CO₂ Emissions Projections for 2100: A Comparative Machine Learning Study of U.S. and Multimodal Approach of Global Trends

CO₂ Emissions Projections for 2100: A Comparative Machine Learning Study of U.S. and Multimodal Approach of Global Trends | 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 Short Report CO₂ Emissions Projections for 2100: A Comparative Machine Learning Study of U.S. and Multimodal Approach of Global Trends Sanjana Murgod, Kartik Garg, Triveni Magadum, Vivek Yadav, Harshit Mittal, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5973641/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The continuous rise in CO₂ emissions is a major contributor to climate change, affecting ecosystems, economies, and public health. Predicting future emissions accurately is crucial for designing effective policies and mitigation strategies. This study explores multiple machine learning models for CO₂ emissions forecasting, comparing traditional methods like Support Vector Machines (SVM), Linear Regression, and Decision Trees with advanced deep learning techniques such as Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU), and Multi-Layer Perceptron (MLP). Using a time-series approach, we forecast emissions up to 2100 and assess model performance through key metrics like Mean Squared Error (MSE), Mean Absolute Error (MAE), and R² score. Our results indicate that deep learning models, especially LSTM and GRU, outperform traditional methods in capturing complex patterns and trends in emissions data. Additionally, we generate geospatial visualizations to highlight regions facing the highest risks. These insights provide valuable guidance for policymakers and environmental researchers, enabling data-driven decisions for emission reduction, resource allocation, and long-term sustainability planning in the fight against climate change. Artificial Intelligence and Machine Learning CO₂ Emissions Climate Change Machine Learning Support Vector Machines (SVM) Linear Regression Decision Trees Long Short-Term Memory (LSTM) Gated Recurrent Units (GRU) Multi-Layer Perceptron (MLP) Time-Series Forecasting Predictive Modeling Environmental Impact Figures Figure 1 Figure 2 1. Introduction The rise in CO₂ emissions is one of the biggest drivers of climate change, contributing to a range of environmental, economic, and social challenges around the world. As the impacts of climate change intensify, reducing greenhouse gas emissions has become an urgent global priority. CO₂, being the most common greenhouse gas, is primarily released through the burning of fossil fuels for energy production, transportation, and industrial activities. Its contribution to global warming is well understood, and addressing this issue is critical in avoiding the severe consequences of climate change, such as rising sea levels, extreme weather, and biodiversity loss (Elshaboury et al., 2021 ; Ghanbari et al., 2021 ; Lackner et al., 1995 ; Sakakura et al., 2007 ). Forecasting CO₂ emissions accurately is essential in the fight against climate change. By predicting future emissions, governments, organizations, and policymakers can make informed decisions about emission reduction strategies and plan for long-term sustainability (Garg et al., 2025 ; Mittal & Kushwaha, 2024a ). While traditional methods for predicting CO₂ emissions like statistical models have been useful, they often fail to capture the complex, non-linear relationships and time-dependent factors within the data. This is where machine learning (ML) and artificial intelligence (AI) shine. These technologies analyse large datasets, identify hidden patterns, and make predictions with minimal human input, offering a powerful tool for environmental forecasting (Mittal & Kushwaha, 2024b ; Varoquaux et al., 2015 ). This paper explores the potential of various machine learning models, from traditional algorithms like Support Vector Machines (SVM), Linear Regression, and Decision Trees to advanced deep learning techniques like Long Short-Term Memory (LSTM) networks, Gated Recurrent Units (GRU), and Multi-Layer Perceptron (MLP) networks (Geladi & Kowalski, 1986 ; Leng et al., 2007 ; Smith et al., 2013 ). We’ll assess their ability to forecast CO₂ emissions for both the United States and globally. In particular, deep learning models like LSTM and GRU are excellent at capturing long-term dependencies and cyclical patterns in time-series data, which is crucial for predicting emissions. Traditional models like SVM, Linear Regression, and Decision Trees, while not as effective in handling time-based patterns, still offer valuable insights and benchmarks due to their simplicity and computational efficiency. By comparing the performance of these machine learning models, we aim to identify the most effective methods for long-term CO₂ emissions forecasting (Keshtegar et al., 2018 ; Pekel, 2020 ; Rodriguez-Galiano et al., 2015 ). Each model's ability to learn from historical data and predict future emissions will be evaluated based on accuracy, computational efficiency, and ease of interpretation. This comparison is crucial for understanding how these models can inform strategies for reducing global emissions and meeting climate goals, such as those outlined in the Paris Agreement (Savaresi, 2016 ). The insights gained could help shape better environmental policies, optimize resource allocation, and ultimately contribute to a more sustainable future by providing reliable emissions projections that decision-makers can trust. 2. Methodology To accurately predict CO₂ emissions, we followed a structured approach that included data preprocessing, model selection, training, evaluation, and visualization. Our goal was to compare different machine learning models and determine which ones were best suited for forecasting emissions both in the United States and globally. Below is a breakdown of each step. 2.1 Data Preprocessing Before feeding the data into our models, we had to clean and prepare it, as raw data is rarely perfect. First, we handled any missing values by using methods like mean imputation or interpolation to fill in the gaps. Since the data came with different measurement units such as energy consumption in exajoules and CO₂ emissions in gigatonnes, we applied feature scaling to ensure all variables were on a comparable scale, preventing any single feature from dominating the model (Abbasi & El Hanandeh, 2016; Adams & Nsiah, 2019 ). We also performed feature selection, carefully choosing the most relevant factors influencing CO₂ emissions, so our models could focus on the variables that truly mattered. 2.2 Model Selection To forecast CO₂ emissions from multiple perspectives, we tested various machine learning models, including traditional and deep learning techniques. For regression tasks, we used Linear Regression as a baseline, Support Vector Regression (SVR) to minimize errors, and Decision Tree Regression to capture historical patterns (Gür, 2022 ; Wang et al., 2012 ). Multi-Layer Perceptron (MLP) Regression handled complex relationships, while LSTM and GRU Regression models effectively captured temporal dependencies in emissions data. For classification tasks, we categorized emissions into high, medium, or low risk using Support Vector Machine (SVM), Decision Tree, and Neural Network Classification. LSTM and GRU Classification models further improved trend detection (Pettersson et al., 2014 ; Sharma, 2011 ). By comparing these approaches, we aimed to identify the most accurate models for emissions forecasting and risk assessment. 2.3 Model Training Once we selected the models, we trained them using historical CO₂ emissions data, dividing the dataset into two parts. The training set, which made up 80% of the data, was used to teach the models how CO₂ emissions have changed over time. The remaining 20% of the data was reserved as the testing set to evaluate how well the models could perform on new, unseen data. For deep learning models like LSTM and GRU, we focused on fine-tuning key parameters such as the number of layers, neurons, batch size, and learning rate to ensure optimal performance. Training took place over multiple epochs, allowing the models to learn the complex patterns in the data and improve their accuracy (Chang, 2015 ; Kruppa et al., 2012 ; Lee et al., 2012 ). 2.4 Model Evaluation To assess how well each model performed, we used different evaluation metrics based on the task at hand. For the regression models, we focused on Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R² (coefficient of determination), which helped us understand how accurately the models predicted CO₂ emissions by comparing their outputs to actual values. On the other hand, for the classification models, we looked at metrics like accuracy, precision, recall, and F1-score to gauge how well the models were able to classify emissions trends into categories (B Ramsundar, 2018 ; Liu et al., 2021 ; Yang et al., 2023 ). After evaluating each model, we compared their performance to find the optimal balance between prediction accuracy and computational efficiency. This allowed us to identify the best-performing models that could deliver reliable results while maintaining practical feasibility. 2.5 Geospatial Visualization To make our findings more intuitive, we created geospatial visualizations that mapped CO₂ emissions projections across different regions. This visual representation helped highlight which areas are expected to experience the highest emissions growth and made our results more accessible for policymakers and researchers. 2.6 Forecasting Future CO₂ Emissions Using the best-performing models, we generated CO₂ emission forecasts up to the year 2100. These projections provided valuable insights into future emissions trends, helping stakeholders assess the impact of current environmental policies and develop more effective strategies for reducing greenhouse gases. 3. Results 3.1 Learning Model Comparison for U.S. CO₂ Emissions We tested multiple machine learning models to predict future CO₂ emissions for the U.S. up to the year 2100, including traditional models like Linear Regression, Support Vector Regression (SVR), Decision Tree, and Multi-Layer Perceptron (MLP), as well as deep learning models like LSTM and GRU. Our findings showed that deep learning models, particularly GRU and LSTM, performed better at capturing long-term patterns in emissions data, while traditional models like Decision Tree and Linear Regression struggled with complex trends. The future projections indicate a potential increase in CO₂ emissions, highlighting the need for policy interventions. 3.2 Global CO₂ Emissions Forecasting and Visualization We used Support Vector Regression (SVR) to predict CO₂ emissions for various countries for the year 2100. The model was trained on historical emissions data for each country, and the predictions were visualized on a global map. The results revealed stark regional differences, with industrialized nations showing higher projected emissions compared to developing regions. The color-coded geospatial visualization provides policymakers with a clear picture of future emissions trends, emphasizing the urgency of global emission reduction efforts. 4. Conclusions This study explored various machine learning models, comparing traditional methods like Linear Regression, Decision Trees, and Support Vector Machines with deep learning models such as LSTM and GRU to predict CO₂ emissions. The results demonstrated that deep learning models, particularly GRU and LSTM, outperformed traditional methods, offering more accurate forecasts by effectively handling sequential data. For the United States, the predictions suggested that CO₂ emissions may continue to rise without significant intervention. On a global scale, regional disparities were evident, with developed nations maintaining high emissions while emerging economies showed potential for rapid increases due to industrialization. These findings, visualized through geospatial maps, underscore the need for targeted climate action. Accurate CO₂ emission predictions are essential for shaping climate policies, setting reduction targets, and promoting sustainability. The study highlights the importance of proactive measures such as clean energy investment and stronger environmental policies. While machine learning offers valuable insights, real-world action is critical in the fight against climate change. By leveraging these advanced predictive models, policymakers can make informed decisions, driving global efforts toward a more sustainable and climate-resilient future. References Abbasi, M., & El Hanandeh, A. (2016). Forecasting municipal solid waste generation using artificial intelligence modelling approaches. Waste Management , 56 , 13–22. https://doi.org/10.1016/j.wasman.2016.05.018 Adams, S., & Nsiah, C. (2019). Reducing carbon dioxide emissions; Does renewable energy matter? Science of The Total Environment , 693 , 133288. https://doi.org/10.1016/j.scitotenv.2019.07.094 B Ramsundar. (2018). Molecular machine learning with DeepChem . Doctoral Dissertation Stanford University . Chang, N. (2015). Changing industrial structure to reduce carbon dioxide emissions: a Chinese application. Journal of Cleaner Production , 103 , 40–48. https://doi.org/10.1016/j.jclepro.2014.03.003 Elshaboury, N., Mohammed Abdelkader, E., Al-Sakkaf, A., & Alfalah, G. (2021). Predictive Analysis of Municipal Solid Waste Generation Using an Optimized Neural Network Model. Processes , 9 (11), 2045. https://doi.org/10.3390/pr9112045 Garg, K., Mittal, H., Yadav, V., Sehrawat, A., Shah, V., & Kushwaha, O. (2025). Municipal Solid Waste (MSW) Management Prediction Through Machine Learning Models: An Ensemble Tree Regressor Analysis . https://doi.org/10.21203/rs.3.rs-5834340/v1 Geladi, P., & Kowalski, B. R. (1986). Partial least-squares regression: a tutorial. Analytica Chimica Acta , 185 , 1–17. https://doi.org/10.1016/0003-2670(86)80028-9 Ghanbari, F., Kamalan, H., & Sarraf, A. (2021). An evolutionary machine learning approach for municipal solid waste generation estimation utilizing socioeconomic components. Arabian Journal of Geosciences , 14 (2), 92. https://doi.org/10.1007/s12517-020-06348-w Gür, T. M. (2022). Carbon Dioxide Emissions, Capture, Storage and Utilization: Review of Materials, Processes and Technologies. Progress in Energy and Combustion Science , 89 , 100965. https://doi.org/10.1016/j.pecs.2021.100965 Keshtegar, B., Mert, C., & Kisi, O. (2018). Comparison of four heuristic regression techniques in solar radiation modeling: Kriging method vs RSM, MARS and M5 model tree. Renewable and Sustainable Energy Reviews , 81 , 330–341. https://doi.org/10.1016/j.rser.2017.07.054 Kruppa, J., Ziegler, A., & König, I. R. (2012). Risk estimation and risk prediction using machine-learning methods. Human Genetics , 131 (10), 1639–1654. https://doi.org/10.1007/s00439-012-1194-y Lackner, K. S., Wendt, C., Butt, D. P., Joyce, E. L., & Sharp, D. H. (1995). Carbon dioxide disposal in carbonate minerals. Energy . https://doi.org/10.1016/0360-5442(95)00071-n Lee, M., Zhang, N., Zhang, N., & Zhang, N. (2012). Technical efficiency, shadow price of carbon dioxide emissions, and substitutability for energy in the Chinese manufacturing industries. Energy Economics . https://doi.org/10.1016/j.eneco.2012.06.023 Leng, L., Zhang, T., Kleinman, L., & Zhu, W. (2007). Ordinary least square regression, orthogonal regression, geometric mean regression and their applications in aerosol science. Journal of Physics: Conference Series , 78 , 012084. https://doi.org/10.1088/1742-6596/78/1/012084 Liu, Y., Zhang, D., Tang, Y., Zhang, Y., Chang, Y., & Zheng, J. (2021). Machine Learning-Enabled Design and Prediction of Protein Resistance on Self-Assembled Monolayers and Beyond. ACS Applied Materials & Interfaces , 13 (9), 11306–11319. https://doi.org/10.1021/acsami.1c00642 Mittal, H., & Kushwaha, O. S. (2024a). Biogas and Biofuel Production from Biowaste: Modelling and Simulation Study. In From Waste to Wealth (pp. 379–400). Springer Nature Singapore. https://doi.org/10.1007/978-981-99-7552-5_18 Mittal, H., & Kushwaha, O. S. (2024b). Machine Learning in Commercialized Coatings. In Functional Coatings (pp. 450–474). Wiley. https://doi.org/10.1002/9781394207305.ch17 Pekel, E. (2020). Estimation of soil moisture using decision tree regression. Theoretical and Applied Climatology , 139 (3–4), 1111–1119. https://doi.org/10.1007/s00704-019-03048-8 Pettersson, F., Maddison, D., Acar, S., & Soderholm, P. (2014). Convergence of Carbon Dioxide Emissions: A Review of the Literature. International Review of Environmental and Resource Economics , 7 (2), 141–178. https://doi.org/10.1561/101.00000059 Rodriguez-Galiano, V., Sanchez-Castillo, M., Chica-Olmo, M., & Chica-Rivas, M. (2015). Machine learning predictive models for mineral prospectivity: An evaluation of neural networks, random forest, regression trees and support vector machines. Ore Geology Reviews , 71 , 804–818. https://doi.org/10.1016/j.oregeorev.2015.01.001 Sakakura, T., Sakakura, T., Choi, J.-C., & Yasuda, H. (2007). Transformation of carbon dioxide. Chemical Reviews . https://doi.org/10.1021/cr068357u Savaresi, A. (2016). The Paris Agreement: a new beginning? Journal of Energy & Natural Resources Law , 34 (1), 16–26. https://doi.org/10.1080/02646811.2016.1133983 Sharma, S. S. (2011). Determinants of carbon dioxide emissions: Empirical evidence from 69 countries. Applied Energy , 88 (1), 376–382. https://doi.org/10.1016/j.apenergy.2010.07.022 Smith, P. F., Ganesh, S., & Liu, P. (2013). A comparison of random forest regression and multiple linear regression for prediction in neuroscience. Journal of Neuroscience Methods , 220 (1), 85–91. https://doi.org/10.1016/j.jneumeth.2013.08.024 Varoquaux, G., Varoquaux, G., Buitinck, L., Buitinck, L., Buitinck, L., Louppe, G., Louppe, G., Grisel, O., Grisel, O., Pedregosa, F., Pedregosa, F., Mueller, A., & Mueller, A. (2015). Scikit-learn: Machine Learning Without Learning the Machinery . https://doi.org/10.1145/2786984.2786995 Wang, Q., Zhou, P., & Zhou, D. (2012). Efficiency measurement with carbon dioxide emissions: The case of China. Applied Energy , 90 (1), 161–166. https://doi.org/10.1016/j.apenergy.2011.02.022 Yang, H., Zou, C., Huang, M., Zang, M., & Chen, S. (2023). High-fidelity computational modeling of scratch damage in automotive coatings with machine learning-driven identification of fracture parameters. Composite Structures , 316 , 117027. https://doi.org/10.1016/j.compstruct.2023.117027 Additional Declarations The authors declare no competing interests. Cite Share Download PDF Status: Posted Version 1 posted 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-5973641","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Short Report","associatedPublications":[],"authors":[{"id":412088923,"identity":"efcda37b-daa4-4db3-a732-d543920a674d","order_by":0,"name":"Sanjana Murgod","email":"","orcid":"","institution":"Pro H2Vis Solutions","correspondingAuthor":false,"prefix":"","firstName":"Sanjana","middleName":"","lastName":"Murgod","suffix":""},{"id":412088924,"identity":"a62e1f24-843a-4ee0-bd74-23f0be4272d0","order_by":1,"name":"Kartik Garg","email":"","orcid":"","institution":"Guru Gobind Singh Indraprastha University","correspondingAuthor":false,"prefix":"","firstName":"Kartik","middleName":"","lastName":"Garg","suffix":""},{"id":412088925,"identity":"43ce6606-b5ac-4c40-ae43-e3186cc702c1","order_by":2,"name":"Triveni Magadum","email":"","orcid":"","institution":"Pro H2Vis Solutions","correspondingAuthor":false,"prefix":"","firstName":"Triveni","middleName":"","lastName":"Magadum","suffix":""},{"id":412088926,"identity":"f1374c72-662e-4152-9158-fd271036f16e","order_by":3,"name":"Vivek Yadav","email":"","orcid":"","institution":"Guru Gobind Singh Indraprastha University","correspondingAuthor":false,"prefix":"","firstName":"Vivek","middleName":"","lastName":"Yadav","suffix":""},{"id":412088927,"identity":"3b4f0593-5742-4fce-9628-9aaae197a1fd","order_by":4,"name":"Harshit Mittal","email":"","orcid":"","institution":"KLE Technological University","correspondingAuthor":false,"prefix":"","firstName":"Harshit","middleName":"","lastName":"Mittal","suffix":""},{"id":412088928,"identity":"ab56aa72-7ab9-40c4-a08e-61a8b4c5d2a1","order_by":5,"name":"Omkar Singh Kushwaha","email":"data:image/png;base64,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","orcid":"","institution":"Indian Institute of Technology Madras","correspondingAuthor":true,"prefix":"","firstName":"Omkar","middleName":"Singh","lastName":"Kushwaha","suffix":""}],"badges":[],"createdAt":"2025-02-06 12:46:24","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-5973641/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5973641/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":75695715,"identity":"eb97174e-2399-4b98-9758-ca9fc1d2d755","added_by":"auto","created_at":"2025-02-07 08:15:44","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":109509,"visible":true,"origin":"","legend":"\u003cp\u003ePredicted Annual CO2 Emissions (U.S.)-Model Comparison (12 Models).\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-5973641/v1/4d9e06eb2a21072e2e6ce93e.png"},{"id":75695716,"identity":"cdb79e44-e119-4f68-9603-fd8838dc09b5","added_by":"auto","created_at":"2025-02-07 08:15:44","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":143180,"visible":true,"origin":"","legend":"\u003cp\u003eFuture CO2 Emissions Projection for 2100.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5973641/v1/9b66fc2ec70379a7bb83cc22.png"},{"id":75697471,"identity":"69faa031-9319-4be5-8c76-3abd046e1c59","added_by":"auto","created_at":"2025-02-07 08:39:49","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":639723,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5973641/v1/032502a8-21fb-41e6-b520-937a556b7d07.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eCO₂ Emissions Projections for 2100: A Comparative Machine Learning Study of U.S. and Multimodal Approach of Global Trends\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eThe rise in CO₂ emissions is one of the biggest drivers of climate change, contributing to a range of environmental, economic, and social challenges around the world. As the impacts of climate change intensify, reducing greenhouse gas emissions has become an urgent global priority. CO₂, being the most common greenhouse gas, is primarily released through the burning of fossil fuels for energy production, transportation, and industrial activities. Its contribution to global warming is well understood, and addressing this issue is critical in avoiding the severe consequences of climate change, such as rising sea levels, extreme weather, and biodiversity loss (Elshaboury et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Ghanbari et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Lackner et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e1995\u003c/span\u003e; Sakakura et al., \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). Forecasting CO₂ emissions accurately is essential in the fight against climate change. By predicting future emissions, governments, organizations, and policymakers can make informed decisions about emission reduction strategies and plan for long-term sustainability (Garg et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Mittal \u0026amp; Kushwaha, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2024a\u003c/span\u003e). While traditional methods for predicting CO₂ emissions like statistical models have been useful, they often fail to capture the complex, non-linear relationships and time-dependent factors within the data. This is where machine learning (ML) and artificial intelligence (AI) shine. These technologies analyse large datasets, identify hidden patterns, and make predictions with minimal human input, offering a powerful tool for environmental forecasting (Mittal \u0026amp; Kushwaha, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e2024b\u003c/span\u003e; Varoquaux et al., \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2015\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThis paper explores the potential of various machine learning models, from traditional algorithms like Support Vector Machines (SVM), Linear Regression, and Decision Trees to advanced deep learning techniques like Long Short-Term Memory (LSTM) networks, Gated Recurrent Units (GRU), and Multi-Layer Perceptron (MLP) networks (Geladi \u0026amp; Kowalski, \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e1986\u003c/span\u003e; Leng et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2007\u003c/span\u003e; Smith et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). We\u0026rsquo;ll assess their ability to forecast CO₂ emissions for both the United States and globally. In particular, deep learning models like LSTM and GRU are excellent at capturing long-term dependencies and cyclical patterns in time-series data, which is crucial for predicting emissions. Traditional models like SVM, Linear Regression, and Decision Trees, while not as effective in handling time-based patterns, still offer valuable insights and benchmarks due to their simplicity and computational efficiency.\u003c/p\u003e \u003cp\u003eBy comparing the performance of these machine learning models, we aim to identify the most effective methods for long-term CO₂ emissions forecasting (Keshtegar et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Pekel, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Rodriguez-Galiano et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Each model's ability to learn from historical data and predict future emissions will be evaluated based on accuracy, computational efficiency, and ease of interpretation. This comparison is crucial for understanding how these models can inform strategies for reducing global emissions and meeting climate goals, such as those outlined in the Paris Agreement (Savaresi, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). The insights gained could help shape better environmental policies, optimize resource allocation, and ultimately contribute to a more sustainable future by providing reliable emissions projections that decision-makers can trust.\u003c/p\u003e"},{"header":"2. Methodology","content":"\u003cp\u003eTo accurately predict CO₂ emissions, we followed a structured approach that included data preprocessing, model selection, training, evaluation, and visualization. Our goal was to compare different machine learning models and determine which ones were best suited for forecasting emissions both in the United States and globally. Below is a breakdown of each step.\u003c/p\u003e \u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Data Preprocessing\u003c/h2\u003e \u003cp\u003eBefore feeding the data into our models, we had to clean and prepare it, as raw data is rarely perfect. First, we handled any missing values by using methods like mean imputation or interpolation to fill in the gaps. Since the data came with different measurement units such as energy consumption in exajoules and CO₂ emissions in gigatonnes, we applied feature scaling to ensure all variables were on a comparable scale, preventing any single feature from dominating the model (Abbasi \u0026amp; El Hanandeh, 2016; Adams \u0026amp; Nsiah, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). We also performed feature selection, carefully choosing the most relevant factors influencing CO₂ emissions, so our models could focus on the variables that truly mattered.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Model Selection\u003c/h2\u003e \u003cp\u003eTo forecast CO₂ emissions from multiple perspectives, we tested various machine learning models, including traditional and deep learning techniques. For regression tasks, we used Linear Regression as a baseline, Support Vector Regression (SVR) to minimize errors, and Decision Tree Regression to capture historical patterns (G\u0026uuml;r, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Wang et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). Multi-Layer Perceptron (MLP) Regression handled complex relationships, while LSTM and GRU Regression models effectively captured temporal dependencies in emissions data. For classification tasks, we categorized emissions into high, medium, or low risk using Support Vector Machine (SVM), Decision Tree, and Neural Network Classification. LSTM and GRU Classification models further improved trend detection (Pettersson et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Sharma, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). By comparing these approaches, we aimed to identify the most accurate models for emissions forecasting and risk assessment.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Model Training\u003c/h2\u003e \u003cp\u003eOnce we selected the models, we trained them using historical CO₂ emissions data, dividing the dataset into two parts. The training set, which made up 80% of the data, was used to teach the models how CO₂ emissions have changed over time. The remaining 20% of the data was reserved as the testing set to evaluate how well the models could perform on new, unseen data. For deep learning models like LSTM and GRU, we focused on fine-tuning key parameters such as the number of layers, neurons, batch size, and learning rate to ensure optimal performance. Training took place over multiple epochs, allowing the models to learn the complex patterns in the data and improve their accuracy (Chang, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Kruppa et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Lee et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2012\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Model Evaluation\u003c/h2\u003e \u003cp\u003eTo assess how well each model performed, we used different evaluation metrics based on the task at hand. For the regression models, we focused on Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R\u0026sup2; (coefficient of determination), which helped us understand how accurately the models predicted CO₂ emissions by comparing their outputs to actual values. On the other hand, for the classification models, we looked at metrics like accuracy, precision, recall, and F1-score to gauge how well the models were able to classify emissions trends into categories (B Ramsundar, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Liu et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Yang et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). After evaluating each model, we compared their performance to find the optimal balance between prediction accuracy and computational efficiency. This allowed us to identify the best-performing models that could deliver reliable results while maintaining practical feasibility.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Geospatial Visualization\u003c/h2\u003e \u003cp\u003eTo make our findings more intuitive, we created geospatial visualizations that mapped CO₂ emissions projections across different regions. This visual representation helped highlight which areas are expected to experience the highest emissions growth and made our results more accessible for policymakers and researchers.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.6 Forecasting Future CO₂ Emissions\u003c/h2\u003e \u003cp\u003eUsing the best-performing models, we generated CO₂ emission forecasts up to the year 2100. These projections provided valuable insights into future emissions trends, helping stakeholders assess the impact of current environmental policies and develop more effective strategies for reducing greenhouse gases.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Learning Model Comparison for U.S. CO₂ Emissions\u003c/h2\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe tested multiple machine learning models to predict future CO₂ emissions for the U.S. up to the year 2100, including traditional models like Linear Regression, Support Vector Regression (SVR), Decision Tree, and Multi-Layer Perceptron (MLP), as well as deep learning models like LSTM and GRU. Our findings showed that deep learning models, particularly GRU and LSTM, performed better at capturing long-term patterns in emissions data, while traditional models like Decision Tree and Linear Regression struggled with complex trends. The future projections indicate a potential increase in CO₂ emissions, highlighting the need for policy interventions.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Global CO₂ Emissions Forecasting and Visualization\u003c/h2\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eWe used Support Vector Regression (SVR) to predict CO₂ emissions for various countries for the year 2100. The model was trained on historical emissions data for each country, and the predictions were visualized on a global map. The results revealed stark regional differences, with industrialized nations showing higher projected emissions compared to developing regions. The color-coded geospatial visualization provides policymakers with a clear picture of future emissions trends, emphasizing the urgency of global emission reduction efforts.\u003c/p\u003e \u003c/div\u003e"},{"header":"4. Conclusions","content":"\u003cp\u003eThis study explored various machine learning models, comparing traditional methods like Linear Regression, Decision Trees, and Support Vector Machines with deep learning models such as LSTM and GRU to predict CO₂ emissions. The results demonstrated that deep learning models, particularly GRU and LSTM, outperformed traditional methods, offering more accurate forecasts by effectively handling sequential data. For the United States, the predictions suggested that CO₂ emissions may continue to rise without significant intervention. On a global scale, regional disparities were evident, with developed nations maintaining high emissions while emerging economies showed potential for rapid increases due to industrialization. These findings, visualized through geospatial maps, underscore the need for targeted climate action. Accurate CO₂ emission predictions are essential for shaping climate policies, setting reduction targets, and promoting sustainability. The study highlights the importance of proactive measures such as clean energy investment and stronger environmental policies. While machine learning offers valuable insights, real-world action is critical in the fight against climate change. By leveraging these advanced predictive models, policymakers can make informed decisions, driving global efforts toward a more sustainable and climate-resilient future.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eAbbasi, M., \u0026amp; El Hanandeh, A. (2016). Forecasting municipal solid waste generation using artificial intelligence modelling approaches. \u003cem\u003eWaste Management\u003c/em\u003e, \u003cem\u003e56\u003c/em\u003e, 13\u0026ndash;22. https://doi.org/10.1016/j.wasman.2016.05.018\u003c/li\u003e\n \u003cli\u003eAdams, S., \u0026amp; Nsiah, C. (2019). Reducing carbon dioxide emissions; Does renewable energy matter? \u003cem\u003eScience of The Total Environment\u003c/em\u003e, \u003cem\u003e693\u003c/em\u003e, 133288. https://doi.org/10.1016/j.scitotenv.2019.07.094\u003c/li\u003e\n \u003cli\u003eB Ramsundar. (2018). Molecular machine learning with DeepChem . \u003cem\u003eDoctoral Dissertation Stanford University\u003c/em\u003e.\u003c/li\u003e\n \u003cli\u003eChang, N. (2015). Changing industrial structure to reduce carbon dioxide emissions: a Chinese application. \u003cem\u003eJournal of Cleaner Production\u003c/em\u003e, \u003cem\u003e103\u003c/em\u003e, 40\u0026ndash;48. https://doi.org/10.1016/j.jclepro.2014.03.003\u003c/li\u003e\n \u003cli\u003eElshaboury, N., Mohammed Abdelkader, E., Al-Sakkaf, A., \u0026amp; Alfalah, G. (2021). Predictive Analysis of Municipal Solid Waste Generation Using an Optimized Neural Network Model. \u003cem\u003eProcesses\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(11), 2045. https://doi.org/10.3390/pr9112045\u003c/li\u003e\n \u003cli\u003eGarg, K., Mittal, H., Yadav, V., Sehrawat, A., Shah, V., \u0026amp; Kushwaha, O. (2025). \u003cem\u003eMunicipal Solid Waste (MSW) Management Prediction Through Machine Learning Models: An Ensemble Tree Regressor Analysis\u003c/em\u003e. https://doi.org/10.21203/rs.3.rs-5834340/v1\u003c/li\u003e\n \u003cli\u003eGeladi, P., \u0026amp; Kowalski, B. R. (1986). Partial least-squares regression: a tutorial. \u003cem\u003eAnalytica Chimica Acta\u003c/em\u003e, \u003cem\u003e185\u003c/em\u003e, 1\u0026ndash;17. https://doi.org/10.1016/0003-2670(86)80028-9\u003c/li\u003e\n \u003cli\u003eGhanbari, F., Kamalan, H., \u0026amp; Sarraf, A. (2021). An evolutionary machine learning approach for municipal solid waste generation estimation utilizing socioeconomic components. \u003cem\u003eArabian Journal of Geosciences\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(2), 92. https://doi.org/10.1007/s12517-020-06348-w\u003c/li\u003e\n \u003cli\u003eG\u0026uuml;r, T. M. (2022). Carbon Dioxide Emissions, Capture, Storage and Utilization: Review of Materials, Processes and Technologies. \u003cem\u003eProgress in Energy and Combustion Science\u003c/em\u003e, \u003cem\u003e89\u003c/em\u003e, 100965. https://doi.org/10.1016/j.pecs.2021.100965\u003c/li\u003e\n \u003cli\u003eKeshtegar, B., Mert, C., \u0026amp; Kisi, O. (2018). Comparison of four heuristic regression techniques in solar radiation modeling: Kriging method vs RSM, MARS and M5 model tree. \u003cem\u003eRenewable and Sustainable Energy Reviews\u003c/em\u003e, \u003cem\u003e81\u003c/em\u003e, 330\u0026ndash;341. https://doi.org/10.1016/j.rser.2017.07.054\u003c/li\u003e\n \u003cli\u003eKruppa, J., Ziegler, A., \u0026amp; K\u0026ouml;nig, I. R. (2012). Risk estimation and risk prediction using machine-learning methods. \u003cem\u003eHuman Genetics\u003c/em\u003e, \u003cem\u003e131\u003c/em\u003e(10), 1639\u0026ndash;1654. https://doi.org/10.1007/s00439-012-1194-y\u003c/li\u003e\n \u003cli\u003eLackner, K. S., Wendt, C., Butt, D. P., Joyce, E. L., \u0026amp; Sharp, D. H. (1995). Carbon dioxide disposal in carbonate minerals. \u003cem\u003eEnergy\u003c/em\u003e. https://doi.org/10.1016/0360-5442(95)00071-n\u003c/li\u003e\n \u003cli\u003eLee, M., Zhang, N., Zhang, N., \u0026amp; Zhang, N. (2012). Technical efficiency, shadow price of carbon dioxide emissions, and substitutability for energy in the Chinese manufacturing industries. \u003cem\u003eEnergy Economics\u003c/em\u003e. https://doi.org/10.1016/j.eneco.2012.06.023\u003c/li\u003e\n \u003cli\u003eLeng, L., Zhang, T., Kleinman, L., \u0026amp; Zhu, W. (2007). Ordinary least square regression, orthogonal regression, geometric mean regression and their applications in aerosol science. \u003cem\u003eJournal of Physics: Conference Series\u003c/em\u003e, \u003cem\u003e78\u003c/em\u003e, 012084. https://doi.org/10.1088/1742-6596/78/1/012084\u003c/li\u003e\n \u003cli\u003eLiu, Y., Zhang, D., Tang, Y., Zhang, Y., Chang, Y., \u0026amp; Zheng, J. (2021). Machine Learning-Enabled Design and Prediction of Protein Resistance on Self-Assembled Monolayers and Beyond. \u003cem\u003eACS Applied Materials \u0026amp; Interfaces\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e(9), 11306\u0026ndash;11319. https://doi.org/10.1021/acsami.1c00642\u003c/li\u003e\n \u003cli\u003eMittal, H., \u0026amp; Kushwaha, O. S. (2024a). Biogas and Biofuel Production from Biowaste: Modelling and Simulation Study. In \u003cem\u003eFrom Waste to Wealth\u003c/em\u003e (pp. 379\u0026ndash;400). Springer Nature Singapore. https://doi.org/10.1007/978-981-99-7552-5_18\u003c/li\u003e\n \u003cli\u003eMittal, H., \u0026amp; Kushwaha, O. S. (2024b). Machine Learning in Commercialized Coatings. In \u003cem\u003eFunctional Coatings\u003c/em\u003e (pp. 450\u0026ndash;474). Wiley. https://doi.org/10.1002/9781394207305.ch17\u003c/li\u003e\n \u003cli\u003ePekel, E. (2020). Estimation of soil moisture using decision tree regression. \u003cem\u003eTheoretical and Applied Climatology\u003c/em\u003e, \u003cem\u003e139\u003c/em\u003e(3\u0026ndash;4), 1111\u0026ndash;1119. https://doi.org/10.1007/s00704-019-03048-8\u003c/li\u003e\n \u003cli\u003ePettersson, F., Maddison, D., Acar, S., \u0026amp; Soderholm, P. (2014). Convergence of Carbon Dioxide Emissions: A Review of the Literature. \u003cem\u003eInternational Review of Environmental and Resource Economics\u003c/em\u003e, \u003cem\u003e7\u003c/em\u003e(2), 141\u0026ndash;178. https://doi.org/10.1561/101.00000059\u003c/li\u003e\n \u003cli\u003eRodriguez-Galiano, V., Sanchez-Castillo, M., Chica-Olmo, M., \u0026amp; Chica-Rivas, M. (2015). Machine learning predictive models for mineral prospectivity: An evaluation of neural networks, random forest, regression trees and support vector machines. \u003cem\u003eOre Geology Reviews\u003c/em\u003e, \u003cem\u003e71\u003c/em\u003e, 804\u0026ndash;818. https://doi.org/10.1016/j.oregeorev.2015.01.001\u003c/li\u003e\n \u003cli\u003eSakakura, T., Sakakura, T., Choi, J.-C., \u0026amp; Yasuda, H. (2007). Transformation of carbon dioxide. \u003cem\u003eChemical Reviews\u003c/em\u003e. https://doi.org/10.1021/cr068357u\u003c/li\u003e\n \u003cli\u003eSavaresi, A. (2016). The Paris Agreement: a new beginning? \u003cem\u003eJournal of Energy \u0026amp; Natural Resources Law\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(1), 16\u0026ndash;26. https://doi.org/10.1080/02646811.2016.1133983\u003c/li\u003e\n \u003cli\u003eSharma, S. S. (2011). Determinants of carbon dioxide emissions: Empirical evidence from 69 countries. \u003cem\u003eApplied Energy\u003c/em\u003e, \u003cem\u003e88\u003c/em\u003e(1), 376\u0026ndash;382. https://doi.org/10.1016/j.apenergy.2010.07.022\u003c/li\u003e\n \u003cli\u003eSmith, P. F., Ganesh, S., \u0026amp; Liu, P. (2013). A comparison of random forest regression and multiple linear regression for prediction in neuroscience. \u003cem\u003eJournal of Neuroscience Methods\u003c/em\u003e, \u003cem\u003e220\u003c/em\u003e(1), 85\u0026ndash;91. https://doi.org/10.1016/j.jneumeth.2013.08.024\u003c/li\u003e\n \u003cli\u003eVaroquaux, G., Varoquaux, G., Buitinck, L., Buitinck, L., Buitinck, L., Louppe, G., Louppe, G., Grisel, O., Grisel, O., Pedregosa, F., Pedregosa, F., Mueller, A., \u0026amp; Mueller, A. (2015). \u003cem\u003eScikit-learn: Machine Learning Without Learning the Machinery\u003c/em\u003e. https://doi.org/10.1145/2786984.2786995\u003c/li\u003e\n \u003cli\u003eWang, Q., Zhou, P., \u0026amp; Zhou, D. (2012). Efficiency measurement with carbon dioxide emissions: The case of China. \u003cem\u003eApplied Energy\u003c/em\u003e, \u003cem\u003e90\u003c/em\u003e(1), 161\u0026ndash;166. https://doi.org/10.1016/j.apenergy.2011.02.022\u003c/li\u003e\n \u003cli\u003eYang, H., Zou, C., Huang, M., Zang, M., \u0026amp; Chen, S. (2023). High-fidelity computational modeling of scratch damage in automotive coatings with machine learning-driven identification of fracture parameters. \u003cem\u003eComposite Structures\u003c/em\u003e, \u003cem\u003e316\u003c/em\u003e, 117027. https://doi.org/10.1016/j.compstruct.2023.117027\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Guru Gobind Singh Indraprastha University","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"CO₂ Emissions, Climate Change, Machine Learning, Support Vector Machines (SVM), Linear Regression, Decision Trees, Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU), Multi-Layer Perceptron (MLP), Time-Series Forecasting, Predictive Modeling, Environmental Impact","lastPublishedDoi":"10.21203/rs.3.rs-5973641/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5973641/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe continuous rise in CO₂ emissions is a major contributor to climate change, affecting ecosystems, economies, and public health. Predicting future emissions accurately is crucial for designing effective policies and mitigation strategies. This study explores multiple machine learning models for CO₂ emissions forecasting, comparing traditional methods like Support Vector Machines (SVM), Linear Regression, and Decision Trees with advanced deep learning techniques such as Long Short-Term Memory (LSTM), Gated Recurrent Units (GRU), and Multi-Layer Perceptron (MLP). Using a time-series approach, we forecast emissions up to 2100 and assess model performance through key metrics like Mean Squared Error (MSE), Mean Absolute Error (MAE), and R\u0026sup2; score. Our results indicate that deep learning models, especially LSTM and GRU, outperform traditional methods in capturing complex patterns and trends in emissions data. Additionally, we generate geospatial visualizations to highlight regions facing the highest risks. These insights provide valuable guidance for policymakers and environmental researchers, enabling data-driven decisions for emission reduction, resource allocation, and long-term sustainability planning in the fight against climate change.\u003c/p\u003e","manuscriptTitle":"CO₂ Emissions Projections for 2100: A Comparative Machine Learning Study of U.S. and Multimodal Approach of Global Trends","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-02-07 08:15:40","doi":"10.21203/rs.3.rs-5973641/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0c7a7ba4-7c28-4681-9065-c69643b65365","owner":[],"postedDate":"February 7th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":43943280,"name":"Artificial Intelligence and Machine Learning"}],"tags":[],"updatedAt":"2025-02-07T08:15:40+00:00","versionOfRecord":[],"versionCreatedAt":"2025-02-07 08:15:40","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5973641","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5973641","identity":"rs-5973641","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}