcid":"","institution":"Polytechnique Montréal","correspondingAuthor":true,"prefix":"","firstName":"Maxime","middleName":"","lastName":"Agez","suffix":""},{"id":550530049,"identity":"df66d40c-5716-474c-a165-9d110e70e197","order_by":1,"name":"Guillaume Majeau-Bettez","email":"","orcid":"","institution":"Polytechnique Montréal","correspondingAuthor":false,"prefix":"","firstName":"Guillaume","middleName":"","lastName":"Majeau-Bettez","suffix":""}],"badges":[],"createdAt":"2025-11-20 00:08:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8159063/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8159063/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":97161938,"identity":"0a9f6539-84ac-4b82-9833-03df57349f96","added_by":"auto","created_at":"2025-12-01 12:50:44","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11818828,"visible":true,"origin":"","legend":"","description":"","filename":"Regioinventv5manuscript.docx","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/12a1947e9d727d62fe0ed064.docx"},{"id":97248954,"identity":"92218f36-494e-4599-99d6-b74748e48ff2","added_by":"auto","created_at":"2025-12-02 13:08:52","extension":"json","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5013,"visible":true,"origin":"","legend":"","description":"","filename":"6df94dd090a149829b9fb0db87f775fd.json","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/2077eb94159c56dbfe3aea16.json"},{"id":97250028,"identity":"eff263ed-becb-4292-b6c7-9ef31590192c","added_by":"auto","created_at":"2025-12-02 13:13:48","extension":"xlsx","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":112211,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryinformationv5.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/0d19bddc5d3fb434277e918c.xlsx"},{"id":97250217,"identity":"004e92a1-d5a2-41a6-b367-8eb8deddd8e7","added_by":"auto","created_at":"2025-12-02 13:14:06","extension":"xml","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":141961,"visible":true,"origin":"","legend":"","description":"","filename":"6df94dd090a149829b9fb0db87f775fd1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/0b2aa094d0aedfcf9b32f1c6.xml"},{"id":97249939,"identity":"b91429c2-3dbe-4965-8421-b8930673309c","added_by":"auto","created_at":"2025-12-02 13:13:41","extension":"eps","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":801189,"visible":true,"origin":"","legend":"","description":"","filename":"Figure31.eps","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/f87ecbf9b53f07e808fee447.eps"},{"id":97248592,"identity":"a8da2254-7856-4a08-b918-a69af524af21","added_by":"auto","created_at":"2025-12-02 13:03:51","extension":"eps","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":179727,"visible":true,"origin":"","legend":"","description":"","filename":"Figure41.eps","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/10837957087cbfad91733f7a.eps"},{"id":97250164,"identity":"2955d559-22c4-4494-a22c-88ac73f325ae","added_by":"auto","created_at":"2025-12-02 13:14:02","extension":"eps","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":226044,"visible":true,"origin":"","legend":"","description":"","filename":"Figure42.eps","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/270a3fa87fdc8a5820b1c94b.eps"},{"id":97161930,"identity":"f8f6fa9d-4a67-4242-a495-51ea9c0e96b5","added_by":"auto","created_at":"2025-12-01 12:50:44","extension":"eps","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":210936,"visible":true,"origin":"","legend":"","description":"","filename":"Figure43.eps","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/d9d08a0d5d4ee8a17e6c2cd3.eps"},{"id":97161927,"identity":"88cc746a-2e88-40c3-9e34-6e80e860b654","added_by":"auto","created_at":"2025-12-01 12:50:44","extension":"eps","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":36143,"visible":true,"origin":"","legend":"","description":"","filename":"Figure44.eps","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/4837d18bdecf8fe2b399617d.eps"},{"id":97248917,"identity":"2fa2a8cc-cbda-4af3-870b-3baf23c4844f","added_by":"auto","created_at":"2025-12-02 13:08:13","extension":"eps","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":37558,"visible":true,"origin":"","legend":"","description":"","filename":"Figure45.eps","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/63a8f1f131ae1af61a8f8c46.eps"},{"id":97248802,"identity":"a7c903ba-86ce-4ab5-b02d-2a61679ef406","added_by":"auto","created_at":"2025-12-02 13:07:21","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":353465,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/f023b8c9c352f2d539191ae7.png"},{"id":97248713,"identity":"e1b5a3bc-7272-430c-85cc-67f973a44947","added_by":"auto","created_at":"2025-12-02 13:06:20","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":4065029,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/e7c2f97a647e511530a277fa.png"},{"id":97161940,"identity":"40fbe27c-427e-4b6d-8451-d0aa42a60021","added_by":"auto","created_at":"2025-12-01 12:50:45","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2956797,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/a80b5cf4ccb503d1d5c656af.png"},{"id":97249955,"identity":"44460034-3762-493a-b958-14f58fad4358","added_by":"auto","created_at":"2025-12-02 13:13:43","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2841245,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/846233a652ced8b49fc55b6c.png"},{"id":97161935,"identity":"a8e9273b-5fa9-48b1-8b13-978ca9189167","added_by":"auto","created_at":"2025-12-01 12:50:44","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":103102,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/63ae676175c4848ff8ea0238.png"},{"id":97248902,"identity":"0e9a0d37-5b72-4417-be17-4d2deeee89c8","added_by":"auto","created_at":"2025-12-02 13:08:03","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":103776,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/eabcf0b3caf5419c9d7fa216.png"},{"id":97161945,"identity":"bc8c1db4-ca98-4301-8784-e4b1370f22b7","added_by":"auto","created_at":"2025-12-01 12:50:45","extension":"png","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":140626,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/11f1848450dcf4a82c80f825.png"},{"id":97250038,"identity":"0a38b34e-6dc5-4e91-8b3e-3837cf04ed21","added_by":"auto","created_at":"2025-12-02 13:13:48","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":299509,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/5924a1d4302b327adbad3ee3.png"},{"id":97161949,"identity":"d98831e6-a75a-48ad-b517-03a1d9cc1685","added_by":"auto","created_at":"2025-12-01 12:50:45","extension":"png","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":231063,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/e12a249af91e6bcd68e3c4eb.png"},{"id":97161946,"identity":"d8842d00-0e52-41d2-af2b-10b2faea15d0","added_by":"auto","created_at":"2025-12-01 12:50:45","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":218144,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/ae23911c50ce0e5e94d81ef5.png"},{"id":97161942,"identity":"212bf626-e2cd-453b-bc03-df45400380af","added_by":"auto","created_at":"2025-12-01 12:50:45","extension":"png","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":48683,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/a6b02a281b9c650bdf203281.png"},{"id":97249850,"identity":"21743727-f6e1-4299-9b3d-da4cc632c4b9","added_by":"auto","created_at":"2025-12-02 13:13:32","extension":"png","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":46899,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/00b0bc8502ef40ada746d02d.png"},{"id":97161948,"identity":"b86f9fc0-25ac-4df9-b1d4-3d3ac396b84c","added_by":"auto","created_at":"2025-12-01 12:50:45","extension":"xml","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":139203,"visible":true,"origin":"","legend":"","description":"","filename":"6df94dd090a149829b9fb0db87f775fd1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/b8cd73ec109693fafb1c3e9b.xml"},{"id":97248890,"identity":"3b1847d5-7f0e-4f90-93e9-048d7700f2ac","added_by":"auto","created_at":"2025-12-02 13:07:57","extension":"html","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":151238,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/064db8b88f9ad5d540be4026.html"},{"id":97161920,"identity":"6ae9c487-016f-4445-a22b-da2fff123332","added_by":"auto","created_at":"2025-12-01 12:50:44","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":456423,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 3‑1: Algorithm representing the methodology followed to regionalize the ecoinvent database\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/e6f91e4fbc68f78dd48232e9.png"},{"id":97161925,"identity":"5f0a4b20-5e04-4c39-a9a5-eda241600081","added_by":"auto","created_at":"2025-12-01 12:50:44","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":3137973,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 4‑1: The consumption market for nitrous dioxide in Morocco. Screenshot taken from the activity-browser software. Transportation value is hidden for copyright reasons.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/39933ac5e81154a48f88ef93.png"},{"id":97161923,"identity":"3fbe2ab7-0f50-46a1-bbc5-5dcfc80999bf","added_by":"auto","created_at":"2025-12-01 12:50:44","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2404370,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 4‑2: The regionalized national production process for nitrous dioxide in Spain. Screenshot taken from the activity-browser software. Values are hidden for copyright reasons.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/6ed6591058ae3683f0ee39c0.png"},{"id":97161921,"identity":"19776b8d-bfcf-4fb4-af6c-35152ac6effd","added_by":"auto","created_at":"2025-12-01 12:50:44","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":90521,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 4‑3: The regionalized version of the ecoinvent process for nuclear electricity in Spain. Screenshot taken from the activity-browser software. Values are hidden for copyright reasons.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/dbb856e8bb9dd7681b3eaae0.png"},{"id":97250096,"identity":"431410f9-be2b-4f0b-9180-d6f511888f79","added_by":"auto","created_at":"2025-12-02 13:13:54","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":2312893,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 4‑4: Variation of the impacts on climate change of IW+v2.1 for the 22 national production processes of nitrous dioxide from regioinvent. On the left-hand side of the graph, the two original production processes from ecoinvent. Results are normalized compare.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/ffcfc653f0c335addcac3da2.png"},{"id":97161922,"identity":"519e6a66-d591-466f-b971-c8caff6e478a","added_by":"auto","created_at":"2025-12-01 12:50:44","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":103157,"visible":true,"origin":"","legend":"\u003cp\u003eFigure 4‑5: Variation of the impacts on water scarcity of IW+v2.1 for the 22 national production processes of nitrous dioxide from regioinvent. On the left-hand side of the graph, the two original production processes from ecoinvent. Results are normalized compare\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/b9217b2f5b077ead0f68a38b.png"},{"id":97664544,"identity":"b58660bf-62b0-42fa-af33-a196b35aa4a4","added_by":"auto","created_at":"2025-12-08 09:09:34","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":10181958,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/affc53be-bb10-4e67-b09f-080869d2d9da.pdf"},{"id":97161928,"identity":"ee254c40-b992-4ba9-b0b2-ed0f9d91eda7","added_by":"auto","created_at":"2025-12-01 12:50:44","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":112211,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryinformationv5.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-8159063/v1/50977ff6bef7aa5c4c9b86f9.xlsx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Regioinvent: a regionalized version of ecoinvent integrating detailed trade data","fulltext":[{"header":"1. INTRODUCTION","content":"\u003cp\u003eRegionalization is a core aspect of conducting any Life Cycle Assessment (LCA) study. In practice, it involves copying an existing process from an LCA database and adapting its inputs to the relevant geographical context (Baitz et al. 2013). However, the term \u003cem\u003eregionalization\u003c/em\u003e itself is broad and can refer to several distinct aspects. Patouillard et al. (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) distinguished between three components. The regionalization of the inventory, pertaining to the modification of the technosphere inputs/outputs to match the required geographical context (e.g., from \"electricity - RER\" to \"electricity - BE\"). The spatialization of the inventory, which consists in specifying a location for the extraction/release of elementary flows (e.g., from \"Water\" to \"Water, BE\"). The impact regionalization which is the characterization of a spatialized inventory with regionalized characterization factors (e.g., the characterization factor of \"Water, BE\" according to the AWARE 2.0 method is 3.56 m\u0026sup3; world-eq., while it is 38.2 m\u0026sup3; world-eq for the global average (Seitfudem et al. \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e2025\u003c/span\u003e)). Typically, LCA practitioners handle the regionalization and spatialization of inventories, whereas LCIA model developers are responsible for impact regionalization (Patouillard et al. \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). These three dimensions of regionalization have been shown to significantly influence LCA results (Mutel and Hellweg 2009), (Yang and Heijungs 2017), (Hung et al. \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), (Schenker, Oberschelp, and Pfister 2022), (Adrianto, Pfister, and Hellweg 2022), highlighting the necessity for regionalization within LCA studies.\u003c/p\u003e\u003cp\u003eThere are multiple ways to apply regionalization and spatialization to an LCA inventory. n its simplest form, regionalizing the inventory consists of changing the origin of a commodity or service\u0026mdash;for example, replacing a European electricity grid mix with a Belgian one. In more advanced applications, additional data are collected to adjust not only the origin of a commodity but also the quantities and types of its inputs and outputs, or even the production technologies used (Patouillard et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Examples include the regionalized modeling of wind turbines in Denmark (Sacchi et al. \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) and of milk production in the United States (Henderson et al. \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). For spatialization, the most basic approach is to specify the region or country where an emission or resource extraction occurs. This can be done by creating new elementary flows (e.g., \u0026ldquo;Water, CA-QC\u0026rdquo;) or by assigning the location during calculation, thereby avoiding the creation of thousands of additional flows (Sacchi et al. \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). More advanced approaches make use of geographic information systems (GIS) (Mutel \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2012\u003c/span\u003e), to pinpoint emissions or extractions at much finer spatial resolution. Li et al. (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) review 105 LCA studies that rely on GIS-based spatialization. Ongoing research also seeks to integrate both regionalization and spatialization within a unified computational framework (Mutel and Hellweg 2023).\u003c/p\u003e\u003cp\u003eThe application of regionalization in LCA is often restricted to the foreground system\u0026mdash;that is, to the data directly collected and modeled by the practitioner. The background database (e.g., ecoinvent) is almost always left unchanged. However, LCA databases themselves offer limited geographical resolution. For example, ecoinvent v3.10.1 (cut-off) covers on average only about 0.85 countries per process, meaning that fewer than one country-specific process is available per activity. Like other LCA databases, ecoinvent therefore relies heavily on broad regional representations such as Europe (RER), Rest of the World (RoW), or Global (GLO). Moreover, each new ecoinvent release updates only a small subset of process models and markets. As a result, most market datasets remain several years\u0026mdash;or even decades\u0026mdash;out of date (Steubing et al. \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). Given that the global economy evolves continually, the market shares used in ecoinvent are unlikely to reflect current average supply chains. This issue is especially pronounced for energy-related markets (e.g., crude oil), which are highly sensitive to temporal changes.\u003c/p\u003e\u003cp\u003eThese limitations, shared by all LCA databases, could in principle be addressed, since the simplest forms of regionalization and spatialization of the inventory can be automated and applied across an entire background database. Alaux et al. (2024), for example, developed an algorithm that systematically screens the ecoinvent database and reconnects regional inputs of country-specific processes to national inputs whenever possible. This improves the geographical representativeness of ecoinvent processes to some extent, but it neither corrects the outdated or misrepresented market structures found in LCA databases nor provides spatialization of elementary flows. Sacchi (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) by contrast, adapted the consequential version of ecoinvent by generating consumption markets based on trade data and even explored how these markets would respond to marginal changes in product demand. However, this approach was applied only to two products and was not generalized to the full database. Sany\u0026eacute; Mengual et al. (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2023\u003c/span\u003e) extended regionalization further by applying it to 164 selected commodities in ecoinvent and Agrifootprint, focusing on adapting energy inputs, waste treatment, feedstock diets, and consumption markets. Yet, their work also does not cover the full set of database commodities, and while the resulting data are open-access, the code used to generate the regionalized dataset is not open-source. Peng \u0026amp; Pfister (2024) released code to generate a regionalized version of ecoinvent using EXIOBASE, a multi-regional input\u0026ndash;output database. Their implementation also created consumption markets based on EXIOBASE trade data, covering 44 countries for each commodity and adapting the corresponding ecoinvent activities, while also spatializing water-related elementary flows. Unlike Sacchi (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) and Sany\u0026eacute; Mengual et al. (\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), their approach was applied to the entire ecoinvent database. Although this represents a major step toward fully automating the regionalization of ecoinvent, several limitations remain. Because the approach relies on EXIOBASE, much of the detailed trade information available at the commodity level is lost, as the model is restricted to the 163 industries and 44 countries covered by EXIOBASE. This leads to well-known aggregation issues in input\u0026ndash;output analyses (Lenzen \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2011\u003c/span\u003e), where commodities within the same sector cannot be distinguished. For instance, EXIOBASE cannot differentiate between individual chemical products, meaning that all trade data for substances such as propanol or benzene collapse into the sector-level average for \u0026ldquo;Chemicals.\u0026rdquo; The reliance on EXIOBASE also automatically excludes all countries not represented among its 44 regions. Another limitation arises from the monetary nature of EXIOBASE trade flows. The authors derived monetary trade shares to build consumption markets and applied these shares to ecoinvent markets, which are expressed in physical units. Since commodity prices vary substantially across countries, this procedure inevitably introduces inconsistencies. Their implementation also has practical constraints, requiring more than 350 GB of RAM\u0026mdash;an amount far beyond what most LCA practitioners can access. Finally, as with previous studies generating consumption markets, their approach relies on proxy regions whenever no country-specific production dataset exists. In other words, the market description introduces a regional process (e.g., RER) whenever the national dataset (e.g., AT) is missing.\u003c/p\u003e"},{"header":"2. AIM AND SCOPE","content":"\u003cp\u003eThis article aims at creating --- in a fully automated, reproducible, highly maintainable and iteratively improvable data fusion approach --- a fully regionalized version of the ecoinvent LCA database at the country level. This entails the automatic creation of thousands of regionalized and spatialized inventories of national production processes of commodities and of national consumption markets. This research is looking to rely on the best detailed available data regarding trade and regionalized characterization factors to achieve the most fine-grained possible regionalization of ecoinvent through an \u003cspan type=\"ItalicUnderline\" class=\"ItalicUnderline\" name=\"Emphasis\"\u003eautomated\u003c/span\u003e process. As a result, in order to automatically apply heuristics across the database, this research will adopt the simplified approach to regionalization/spatialization, which means that only the origin of inputs/outputs will be adapted in national production processes (not the quantities) and that we will create new elementary flows for spatialization, which also means no GIS will be involved. This research thus does not aim at generating new processes through additional data collection, but rather through integration of already existing sources of data.\u003c/p\u003e\u003cp\u003eIn this article, the resulting \u0026ldquo;regioinvent\u0026rdquo; adaptation of the ecoinvent database described requires a significant amount of computation resources. However, another goal of this research is to ensure that any LCA practitioner be able to use the fruits of this research. This implies that any typical laptop must be able to generate the resulting \"regioinvent\" adaptation of ecoinvent within a few hours and that regioinvent must be usable in mainstream LCA software. In the article, we will thus also study the effect of certain strategies to reduce the size of the regioinvent database on the results, in order to reduce required computation resources and ensure the operationalization of the research.\u003c/p\u003e\u003cp\u003eThrough the comparison of ecoinvent with the resulting regioinvent database, the effect of the regionalization of inventories and of the spatialization of inventories will be estimated on a broad sample.\u003c/p\u003e"},{"header":"3. METHODS","content":"\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\n \u003ch2\u003e3.1 Data sources\u003c/h2\u003e\n \u003cp\u003eTo perform regionalization as implemented in regioinvent, four components are required: life cycle inventories, trade data, production volumes, and a reference list of elementary flows to be spatialized.\u003c/p\u003e\n \u003cp\u003eIn this study, we focus on the ecoinvent database (cutoff v3.10.1), as it provides the most comprehensive coverage of commodities and geographies among existing LCA databases. Additionally, substantial efforts by the ecoinvent team have resulted in detailed geographical differentiation for energy-related processes, such as electricity and heat production, which this research leverages. Nevertheless, the methodology developed here for automated regionalization can be applied to other LCA databases as well.\u003c/p\u003e\n \u003cp\u003eWe selected the cutoff system model because it is the most widely used, but regioinvent is also compatible with the APOS (Allocation at Point of Substitution) version. Applying regioinvent to the consequential version of ecoinvent, however, would require the use of marginal trade data, whereas the current implementation relies on average trade data. Similarly, applying regioinvent to prospective versions of ecoinvent\u0026mdash;such as those generated through the \u003cem\u003epremise\u003c/em\u003e package (Sacchi et al. \u003cspan class=\"CitationRef\"\u003e2022\u003c/span\u003e)\u0026mdash;would require prospective trade data to capture the evolving dynamics of international trade.\u003c/p\u003e\n \u003cp\u003eThe trade data used in regioinvent are sourced from the BACI (Base pour l\u0026rsquo;Analyse du Commerce International) database, which itself is based on the United Nations\u0026rsquo; UN COMTRADE dataset. BACI provides detailed records of imports and exports for thousands of commodities traded between virtually all countries worldwide.\u003c/p\u003e\n \u003cp\u003eFor domestic production\u0026mdash;that is, commodities that do not cross borders and therefore do not appear in BACI\u0026mdash;we rely on additional data sources. FAOSTAT is used to obtain production volumes for agricultural commodities. Production volumes for mineral extraction and base metal production are drawn from the British Geological Survey (BGS) and the United States Geological Survey (USGS). For all remaining commodities, we use EXIOBASE to estimate the share of domestic production consumed domestically versus exported.\u003c/p\u003e\n \u003cp\u003eFor the list of elementary flows to be spatialized, we rely on three regionalized LCIA methods: IMPACT World+ (IW+), ReCiPe, and Environmental Footprint (EF). These methods are already implemented in mainstream LCA software and are among the most widely used LCIA frameworks.\u003c/p\u003e\n \u003cp\u003eCompared with earlier studies that introduced trade data into ecoinvent, (Peng and Pfister 2024) solely relied on EXIOBASE as a source of data, (Sacchi \u003cspan class=\"CitationRef\"\u003e2018\u003c/span\u003e) used the UN COMTRADE, FAOSTAT and USGS databases and (Sany\u0026eacute; Mengual et al. \u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e) relied on Eurostat data.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\n \u003ch2\u003e3.2 Overall methodology\u003c/h2\u003e\n \u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e\u0026thinsp;\u0026minus;\u0026thinsp;1 illustrates the algorithm followed by \u003cem\u003eregioinvent\u003c/em\u003e to generate a regionalized version of the ecoinvent database. The colors in the figure correspond to the different components of the methodology: extraction and processing of trade data (yellow), creation of national production processes (orange), creation of national consumption markets and technology mixes (blue), relinking (purple), and spatialization of elementary flows (red). Each of these steps is detailed in the subsequent subsections of the methodology.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\n \u003ch2\u003e3.3 Trade data extraction and treatment (in yellow)\u003c/h2\u003e\n \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eAim of the section\u003c/span\u003e: The objective is to determine, for each ecoinvent commodity, the distribution of producing countries within the global production market and within each national consumption market. In other words: Which countries produce commodity \u003cem\u003eA\u003c/em\u003e globally? For the commodity \u003cem\u003eA\u003c/em\u003e consumed in country \u003cem\u003eB\u003c/em\u003e, what share originates from domestic production versus imports? And, if imported, from which countries and in what proportions?\u003c/p\u003e\n \u003cp\u003eThe ecoinvent cutoff v3.10.1 database includes 4,031 commodities (products or services) represented through 23,523 processes. Of these commodities, 1,982 are traded internationally, while the remaining 2,049 are not\u0026mdash;for example, \u0026ldquo;building, hall,\u0026rdquo; \u0026ldquo;diesel, burned in machinery,\u0026rdquo; or \u0026ldquo;metal working, average for aluminium product.\u0026rdquo; The trade data extraction described in this section concerns only the 1,982 internationally traded commodities.\u003c/p\u003e\n \u003cp\u003eInternational import and export data for these commodities were extracted from the BACI database using a concordance between ecoinvent product names and the Harmonized System (HS) classification employed by BACI. This concordance was initially generated with the assistance of ChatGPT-4o to accelerate the process, and each mapping was subsequently checked and validated manually by the authors. The final mapping is provided in the Supplementary Information.\u003c/p\u003e\n \u003cp\u003eTrade data were extracted for the five most recent available years (2018\u0026ndash;2022) from the HS17 202401b version of BACI and averaged arithmetically over this period. Using a five-year average mitigates the influence of atypical years and reduces the risk of missing data points. We used BACI\u0026rsquo;s physical data (tonnes of traded products), thereby avoiding inconsistencies that would arise from applying monetary trade ratios to ecoinvent\u0026rsquo;s physically based flows. The extracted and formatted dataset is available on Zenodo and can be reused in other projects.\u003c/p\u003e\n \u003cp\u003eThe BACI database does not distinguish between exports and re-exports. Re-exporting occurs when a country imports a commodity and subsequently exports it without any transformation. As a result, BACI may identify countries as major exporters of goods they do not produce. For example, the Netherlands appears as the 7th largest exporter of bananas, even though bananas do not grow in its climate. Directly using raw export data would therefore introduce substantial inconsistencies.\u003c/p\u003e\n \u003cp\u003eTo address this issue, regioinvent relies on net exports rather than raw export values. Net exports are calculated as exports minus imports. When a country imports large quantities of a commodity for the purpose of re-exporting it, its net export value will become zero or negative. Such a condition signals that the country is not a genuine producer or exporter of the commodity but primarily a re-exporter. For all commodities with identified re-exporting behavior, we correct the corresponding import data by redistributing the import volume to the top five global net exporters of that commodity. The selection of the five largest net exporters is arbitrary; however, because the imported quantities are simply reallocated, the global balance between supply and consumption remains conserved.\u003c/p\u003e\n \u003cp\u003eNote that Sany\u0026eacute; Mengual et al. (\u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e) proposed a more refined method to correct for re-exports. Instead of reallocating the entire export value, their approach accounts for the share that may originate from domestic production using the national import ratio. For example, if France has a negative net export balance for fridges and imports 25% of the fridges sold domestically, their method assumes that 25% of France\u0026rsquo;s fridge exports originate from domestic production, while the remaining 75% stem from re-exports. In regioinvent, by contrast, we implicitly assume that 100% of such exports are due to re-exports.\u003c/p\u003e\n \u003cp\u003eThe BACI database provides information only on commodities that cross national borders. As a result, commodities that are produced and consumed domestically do not appear in BACI. To the best of the authors\u0026rsquo; knowledge, no open-access database offers comprehensive production volume data at a sufficiently detailed level (i.e., comparable to HS6 resolution). We therefore rely on multiple data sources. When total domestic production data are available\u0026mdash;as is the case for agricultural commodities (via FAOSTAT) and for mineral extraction and metal production (via BGS and USGS)\u0026mdash;we estimate domestically consumed production by subtracting net exports from the reported domestic production volume.\u003c/p\u003e\n \u003cp\u003eWhen domestic production volumes are not available, we estimate domestic production and consumption using the EXIOBASE database. For each commodity and each EXIOBASE country or region, we compute the ratio of exports to domestic consumption. This ratio is then applied to the BACI net export values to estimate the domestic consumption of domestically produced goods in physical units, following the equation presented below.\u003c/p\u003e\n \u003cdiv id=\"Equa\" class=\"Equation\"\u003e\n \u003cdiv class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e$$\\:{D}_{x,y}={E}_{x,y}(\\frac{1}{{R}_{X,Y}}-1)$$\u003c/div\u003e\n \u003c/div\u003e\n \u003cp\u003eWhere:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eD is the amount (in tonnes) of domestically produced commodity \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:x\\)\u003c/span\u003e\u003c/span\u003e that is consumed within country \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:y\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eE is the amount (in tonnes) of net exports of commodity \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:x\\)\u003c/span\u003e\u003c/span\u003e of the country \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:y\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eR is the ratio of domestically consumed vs exported of the sector \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:X\\)\u003c/span\u003e\u003c/span\u003e to which belongs commodity \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:x\\)\u003c/span\u003e\u003c/span\u003e in region \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Y\\)\u003c/span\u003e\u003c/span\u003e to which belongs country \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:y\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003cp\u003eTo illustrate this approach, consider the example of benzene in South Korea. We do not know the country\u0026rsquo;s benzene production volume directly. However, BACI reports that South Korea exports 630,000 t of benzene, and EXIOBASE indicates that, on average, South Korea exports 54% of its domestic production of chemicals. We can therefore estimate the national benzene production by dividing the net exports by this export share, yielding an estimated domestically consumed production of approximately 536,000 t of benzene.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\n \u003ch2\u003e3.4 Creation of national production processes (in orange)\u003c/h2\u003e\n \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eAim of the section\u003c/span\u003e: The goal of this section is to create national production processes, that is, regionalized copies of the corresponding ecoinvent processes.\u003c/p\u003e\n \u003cp\u003eAfter identifying the commodities to be regionalized, the next step is to determine for which countries each commodity should be regionalized. Peng \u0026amp; Pfister (2024) chose to regionalize every activity for all 44 EXIOBASE countries. However, this produces a large number of irrelevant datasets\u0026mdash;such as a \u0026ldquo;production of banana\u0026rdquo; process for Canada, which is implausible given the country\u0026rsquo;s climate. A more pragmatic approach is to determine the relevant countries commodity by commodity, based on the major global producers.\u003c/p\u003e\n \u003cp\u003eFor internationally traded commodities, we therefore rely on the production data (i.e., domestic production plus net exports, as determined in the previous section) to identify which countries should be included in the regionalization. Countries are added in descending order of their production volumes until their cumulative production reaches 99% of the global total for that commodity. All remaining countries are aggregated into a Rest-of-the-World (RoW) region.\u003c/p\u003e\n \u003cp\u003eFor example, if the global production of commodity A is distributed as follows\u0026mdash;China (55%), United States (24.5%), Australia (12.5%), Chile (7.5%), Guatemala (0.3%), and Spain (0.2%)\u0026mdash;then regionalized production processes would be created for China, the United States, Australia, and Chile. Guatemala and Spain would be grouped into a RoW process.\u003c/p\u003e\n \u003cp\u003eFor commodities that are not traded internationally, this country-selection method cannot be applied. In these cases, we generate regionalized versions of each process for all countries covered by the BACI database (225 countries).\u003c/p\u003e\n \u003cp\u003eFor each selected commodity\u0026ndash;country combination, we create a regionalized production process by copying the ecoinvent process corresponding to the region in which that country is located (e.g., \u0026ldquo;RER\u0026rdquo; for Germany, \u0026ldquo;RME\u0026rdquo; for Lebanon). No assumptions are made regarding technological equivalency between countries. Thus, if a country does not have its own dataset in ecoinvent, we use the broader regional dataset rather than selecting another country with potentially similar technology (e.g., we do not model Germany based on France, but on the RER region).\u003c/p\u003e\n \u003cp\u003eTo adapt this regional process to the specific country, three key types of inputs are systematically replaced: electricity, heat, and municipal solid waste treatment. These inputs are substituted with their country-specific counterparts in each regionalized process. They were selected because they appear in virtually all ecoinvent processes and are already provided with detailed geographical differentiation in ecoinvent.\u003c/p\u003e\n \u003cp\u003eWhen multiple production technologies exist for the same commodity, we generate a regionalized copy for each technology in every relevant country. For example, 1-butanol in ecoinvent can be produced either via hydroformylation of propylene or via a Fischer\u0026ndash;Tropsch process. Regardless of which technology is used in a given country in reality, regioinvent creates a regionalized version of each technology for each country.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\n \u003ch2\u003e3.5 Creation of national consumption markets and technology mixes (in blue)\u003c/h2\u003e\n \u003cp\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003eAim of the section\u003c/span\u003e: The objective is to create national consumption market processes, i.e., processes that represent the average origin of a commodity purchased within a given country.\u003c/p\u003e\n \u003cp\u003eAs with the creation of national production processes, there is no need to create consumption markets for countries that do not meaningfully consume a given commodity. Countries included in the regionalization are therefore selected based on their consumption values, sorted in descending order, until their cumulative consumption reaches 99% of global consumption for that commodity.\u003c/p\u003e\n \u003cp\u003eRegarding commodities produced through multiple technologies, no available dataset provides trade information disaggregated by technology. In the absence of such detail, we assign technologies according to their global market shares in ecoinvent. For example, in the case of 1-butanol, the hydroformylation of propylene accounts for 94.5% of the global market in ecoinvent, while the Fischer\u0026ndash;Tropsch route accounts for 5.5%. Consequently, every national consumption market for 1-butanol adopts this 94.5/5.5 technology split. Transportation inputs are copied directly from the corresponding global market process in ecoinvent.\u003c/p\u003e\n \u003cp\u003eFor commodities that are not traded internationally, we create \u0026ldquo;technology mix\u0026rdquo; processes to represent the distribution of production technologies. These mixes simply reuse the technology shares already present in ecoinvent.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\n \u003ch2\u003e3.6 Relinking (in purple)\u003c/h2\u003e\n \u003cp\u003eWe then relink all created processes\u0026mdash;both national production processes and national consumption markets\u0026mdash;to one another within regioinvent. This ensures, for example, that the newly created Algerian brass production process uses Algerian consumption markets for zinc and copper, rather than relying on global markets.\u003c/p\u003e\n \u003cp\u003eIn addition, we relink all created processes to the original ecoinvent datasets in order to improve the regionalization of the three key inputs central to our approach (electricity, heat, and waste). As a result, for instance, the Austrian electricity-from-coal process now draws on the Austrian consumption market for coal instead of using the European market.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\n \u003ch2\u003e3.7 Spatialization of elementary flows (in red)\u003c/h2\u003e\n \u003cp\u003eFor spatialization, regioinvent applies a straightforward rule: each elementary flow is assigned the geography of the process that emits or extracts it. In other words, if 1-butanol is produced in Belgium, all associated water use and water emissions are assumed to occur in Belgium. This assumption is already widely used in LCA practice, including in SimaPro and, previously, in openLCA.\u003c/p\u003e\n \u003cp\u003eTo determine which elementary flows should be spatialized, we compiled the spatialized flows from three regionalized LCIA methods\u0026mdash;IMPACT World+, ReCiPe, and Environmental Footprint (EF)\u0026mdash;at both midpoint and endpoint/damage levels. Consequently, the following categories of elementary flows were spatialized: water, land, acidification, eutrophication, particulate matter formation, and photochemical ozone formation. Spatialization was applied to both regioinvent processes and the original ecoinvent processes.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\n \u003ch2\u003e3.8 Operationalization of regioinvent\u003c/h2\u003e\n \u003cp\u003eThe version of regioinvent used in this article consists of 669,571 newly created processes and 110,559 newly spatialized elementary flows, in addition to the 23,523 original processes and 4,362 elementary flows in ecoinvent 3.10.1 (cutoff). Generating this full database requires substantial computational resources\u0026mdash;approximately 10 hours of processing time and 128 GB of RAM. Once generated, running a simple cradle-to-gate LCA on an ecoinvent process requires a machine with at least 64 GB of RAM and takes roughly 30 minutes on a 2.5 GHz CPU.\u003c/p\u003e\n \u003cp\u003eThis full version, while appropriate for research purposes, is not practical for everyday LCA use in unit-process format. When converted to system processes, computational requirements decrease significantly and become manageable on typical laptops, but this conversion also entails a loss of information. To enable the use of regioinvent in unit-process form, we therefore introduce several rules and parameters to reduce the size of the database, thereby lowering both the computational burden of generating it and the resources needed to run LCAs with it.\u0026nbsp;\u003c/p\u003e\n \u003cp\u003eFirst, we reduce the number of non-internationally traded commodities to be regionalized. Instead of regionalizing all 2,049 such commodities, we focus only on those that are most relevant. To identify which commodities should be included, we perform an LCA of every process in ecoinvent 3.10.1 cutoff (excluding markets, i.e., 14,975 processes) and analyze the contribution of each of the 2,049 commodities across all unit processes and across the two areas of protection in the IMPACT World+ method: Human Health and Ecosystem Quality. If, in any of the 14,975 processes and for either area of protection, a commodity contributes more than 0.3% to the total impact, it is classified as relevant. If a commodity never exceeds this 0.3% contribution threshold, it is deemed irrelevant and is no longer regionalized. The 0.3% threshold is arbitrary: a 1% threshold resulted in too few relevant commodities, while a 0.1% threshold produced too many. At 0.3%, only 67 of the 2,049 commodities qualify as relevant and are therefore regionalized. This reduction brings the database size down to 228,013 processes, making it feasible to generate and run calculations on a typical machine with 16 GB of RAM.\u003c/p\u003e\n \u003cp\u003eWe also introduce a cut-off threshold for determining which countries are included in the regionalization of the 1,982 internationally traded commodities. This corresponds to the 99% cumulative coverage threshold described in sections 3.4 \u0026amp; 3.5. Users may choose to lower this threshold, thereby reducing the number of countries for which each commodity is regionalized. For example, selecting a 90% cut-off reduces the database size to 95,871 processes, while a 75% cut-off results in a database of 51,327 processes. With a 75% threshold, an LCA performed with regioinvent in brightway2 takes approximately 30 seconds for the first run and less than 10 seconds for subsequent runs on a standard laptop with 16 GB RAM and a 4-core 3.3 GHz CPU.\u003c/p\u003e\n \u003cp\u003eAt the time of writing, regioinvent operates exclusively within the brightway framework. Importers for SimaPro and openLCA are planned, but it remains uncertain to what extent these software platforms will be able to handle datasets of this size.\u003c/p\u003e\n \u003ch2\u003e3.9 Data availability\u003c/h2\u003e\n \u003cp\u003eRegioinvent is a completely open-source and free project that anyone can install and use, provided they have access to an ecoinvent license. The version used in this article is the v1.3.0 (Agez 2025a)\u003ca href=\"#_ftn1\" name=\"_ftnref1\" title=\"\"\u003e\u003c/a\u003e\u003csup\u003e1\u003c/sup\u003e and operates with BACI HS17 202401b data (Gaulier and Zignago 2010) that was already extracted and formatted for use with regioinvent (Agez 2025b). We relied on production volumes from FAOSTAT (FAOSTAT 2024) and from BGS/USGS which were previously extracted in (Bucciarelli, Hache, and Mignon 2025).The v3.9.5 of EXIOBASE (Stadler et al. 2021) is used to determine export/production ratios. Regioinvent currently only supports the regionalization of the versions 3.9/1 and 3.10/1 cut-off of ecoinvent (Wernet et al. 2016). It works alongside the brightway2 LCA software (Mutel 2017) and relies on the wurst Python package\u003ca href=\"#_ftn2\" name=\"_ftnref2\" title=\"\"\u003e\u003c/a\u003e\u003csup\u003e2\u003c/sup\u003e, to speed up the generation of the database. A specific fully regionalized version of IMPACT World+ v2.1 (Agez et al. 2024) is used for characterization in this research, but fully regionalized versions of ReCiPe 2016 v1.03 (H) and EF3.1 are also available within the regioinvent data package.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003cbr\u003e\u003c/div\u003e"},{"header":"4. RESULTS AND DISCUSSION","content":"\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\n \u003ch2\u003e4.1 Database description\u003c/h2\u003e\n \u003cp\u003eIn this results section, we first illustrate the level of interpretability that regioinvent offers to LCA practitioners. For this purpose, we use \u0026ldquo;nitrous dioxide\u0026rdquo; as an example, as it provides a clear and relevant system for comparison. Following this illustrative example, the remainder of the results section examines: (i) the effects of regioinvent on the database as a whole, (ii) the relative influence of spatializing elementary flows versus regionalizing technosphere flows, and (iii) the impact of the chosen cut-off threshold on the results.\u003c/p\u003e\n \u003cp\u003eA key contribution of regioinvent is the creation of consumption markets for each commodity\u0026mdash;that is, markets representing the distribution of the countries of origin for a commodity purchased within a given country. Figure \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e\u0026thinsp;\u0026minus;\u0026thinsp;1 displays the consumption market for nitrous dioxide in Morocco. It shows that Morocco primarily imports nitrous dioxide from Spain (28%), followed by China (13%), Sweden (11%), and other countries.\u003c/p\u003e\n \u003cp\u003eAll production processes referenced within the national consumption markets of regioinvent are themselves regionalized. Figure \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e\u0026thinsp;\u0026minus;\u0026thinsp;2 illustrates such a regionalized production process for Spain. As shown, the electricity input is regionalized by substituting the corresponding country-specific process from the original ecoinvent database. In addition, existing consumption markets for other commodities\u0026mdash;such as liquid oxygen or nitrous oxide\u0026mdash;are used to further adapt the process to the Spanish context. Non-internationally traded commodities are also regionalized for Spain. For instance, the \u0026ldquo;chemical factory\u0026rdquo; input is regionalized, which means it ultimately relies on Spanish cement and other Spain-specific upstream processes. Relevant elementary flows are likewise spatialized to Spain, including direct emissions of nitrogen oxides and water-related flows. Note that numerical values have been hidden in the figure for licensing reasons; however, the underlying values remain identical to those in the original ecoinvent process.\u003c/p\u003e\n \u003cp\u003eThe original ecoinvent database is also modified so that its processes rely on the corresponding regioinvent processes whenever these are available and relevant. For example, Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e\u0026thinsp;\u0026minus;\u0026thinsp;3 shows that nuclear electricity produced in Spain now draws on the Spanish technology mix for decarbonized water, as well as on the Spanish consumption market for liquid nitrogen. In addition, its water extractions are spatialized to Spain.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\n \u003ch2\u003e4.2 Results analysis for nitrous dioxide production processes\u003c/h2\u003e\n \u003cp\u003eTo illustrate how regioinvent affects LCA results, the next two figures present outcomes for two impact categories and their main contributors using the IMPACT World\u0026thinsp;+\u0026thinsp;v2.1 LCIA method. The comparison includes the 22 regionalized production processes created in regioinvent for nitrous dioxide, along with the two original (i.e., non-regionalized and non-spatialized) ecoinvent production processes. Results are normalized to the RoW production process of ecoinvent, which serves as the reference (left side of the graph).\u003c/p\u003e\n \u003cp\u003eFor the climate change indicator (Fig.\u0026nbsp;\u0026lt;link rid=\u0026quot;fig4\u0026quot;\u0026gt;\u003cspan class=\"InternalRef\"\u003e4\u0026lt;/link\u0026gt;\u003c/span\u003e\u0026ndash;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e), the regioinvent processes range from 52% (Germany) to 95% (China) of the reference ecoinvent value. The climate change impact of nitrous dioxide production is primarily driven by nitrous oxide production (blue) and oxygen production (orange). Their contributions vary across all regionalized processes because each country relies on different consumption markets for these inputs.\u003c/p\u003e\n \u003cp\u003eFurther down the supply chain, most greenhouse gas emissions stem from the electricity used in air separation to produce oxygen\u0026mdash;an input that is also required for the production of nitrous oxide. Germany shows the lowest climate change impact because its nitrous oxide supply is predominantly domestic and, in turn, relies on oxygen that is mostly imported from France, where electricity has a relatively low carbon intensity. At the opposite end, China relies entirely on domestic nitrous oxide and oxygen production, which means that Chinese electricity\u0026mdash;typically more carbon intensive\u0026mdash;is used for air separation, driving the higher impact.\u003c/p\u003e\n \u003cp\u003eAn interesting observation is that none of the 22 regionalized production processes exceeds the climate change impact of the original ecoinvent RoW process. This is largely because the RoW liquid oxygen dataset in ecoinvent relies heavily on the RAS electricity market group. The RAS market mixes electricity based on total electricity production shares in Asia rather than on production shares specific to liquid oxygen, resulting in an inflated impact for the RoW dataset.\u003c/p\u003e\n \u003cp\u003eFor the water scarcity indicator (Fig. \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e\u0026ndash;5), the influence of regioinvent is more pronounced\u0026mdash;an expected outcome given that water scarcity is a regionalized indicator. Results range from 25% (Finland) to 147% (Iran) of the reference ecoinvent value. Most water consumption occurs during the production of oxygen, either directly for the nitrous dioxide process (orange) or indirectly through the production of nitrous oxide (blue). The process itself also consumes substantial amounts of water directly (purple). In Finland, the oxygen supply ultimately comes from Finland and Sweden, both of which have very low water scarcity characterization factors (2.2 and 2.4 m\u0026sup3; world-eq). In contrast, for Iran, oxygen is produced domestically, where water scarcity is high (70 m\u0026sup3; world-eq), explaining the large differences observed.\u003c/p\u003e\n \u003cp\u003eA notable aspect of this figure is that the two ecoinvent processes show nearly identical results. This occurs because ecoinvent does not, by default, spatialize water flows. Even if water flows were spatialized\u0026mdash;as implemented in SimaPro, for example\u0026mdash;the resulting datasets would apply RoW and RER characterization factors (e.g., from the AWARE method), which are very similar (39.5 and 40.5 m\u0026sup3; world-eq). Therefore, spatialization without the creation of national production processes and consumption markets would not substantially improve the accuracy of water scarcity assessments for nitrous dioxide production.\u003c/p\u003e\n \u003cp\u003eOverall, these results show that regioinvent, through its systemic regionalization approach, enables a much deeper interpretation of LCA outcomes and provides a more accurate representation of average supply chains.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec16\" class=\"Section2\"\u003e\n \u003ch2\u003e4.3 Whole database results analysis\u003c/h2\u003e\n \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e\n \u003ch2\u003e4.3.1 The effect of regioinvent\u003c/h2\u003e\n \u003cp\u003eRegioinvent adds 669,571 new processes, covering 4,031 commodities across 225 countries. What is the effect of these additions on LCA results? To assess this, we compute impacts for all regioinvent production processes of internationally traded commodities (62,423 processes) and compare them with the corresponding original ecoinvent production processes\u0026mdash;specifically, those for which a GLO or RoW dataset is available to serve as a reference. For each product/technology combination, we evaluate the spread of the regioinvent results relative to the ecoinvent reference value.\u003c/p\u003e\n \u003cp\u003eTo do so we calculate the relative standard deviation of each \u003cem\u003eregioinvent\u003c/em\u003e process with respect to its corresponding ecoinvent reference. Outliers are removed using the interquartile range (IQR) criterion, retaining only observations within \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\left[{Q}_{1}-1.5\\cdot\\:\\text{I}\\text{Q}\\text{R},\\text{\\hspace{0.17em}}{Q}_{3}+1.5\\cdot\\:\\text{I}\\text{Q}\\text{R}\\right]\\)\u003c/span\u003e\u003c/span\u003e, where \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{Q}_{1}\\)\u003c/span\u003e\u003c/span\u003eand \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{Q}_{3}\\)\u003c/span\u003e\u003c/span\u003eare the first and third quartiles. A simple arithmetic mean of the remaining values is then computed to obtain an overall relative standard deviation for each impact category. This value reflects the typical spread in results introduced by regioinvent.\u003c/p\u003e\n \u003cp\u003eTo separate the influence of regionalization of technosphere flows from the influence of spatialization of elementary flows, we also generate a version of regioinvent without spatialization. Applying the same relative standard deviation calculation and subtracting the resulting values from those of the fully spatialized version allows us to quantify the distinct contributions of inventory regionalization and inventory spatialization across a large sample of data.\u003c/p\u003e\n \u003cp\u003eTable\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e\u0026thinsp;\u0026minus;\u0026thinsp;1 presents these differences for 34 IMPACT World\u0026thinsp;+\u0026thinsp;indicators; the full table (53 indicators) is provided in the Supplementary Information. On average, regioinvent introduces a spread of 10.8% for the climate change (GWP100) indicator, 13.9% for total human health, and 11.6% for total ecosystem quality. The latter two correspond to the aggregated damage indicators of IMPACT World+. These values therefore represent the average effect of applying regioinvent.\u003c/p\u003e\n \u003cp\u003eSome indicators show very limited sensitivity to regioinvent\u0026mdash;notably fisheries impacts, mineral resource use, and short-term toxicity and ecotoxicity indicators. Others are strongly affected, particularly those that are inherently regionalized: freshwater and terrestrial acidification, freshwater eutrophication, particulate matter formation, and water-related indicators. The results also show that spatialization of elementary flows generally has a larger influence on the outcomes than the regionalization of technosphere flows. Exceptions include land occupation, biodiversity, water scarcity (AWARE 2.0), and water availability (human health). These exceptions can be explained by the fact that regionalizing the inventory modifies electricity mixes throughout entire supply chains. This in turn alters the use of hydroelectricity, leading to substantial differences in associated water and land consumption. For the water availability (human health) indicator in particular, spatialization even reduces the spread. This is likely because non-spatialized water flows receive a positive characterization factor (i.e., they contribute to impact), whereas spatialized water flows for most developed countries receive a zero characterization factor in this category. These countries are assumed to be capable of mitigating water shortages through technological or economic means, such that water availability does not significantly affect human health.\u003c/p\u003e\n \u003cp\u003eFinally, ionizing radiation results are also strongly affected. This is due both to the small absolute magnitude of the indicator and to the regionalization of electricity mixes along supply chains, which may result in greater reliance on nuclear energy for some countries.\u003c/p\u003e\n \u003cp\u003eIt is important to note that the spatialization effects reported in Table\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e\u0026thinsp;\u0026minus;\u0026thinsp;1 are in addition to trade-based regionalization; they do not represent the effect of applying spatialization directly to the original ecoinvent database.\u003c/p\u003e\n \u003cp\u003eSector-level results (based on ISIC classifications) are provided in the Supplementary Information. ISIC sectors with fewer than 10 corresponding ecoinvent processes were excluded. These sectoral results can help identify where regionalization efforts should be prioritized. For example, processes related to crop and animal production are much more strongly affected by regionalization than processes related to repair and maintenance or waste treatment.\u003c/p\u003e\n \u003cp\u003eUnfortunately, a direct comparison with previous studies that regionalized ecoinvent is not possible. Peng and Pfister (2024), despite spatializing water flows, applied only the global AWARE characterization factor (except in their wheat case study), and their workflow cannot be replicated here due to its high computational requirements. Sacchi (\u003cspan class=\"CitationRef\"\u003e2018\u003c/span\u003e) did not spatialized elementary flows. For Sany\u0026eacute; Mengual et al. (\u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e), it is unclear whether elementary flows were spatialized, and their database is not available in brightway, preventing replication and comparison.\u003c/p\u003e\n \u003cp\u003e\u003c/p\u003e\n \u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 4-1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003e\u0026thinsp;Median differences of minimum and maximum of regioinvent processes compared to a reference ecoinvent process for 34 IMPACT World\u0026thinsp;+\u0026thinsp;v2.1 indicators.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTrade regionalization\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSpatialization\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eTotal regionalization\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eClimate change\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eFisheries impact\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e0.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eFossil and nuclear energy use\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eFreshwater acidification\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e13.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e68.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e82.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eFreshwater ecotoxicity, long term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eFreshwater ecotoxicity, short term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eFreshwater eutrophication\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e25.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e33.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eHuman toxicity cancer, long term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eHuman toxicity cancer, short term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e3.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eHuman toxicity non-cancer, long term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eHuman toxicity non-cancer, short term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e5.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eIonizing radiations, ecosystem quality\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e55.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e55.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eIonizing radiations, human health\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e58.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e58.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eLand occupation, biodiversity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e12.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eLand transformation, biodiversity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e6.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e17.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMarine acidification\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMarine ecotoxicity, long term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMarine ecotoxicity, short term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMarine eutrophication\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e21.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eMineral resources use\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e18.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e18.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eOzone layer depletion\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e8.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eParticulate matter formation\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e13.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e40.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003ePhotochemical ozone formation, ecosystem quality\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e10.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003ePhotochemical ozone formation, human health\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTerrestrial acidification\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e13.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e37.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTerrestrial ecotoxicity, long term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTerrestrial ecotoxicity, short term\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e4.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eThermally polluted water\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e35.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e35.3%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eWater availability, freshwater ecosystem\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eWater availability, human health\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e-2.9%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e17.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eWater availability, terrestrial ecosystem\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e14.5%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e19.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e34.1%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eWater scarcity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e20.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e28.7%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal human health\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e9.0%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e13.8%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003eTotal ecosystem quality\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e7.2%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.4%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"char\"\u003e\n \u003cp\u003e11.6%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003e\u003c/p\u003e\n \u003cp\u003e\u003cbr\u003e\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec18\" class=\"Section3\"\u003e\n \u003ch2\u003e4.3.2 The effect of the cutoff of regioinvent\u003c/h2\u003e\n \u003cp\u003eAs mentioned in section 3.8, the regioinvent database generated for this article is too large for most current machines to handle efficiently. We therefore introduced several simplifications to enable its use on typical computers. These simplifications introduce an aggregation bias at the country level. To assess the magnitude of this bias, we evaluate four different \u0026ldquo;versions\u0026rdquo; of regioinvent:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003e\n \u003cp\u003eFull version: the version used throughout the article, in which all eligible processes are regionalized and the country coverage cut-off is set at 99%\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eregio99: a version with the same 99% cut-off, but in which only the most relevant non-internationally traded commodities are regionalized.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eregio90: a version in which the country coverage cut-off is reduced to 90%.\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003eregio75: a version in which the cut-off is reduced to 75%.\u003c/p\u003e\n \u003c/li\u003e\n \u003c/ul\u003e\n \u003cp\u003eWe generated these versions and calculated impacts for all processes in each. We then compared results only for the processes that appear in all versions, yielding a sample of 15,154 common processes. For each of these processes, we calculated the difference between the full version\u0026mdash;which we assume provides the most accurate results\u0026mdash;and each of the three reduced versions.\u003c/p\u003e\n \u003cp\u003eReducing the regionalization scope from the full version to regio99 results in approximately a 0.5% difference in the two areas of protection (Human Health and Ecosystem Quality). Moving to regio90 introduces a 1.5\u0026ndash;2% difference, and reducing further to regio75 leads to a 2\u0026ndash;3% difference. These results indicate that using smaller versions of regioinvent does not substantially compromise overall accuracy.\u003c/p\u003e\n \u003cp\u003eHowever, some midpoint categories exhibit substantially larger deviations\u0026mdash;most notably freshwater eutrophication and water availability, with differences of roughly 30% and 50%, respectively. The Supplementary Information provides detailed comparisons for all IMPACT World\u0026thinsp;+\u0026thinsp;indicators.\u003c/p\u003e\n \u003c/div\u003e\n \u003cdiv id=\"Sec19\" class=\"Section3\"\u003e\n \u003ch2\u003e4.3.3 The case of regionalized water processes\u003c/h2\u003e\n \u003cp\u003eSpatialization of water flows is already implemented in mainstream LCA software such as SimaPro. However, the way this spatialization is currently carried out can introduce inconsistencies because the geographies of water extraction and water release do not always match. These inconsistencies reduce the robustness of the assessments. For example, in ecoinvent the production of apples in Chile uses the process \u0026ldquo;market for irrigation, RoW.\u0026rdquo; When spatialization is applied, this process extracts water from RoW (characterization factor: 39.5 m\u0026sup3; world-eq). This irrigation water is then used within the Chilean apple production process, where it is released as Chilean water (88.1 m\u0026sup3; world-eq). Under these conditions, the production of apples in Chile appears environmentally beneficial, which is clearly an artefact of inconsistent spatialization.\u003c/p\u003e\n \u003cp\u003eIn regioinvent, these inconsistencies are almost entirely eliminated. Regioinvent generates thousands of nationally contextualized processes and systematically relinks them throughout the database. As a result, Chilean apple production now relies on a Chilean irrigation process that extracts Chilean water, rather than water from a Rest-of-the-World aggregate. This ensures that mismatches in water geographies do not occur.\u003c/p\u003e\n \u003cp\u003eOne exception remains: RoW national production processes in regioinvent can still exhibit inconsistencies. This is because an \u0026ldquo;apple production RoW\u0026rdquo; process represents a specific set of countries that does not perfectly match the set of countries represented by an \u0026ldquo;irrigation RoW\u0026rdquo; process. In such cases, the geography of an input (e.g., irrigation) may still differ from the geography represented by the process itself.\u003c/p\u003e\n \u003c/div\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec20\" class=\"Section2\"\u003e\n \u003ch2\u003e4.4 Limitations and outlook\u003c/h2\u003e\n \u003cp\u003eWhile regioinvent represents a significant advancement in the representation of supply chains within LCA, its current implementation still has several limitations, which also represent clear avenues for future improvement. Although the uncertainty introduced by each limitation is difficult to quantify, all contribute some degree of uncertainty to the results produced by regioinvent. Nevertheless, we believe that the models generated by regioinvent still improve overall accuracy compared with the original ecoinvent processes, as they incorporate substantial additional information that more finely resolves supply chain structures.\u003c/p\u003e\n \u003cp\u003eAs specified in section 3.3, our approach relies on a combination of actual production volume data and estimates derived by applying export/domestic consumption ratios from EXIOBASE to BACI net-export values. This estimation method is highly dependent on the EXIOBASE ratios and is therefore subject to aggregation bias\u0026mdash;both in terms of commodities (EXIOBASE includes only\u0026thinsp;~\u0026thinsp;200 commodity groups) and geography (49 regions). Moreover, EXIOBASE ratios are expressed in monetary units (euros), which we then apply to physical quantities (tonnes). This creates inconsistencies because commodity prices vary significantly depending on origin. These issues mirror the limitations identified in Peng and Pfister (2024) although in our case they affect only domestic production estimate. Most importantly, this estimation approach fails when countries produce a commodity exclusively for domestic consumption and do not export it. For example, if China produces benzene but consumes it entirely domestically, the method would incorrectly infer that China produces no benzene, simply because it exports none. Correcting this issue requires supplementing the workflow with additional production volume data. However, integrating comprehensive production statistics for thousands of commodities across all countries represents a substantial effort and remains an open challenge.\u003c/p\u003e\n \u003cp\u003eWhenever multiple technologies exist for producing a given commodity, we distribute these technologies uniformly across all national consumption markets, using the global technology shares provided in ecoinvent. However, this approach introduces limitations: outdated technologies used in certain regions are inadvertently attributed to countries that predominantly rely on more advanced technologies. For example, due to this approximation, European countries are modeled in regioinvent as producing a portion of their butanol via Fischer\u0026ndash;Tropsch synthesis, even though this technology is largely obsolete in Europe. Addressing this limitation would require extensive expert knowledge for the many commodities in ecoinvent that have multiple production technologies.\u003c/p\u003e\n \u003cp\u003eAnother major area for improvement concerns the contextualization of transportation. In its current form, regioinvent simply copies the transport distances and modes from the global market dataset of each commodity in ecoinvent. This leads to unrealistic transport patterns\u0026mdash;for example, sea tanker transport between Switzerland and Germany. Ideally, transport distances and modes should be adapted for each national consumption market based on the actual origin-destination pairs, as done at the European scale in (Sany\u0026eacute; Mengual et al. \u003cspan class=\"CitationRef\"\u003e2023\u003c/span\u003e). The automated approach of (Fry et al. \u003cspan class=\"CitationRef\"\u003e2024\u003c/span\u003e) could thus be adapted and integrated in regioinvent in the future to correct these inconsistencies.\u003c/p\u003e\n \u003cp\u003eCentralized trade databases such as BACI provide only national-level information and do not distinguish between sub-national regions (e.g., states or provinces). However, finer-scale data are often available from national statistics agencies\u0026mdash;for instance, Statistics Canada provides international merchandise trade information for each Canadian province. Incorporating such sub-national data into regioinvent would refine regionalization for large countries with substantial internal variation in production technologies, such as Canada, the United States or China.\u003c/p\u003e\n \u003cp\u003eFinally, BACI provides annual time-series data, enabling the tracking of year-to-year fluctuations in trade flows. These temporal dynamics could be incorporated into regioinvent, either through regular annual updates of the trade data or through less frequent updates that include interannual variability as an explicit source of uncertainty within the datasets.\u003c/p\u003e\n\u003c/div\u003e"},{"header":"5. RECOMMENDATIONS \u0026 CONCLUSION","content":"\u003cp\u003eFor LCA professionals, we recommend using regioinvent rather than relying solely on ecoinvent, as regioinvent provides a more accurate representation of supply chains and consistently spatializes elementary flows across multiple impact categories. When modeling with regioinvent, users should distinguish between the three types of processes it provides: national production processes, national consumption markets, and global production markets (the latter are provided as default options).\u003c/p\u003e\u003cp\u003e\u003cul\u003e\u003cli\u003e\u003cp\u003eIf the country of production is known, the corresponding national production process should be used.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eIf the country of production is unknown but the country of purchase is known, the national consumption market should be selected, as it represents the average origin of the commodity sold in that country.\u003c/p\u003e\u003c/li\u003e\u003cli\u003e\u003cp\u003eIf neither production nor purchase country is known, users can rely on the global production markets provided in regioinvent.\u003c/p\u003e\u003c/li\u003e\u003c/ul\u003e\u003c/p\u003e\u003cp\u003eFor policy-makers, regioinvent allows rapid generation of country-specific supply chains, effectively creating a tailored version of ecoinvent. This is especially valuable for developing countries, where dedicated datasets and studies are often scarce. Using regioinvent\u0026rsquo;s supply chains can therefore enhance the modeling and evaluation of policies.\u003c/p\u003e\u003cp\u003eFor LCA database providers, regioinvent offers an opportunity to strengthen the completeness and realism of their databases. It enables more accurate modeling of supply chains through explicit representation of international trade. Because BACI trade data are updated annually, consumption markets can also be refreshed automatically each year. In addition, our work highlights an increasing need for accessible production volume data to improve market modeling.\u003c/p\u003e\u003cp\u003eFor LCA software providers, there is a growing need to rethink how large-scale LCA data are handled. Many current LCA tools cannot accommodate regioinvent because of its size. While mainstream LCA databases are currently limited to roughly 25,000 processes, future datasets will inevitably become much larger. If this expansion is not driven by database providers, it will come from practitioners themselves, as LCA increasingly moves toward big-data applications. Most existing LCA software is not yet equipped for this future.\u003c/p\u003e\u003cp\u003eRegioinvent is a regionalized version of ecoinvent in which supply chains are modeled using international trade data from the BACI database and elementary flows are spatialized. The resulting database contains 669,571 processes and 110,559 spatialized elementary flows. Comparing regioinvent with the corresponding ecoinvent processes shows that regionalization introduces average differences of 10.8%, 13.9%, and 11.6% for climate change, total human health, and total ecosystem quality indicators, respectively. This research also quantifies, for the first time, the specific contribution of spatialization relative to inventory regionalization, on a large sample of data. Regioinvent builds on recent efforts to regionalize the ecoinvent database and provides a substantial improvement over current practice. Nevertheless, important work remains\u0026mdash;such as consolidating domestic production data and improving the regionalization of commodity distributions. Furthermore, significant updates will be necessary from LCA software developers to ensure that large-scale datasets like regioinvent can be handled efficiently.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eConflict of Interest Statement:\u003c/h2\u003e\u003cp\u003eThe authors declare no conflict of interest.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eM. Agez conceptualized and coded regioinvent, and wrote the main manuscript text. G. Majeau-Bettez contributed to the ideation of some aspects. All authors reviewed the manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThe authors would like to thank Titouan Greffe for his assistance in collecting production volume data for minerals and metals and Matthieu Souttre for code testing and the interesting discussions.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eThe data that support the findings of this study are openly available in Zenodo at [https://doi.org/10.5281/zenodo.15474318](https:/doi.org/10.5281/zenodo.15474318) , \u0026ldquo;Trade data from BACI to-be-used with regioinvent\u0026rdquo; v4. The code and all other required data are available on Github at [https://github.com/CIRAIG/Regioinvent](https:/github.com/CIRAIG/Regioinvent) , v1.3.0.\u003c/p\u003e\u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eSupporting Information\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSupporting information is linked to this article on the \u003cem\u003eJIE\u003c/em\u003e website:\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupporting Information S1:\u0026nbsp;\u003c/strong\u003eThis supporting information provides the data plotted in Figure 4-4 and Figure 4-5, the full results for Table 4-1 and the table for the effect of cutoff on regioinvent.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAdrianto, Lugas Raka, Stephan Pfister, and Stefanie Hellweg. 2022. \u0026ldquo;Regionalized Life Cycle Inventories of Global Sulfidic Copper Tailings.\u0026rdquo; \u003cem\u003eEnvironmental Science \u0026amp; Technology\u003c/em\u003e 56(7):4553\u0026ndash;64. doi:10.1021/acs.est.1c01786.\u003c/li\u003e\n\u003cli\u003eAgez, Maxime. 2025a. \u0026ldquo;Regioinvent v1.3.0.\u0026rdquo;\u003c/li\u003e\n\u003cli\u003eAgez, Maxime. 2025b. \u003cem\u003eTrade data from BACI to-be-used with regioinvent.\u003c/em\u003e: Version 4. doi:https://doi.org/10.5281/zenodo.11583814.\u003c/li\u003e\n\u003cli\u003eAgez, Maxime, Elliot Muller, C\u0026eacute;cile Bulle, Laura Debarre, Georg Seitfudem, Ivan Viveros Santos, and Lisa Duval. 2024. \u003cem\u003eIMPACT World+ / a globally regionalized method for life cycle impact assessment\u003c/em\u003e: Version 2.1. doi:https://doi.org/10.5281/zenodo.14041258.\u003c/li\u003e\n\u003cli\u003eAlaux, Nicolas, Marcella Ruschi Mendes Saade, and Alexander Passer. 2024. \u0026ldquo;Inventory Regionalization of Background Data: Influence on Building Life Cycle Assessment and Carbon Reduction Strategies.\u0026rdquo; \u003cem\u003eJournal of Cleaner Production\u003c/em\u003e 459:142434. doi:10.1016/j.jclepro.2024.142434.\u003c/li\u003e\n\u003cli\u003eBaitz, Martin, Stefan Albrecht, Eloise Brauner, Clare Broadbent, Guy Castellan, Pierre Conrath, James Fava, Matthias Finkbeiner, Matthias Fischer, Pere Fullana I Palmer, Stephan Krinke, Christian Leroy, Oliver Loebel, Phil McKeown, Ivo Mersiowsky, Bernhard M\u0026ouml;ginger, Marcus Pfaadt, Gerald Rebitzer, Elmar Rother, Klaus Ruhland, Aafko Schanssema, and Ladji Tikana. 2013. \u0026ldquo;LCA\u0026rsquo;s Theory and Practice: Like Ebony and Ivory Living in Perfect Harmony?\u0026rdquo; \u003cem\u003eThe International Journal of Life Cycle Assessment\u003c/em\u003e 18(1):5\u0026ndash;13. doi:10.1007/s11367-012-0476-x.\u003c/li\u003e\n\u003cli\u003eBucciarelli, Pauline, Emmanuel Hache, and Val\u0026eacute;rie Mignon. 2025. \u0026ldquo;Evaluating Criticality of Strategic Metals: Are the Herfindahl\u0026ndash;Hirschman Index and Usual Concentration Thresholds Still Relevant?\u0026rdquo; \u003cem\u003eEnergy Economics\u003c/em\u003e 143:108208. doi:10.1016/j.eneco.2025.108208.\u003c/li\u003e\n\u003cli\u003eFAOSTAT. 2024. \u0026ldquo;FAOSTAT.\u0026rdquo; https://www.fao.org/faostat.\u003c/li\u003e\n\u003cli\u003eFry, Jacob, Keiichiro Kanemoto, Alastair Fraser, and Keisuke Nansai. 2024. \u0026ldquo;Global Freight Transport Emissions Responsibility.\u0026rdquo; \u003cem\u003eEnvironmental Science \u0026amp; Technology\u003c/em\u003e 58(43):19231\u0026ndash;42. doi:10.1021/acs.est.4c05658.\u003c/li\u003e\n\u003cli\u003eGaulier, Guillaume, and Soledad Zignago. 2010. \u0026ldquo;BACI: International Trade Database at the Product-Level (the 1994-2007 Version).\u0026rdquo; \u003cem\u003eSSRN Electronic Journal\u003c/em\u003e. doi:10.2139/ssrn.1994500.\u003c/li\u003e\n\u003cli\u003eHenderson, Andrew D., Anne Asselin-Balen\u0026ccedil;on, Martin C. Heller, Jasmina Burek, Daesoo Kim, Lindsay Lessard, Manuele Margni, Rosie Saad, Marty D. Matlock, Greg Thoma, Ying Wang, and Olivier Jolliet. 2023. \u0026ldquo;Spatialized Life Cycle Assessment of Fluid Milk Production and Consumption in the United States.\u0026rdquo; \u003cem\u003eSustainability\u003c/em\u003e 15(3):1890. doi:10.3390/su15031890.\u003c/li\u003e\n\u003cli\u003eHung, Christine Roxanne, Steve V\u0026ouml;ller, Maxime Agez, Guillaume Majeau-Bettez, and Anders Hammer Str\u0026oslash;mman. 2021. \u0026ldquo;Regionalized Climate Footprints of Battery Electric Vehicles in Europe.\u0026rdquo; \u003cem\u003eJournal of Cleaner Production\u003c/em\u003e 322:129052. doi:10.1016/j.jclepro.2021.129052.\u003c/li\u003e\n\u003cli\u003eLenzen, Manfred. 2011. \u0026ldquo;AGGREGATION VERSUS DISAGGREGATION IN INPUT\u0026ndash;OUTPUT ANALYSIS OF THE ENVIRONMENT.\u0026rdquo; \u003cem\u003eEconomic Systems Research\u003c/em\u003e 23(1):73\u0026ndash;89. doi:10.1080/09535314.2010.548793.\u003c/li\u003e\n\u003cli\u003eLi, Junjie, Yajun Tian, Yueling Zhang, and Kechang Xie. 2021. \u0026ldquo;Spatializing Environmental Footprint by Integrating Geographic Information System into Life Cycle Assessment: A Review and Practice Recommendations.\u0026rdquo; \u003cem\u003eJournal of Cleaner Production\u003c/em\u003e 323:129113. doi:10.1016/j.jclepro.2021.129113.\u003c/li\u003e\n\u003cli\u003eMutel, Chris. 2017. \u0026ldquo;Brightway: An Open Source Framework for Life Cycle Assessment.\u0026rdquo; \u003cem\u003eThe Journal of Open Source Software\u003c/em\u003e 2(12):236. doi:10.21105/joss.00236.\u003c/li\u003e\n\u003cli\u003eMutel, Chris, and Stefanie Hellweg. 2023. \u0026ldquo;Matrix-Based Methods for Regionalized Life Cycle Assessment.\u0026rdquo;\u003c/li\u003e\n\u003cli\u003eMutel, Christopher. 2012. \u0026ldquo;Framework and Tools for Regionalization in Life Cycle Assessment.\u0026rdquo; University of Iowa.\u003c/li\u003e\n\u003cli\u003eMutel, Christopher L., and Stefanie Hellweg. 2009. \u0026ldquo;Regionalized Life Cycle Assessment: Computational Methodology and Application to Inventory Databases.\u0026rdquo; \u003cem\u003eEnvironmental Science \u0026amp; Technology\u003c/em\u003e 43(15):5797\u0026ndash;5803. doi:10.1021/es803002j.\u003c/li\u003e\n\u003cli\u003ePatouillard, Laure, C\u0026eacute;cile Bulle, C\u0026eacute;cile Querleu, Dominique Maxime, Philippe Osset, and Manuele Margni. 2018. \u0026ldquo;Critical Review and Practical Recommendations to Integrate the Spatial Dimension into Life Cycle Assessment.\u0026rdquo; \u003cem\u003eJournal of Cleaner Production\u003c/em\u003e 177:398\u0026ndash;412. doi:10.1016/j.jclepro.2017.12.192.\u003c/li\u003e\n\u003cli\u003ePatouillard, Laure, Pierre Collet, Pascal Lesage, Pablo Tirado Seco, C\u0026eacute;cile Bulle, and Manuele Margni. 2019. \u0026ldquo;Prioritizing Regionalization Efforts in Life Cycle Assessment through Global Sensitivity Analysis: A Sector Meta-Analysis Based on Ecoinvent V3.\u0026rdquo; \u003cem\u003eThe International Journal of Life Cycle Assessment\u003c/em\u003e 24(12):2238\u0026ndash;54. doi:10.1007/s11367-019-01635-5.\u003c/li\u003e\n\u003cli\u003ePeng, Sidi, and Stephan Pfister. 2024. \u0026ldquo;Regionalizing the Supply Chain in Process Life Cycle Inventory with Multiregional Input\u0026ndash;Output Data: An Implementation for Ecoinvent with EXIOBASE.\u0026rdquo; \u003cem\u003eJournal of Industrial Ecology\u003c/em\u003e 28(4):680\u0026ndash;94. doi:10.1111/jiec.13491.\u003c/li\u003e\n\u003cli\u003eSacchi, R., T. Terlouw, K. Siala, A. Dirnaichner, C. Bauer, B. Cox, C. Mutel, V. Daioglou, and G. Luderer. 2022. \u0026ldquo;PRospective EnvironMental Impact asSEment (Premise): A Streamlined Approach to Producing Databases for Prospective Life Cycle Assessment Using Integrated Assessment Models.\u0026rdquo; \u003cem\u003eRenewable and Sustainable Energy Reviews\u003c/em\u003e 160:112311. doi:10.1016/j.rser.2022.112311.\u003c/li\u003e\n\u003cli\u003eSacchi, Romain. 2018. \u0026ldquo;A Trade-Based Method for Modelling Supply Markets in Consequential LCA Exemplified with Portland Cement and Bananas.\u0026rdquo; \u003cem\u003eThe International Journal of Life Cycle Assessment\u003c/em\u003e 23(10):1966\u0026ndash;80. doi:10.1007/s11367-017-1423-7.\u003c/li\u003e\n\u003cli\u003eSacchi, Romain, Romain Besseau, Paula P\u0026eacute;rez-L\u0026oacute;pez, and Isabelle Blanc. 2019. \u0026ldquo;Exploring Technologically, Temporally and Geographically-Sensitive Life Cycle Inventories for Wind Turbines: A Parameterized Model for Denmark.\u0026rdquo; \u003cem\u003eRenewable Energy\u003c/em\u003e 132:1238\u0026ndash;50. doi:10.1016/j.renene.2018.09.020.\u003c/li\u003e\n\u003cli\u003eSacchi, Romain, Alvaro Hahn Menacho, Georg Seitfudem, Maxime Agez, Joanna Schlesinger-Martinat, Anish Koyamparambath, Jair Santillan Saldivar, Philippe Loubet, and Christian Bauer. 2025. \u0026ldquo;Contextual LCIA without the Overhead: An Exchange-Based Framework for Flexible Impact Assessment.\u0026rdquo; \u003cem\u003eThe International Journal of Life Cycle Assessment\u003c/em\u003e. doi:10.1007/s11367-025-02551-7.\u003c/li\u003e\n\u003cli\u003eSany\u0026eacute; Mengual, E., J. Foschi, V. Orza, T. Sinkko, P. Wierzgala, and S. Sala. 2023. \u003cem\u003eConsumption Footprint: Methodological Overview\u003c/em\u003e. Joint Research Centre - European Comission. https://dx.doi.org/10.2760/413081.\u003c/li\u003e\n\u003cli\u003eSchenker, Vanessa, Christopher Oberschelp, and Stephan Pfister. 2022. \u0026ldquo;Regionalized Life Cycle Assessment of Present and Future Lithium Production for Li-Ion Batteries.\u0026rdquo; \u003cem\u003eResources, Conservation and Recycling\u003c/em\u003e 187:106611. doi:10.1016/j.resconrec.2022.106611.\u003c/li\u003e\n\u003cli\u003eSeitfudem, Georg, Markus Berger, Hannes M\u0026uuml;ller Schmied, and Anne‐Marie Boulay. 2025. \u0026ldquo;The Updated and Improved Method for Water Scarcity Impact Assessment in LCA, AWARE2.0.\u0026rdquo; \u003cem\u003eJournal of Industrial Ecology\u003c/em\u003e 29(3):891\u0026ndash;907. doi:10.1111/jiec.70023.\u003c/li\u003e\n\u003cli\u003eStadler, Konstantin, Richard Wood, Tatyana Bulavskaya, Carl-Johan S\u0026ouml;dersten, Moana Simas, Sarah Schmidt, Arkaitz Usubiaga, Jose Acosta-Fernandez, Jeroen Kuene, Martin Bruckner, Stefan Giljum, Stephan Lutter, Stefano Merciai, Jannick Schmidt, Michaela Theurl, Christoph Plutzar, Thomas Kastner, Nina Eisenmenger, Karl-Heinz Erb, Arjan Koning, and Arnold Tukker. 2021. \u003cem\u003eEXIOBASE 3 - v3.8.2. https://doi.org/10.5281/zenodo.5589597\u003c/em\u003e: Version v3.8.2. doi:https://doi.org/10.5281/zenodo.5589597.\u003c/li\u003e\n\u003cli\u003eSteubing, Bernhard, Gregor Wernet, J\u0026uuml;rgen Reinhard, Christian Bauer, and Emilia Moreno-Ruiz. 2016. \u0026ldquo;The Ecoinvent Database Version 3 (Part II): Analyzing LCA Results and Comparison to Version 2.\u0026rdquo; \u003cem\u003eThe International Journal of Life Cycle Assessment\u003c/em\u003e 21(9):1269\u0026ndash;81. doi:10.1007/s11367-016-1109-6.\u003c/li\u003e\n\u003cli\u003eWernet, Gregor, Christian Bauer, Bernhard Steubing, J\u0026uuml;rgen Reinhard, Emilia Moreno-Ruiz, and Bo Weidema. 2016. \u0026ldquo;The Ecoinvent Database Version 3 (Part I): Overview and Methodology.\u0026rdquo; \u003cem\u003eThe International Journal of Life Cycle Assessment\u003c/em\u003e 21(9):1218\u0026ndash;30. doi:10.1007/s11367-016-1087-8.\u003c/li\u003e\n\u003cli\u003eYang, Yi, and Reinout Heijungs. 2017. \u0026ldquo;A Generalized Computational Structure for Regional Life-Cycle Assessment.\u0026rdquo; \u003cem\u003eThe International Journal of Life Cycle Assessment\u003c/em\u003e 22(2):213\u0026ndash;21. doi:10.1007/s11367-016-1155-0.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Footnotes","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003e The Github page can be found at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/CIRAIG/Regioinvent\u003c/span\u003e\u003cspan address=\"https://github.com/CIRAIG/Regioinvent\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e The Github page of the wurst package: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/polca/wurst\u003c/span\u003e\u003cspan address=\"https://github.com/polca/wurst\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"journal-of-industrial-ecology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"44498","submissionUrl":"https://submission.springernature.com/new-submission/44498/3","title":"Journal of Industrial Ecology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"“industrial ecology”, “life cycle assessment (LCA)”, “regionalization”, “tool”, “international trade”, “ecoinvent”","lastPublishedDoi":"10.21203/rs.3.rs-8159063/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8159063/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eLife cycle inventory databases currently offer limited country-level coverage and often rely on outdated supply chain representations in their market processes. In this research, we combine ecoinvent with the BACI database\u0026mdash;which provides detailed international trade statistics\u0026mdash;to regionalize ecoinvent and model average supply chains based on up-to-date data. To achieve this, we automatically duplicate regional processes and adapt them to specific national contexts by modifying three key inputs in the ecoinvent database: electricity, heat, and municipal solid waste treatment. We also construct national consumption markets based on international trade and domestic production data to capture country-specific supply chain characteristics. These markets are fully relinked throughout the database, and elementary flows are spatialized. The resulting regioinvent adaptation of ecoinvent comprises 4,031 products regionalized across 225 countries, yielding 669,571 newly created processes. We assess the effect of regionalization by comparing each regionalized process with its corresponding original ecoinvent process. On average, regionalization introduces differences of 10.8% for climate change, 13.9% for human health, and 11.6% for ecosystem quality. Spatialization of elementary flows is found to exert a stronger influence on regionalized impact categories than inventory regionalization alone. Future developments of regioinvent will focus on addressing its current limitations, particularly the estimation of domestic production data and the contextualization of transportation within national consumption markets. Despite these limitations, we recommend that LCA practitioners use regioinvent, as it is expected to provide more accurate results than the traditional ecoinvent database.\u003c/p\u003e","manuscriptTitle":"Regioinvent: a regionalized version of ecoinvent integrating detailed trade data","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-01 12:50:39","doi":"10.21203/rs.3.rs-8159063/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-02-01T22:15:36+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-01-11T08:19:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"322604397270559472540676607008690304839","date":"2025-12-02T15:04:19+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-11-25T11:13:58+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-11-22T08:09:25+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-11-22T08:08:01+00:00","index":"","fulltext":""},{"type":"submitted","content":"Journal of Industrial Ecology","date":"2025-11-19T23:59:13+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"journal-of-industrial-ecology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"44498","submissionUrl":"https://submission.springernature.com/new-submission/44498/3","title":"Journal of Industrial Ecology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"5ff82e11-143b-48a3-9ab5-96631fd3bec6","owner":[],"postedDate":"December 1st, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-02-24T22:24:02+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-01 12:50:39","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8159063","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8159063","identity":"rs-8159063","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}