Global warming and the Pacific Decadal Oscillation drive seasonally varying increases in extreme fire weather over the southwestern US

Global warming and the Pacific Decadal Oscillation drive seasonally varying increases in extreme fire weather over the southwestern US | 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 Article Global warming and the Pacific Decadal Oscillation drive seasonally varying increases in extreme fire weather over the southwestern US Jiale Lou, Youngji Joh, Darri Stubber, Thomas Delworth, Andrew Wittenberg This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7621598/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Wildfire activity has increased across the western United States (US), with evidence linking these trends to anthropogenic climate forcing. Here, we analyze the frequency, intensity, and drivers of fire weather extremes in the southwestern US over the past 83 years. These extremes exhibit strong seasonality: wintertime changes align with long-term warming trends, while summertime variations are dominated by decadal fluctuations modulated by the Pacific Decadal Oscillation (PDO). Fire risk has increased steadily during December–May, outside the traditional fire season, primarily due to rising temperatures and declining relative humidity (RH). In contrast, summertime fire risk—largely governed by humidity—follows a roughly V-shaped trajectory, decreasing from 1940–1980 and then rising from 1981–2022, in close alignment with decadal shifts in the PDO. Although Earth system models (ESMs) reliably capture temperature-related changes, they underestimate interdecadal variability in humidity, manifested by a weak representation of the observed PDO–RH relationship and low-RH extremes in summer. Earth and environmental sciences/Natural hazards Earth and environmental sciences/Climate sciences/Climate change Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Notable increase in wildfire incidents in the western United States (US) over the past few decades 1 – 8 highlight the urgent need to better understand wildfire behavior and to improve seasonal-to-interannual fire weather predictions. Wildfire occurrence and spread depend on three fundamental factors: ignition sources, vegetation availability, and conducive fire weather conditions 9 . Western US is one of many around the world that have been experiencing longer dry seasons 10 – 12 , increased temperatures 13 , decreased humidity 14 , and more extreme weather events such as dry lightning 15 , all of which exacerbate fire-prone conditions and make wildfires both more likely and more destructive. In addition to these climate-driven factors, non-climatic influences such as population and infrastructure growth in fire-prone areas have increased ignition sources and amplified wildfire impacts. Moreover, a century of fire suppression policies 16 , 17 may have led to the accumulation of combustible vegetation, further intensifying fire behavior. These anthropogenic and land-management factors may also contribute to the observed trends in wildfire frequency and intensity. However, current Earth system models do not explicitly simulate wildfire processes, raising important questions about the extent to which these non-climatic drivers are accounted for in projections of future fire risk. Extreme fire weather is typically assessed using indices of daily meteorological variables relevant to fuel moisture content 18 and fire behavior. A wide range of indices are used globally 18 , 19 , as summarized concisely in Table A1 of ref 20 . In this study, we focus on the Canadian Forest Fire Weather Index (FWI) System (see a summary in Supplementary Table 1) — one of the most widely used fire danger rating systems worldwide — to investigate fire weather extremes in the Southwestern US (SWUS). Jain et al. 13 show that the observed trend in FWI during the fire season is mainly driven by changes in atmospheric temperature and humidity. Other studies 1 , 21 , 22 have also highlighted the strong connection between temperature-controlled saturation vapor pressure, humidity-controlled actual vapor pressure, and forest burned area in the western US. Despite their strong relevance to wildfire activity, hydroclimatic processes—particularly humidity-related variables—are often misrepresented in Earth system models (ESM) 14 , 23 – 25 . For example, Simpson et al. 14 noted that the observed decline in water vapor over the SWUS regions during the past four decades (1980–2020) is not simulated in ESMs, which instead show an increasing trend consistent with theoretical expectations in accordance with Clausius–Clapeyron scaling. Additionally, Lou et al. 25 reported that while state-of-the-art ESMs can reasonably capture the month-to-month spatial patterns of hydroclimatic variables, such as vapor pressure deficit (VPD), over the SWUS, they struggle to represent the interdecadal variability of VPD. This limitation stems primarily from deficiencies in simulating actual vapor pressure, which is governed by humidity, as opposed to saturation vapor pressure, which is primarily temperature-dependent. Various thresholds have been used to define fire weather extremes, leading to different interpretations of similarevents 26 . For example, two common approaches to defining hot or warm days include: (1) a relative threshold, such as the 90th percentile of daily maximum temperature for a given calendar day over a baseline period—allowing such events to occur year-round with seasonally varying impacts; and (2) an absolute threshold, such as exceeding 35°C, which is often linked to potential health risks. In the context of fire weather extremes, existing literature highlights the use of either absolute (see Supplementary Table 1) or relative measures across both research and operational applications. Here, we begin our study by comparing the different threshold definitions to examine whether any consensus emerges, regardless of the definition used. Although extreme fire weather conditions during the fire season have been extensively studied, their seasonal evolution remains unclear. This study investigates the spatiotemporal characteristics and seasonal behavior of fire weather extremes from 1940 to 2022 (83 years), with a focus on identifying the key meteorological drivers underlying these changes. In addition, we examine the potential contributions of anthropogenic forcing and internal climate variability to the distinct seasonal patterns observed in fire weather extremes. Specifically, we address four key questions: (1) How have fire weather extremes evolved over time? (2) How do they vary across seasons? (3) What are the primary meteorological drivers of these seasonal changes? and (4) To what extent are these changes captured by ESMs? We use the ERA5 reanalysis 27 , 28 to analyze the frequency, intensity, and duration of daily fire weather extremes (Methods). We also examine the daily outputs from historical prescribed sea surface temperature (SST) simulations of the GFDL Seamless System for Prediction and EArth System Research (SPEAR) model 29 (Methods) to assess the model’s performance. The prescribed SST experiments are conducted using historical forcing and observation-based SST. This experimental design is crucial, as prescribing SST ensures that the models' seasonal and decadal variability remains in phase with real-world conditions. Nevertheless, similar model biases have also been identified in fully coupled historical and pre-industrial control simulations 14 , 25 . Changes in observed fire weather extremes We begin by comparing the spatiotemporal characteristics of extreme fire weather conditions across the US using both fixed and varying FWI thresholds. Figures 1 (a–c) present results based on a fixed threshold (FWI₉₀ = 44; see Methods), applied uniformly across space and time. Figure 1 a shows the spatial distribution of the relative frequency of extreme fire weather events lasting ≥ 5 days. The SWUS emerges as a prominent fire-prone region, with more than 20% of periods exceeding the extreme threshold (FWI > 30, based on Canadian Wildland Fire Information System rating scale; https://cwfis.cfs.nrcan.gc.ca/ ). Notably, the California–Arizona border consistently appears as a hotspot, with over half of the years classified as extreme. In contrast, when extremes are defined using varying thresholds (Fig. 1 d), they appear more evenly distributed across the southern US, including Texas, the Gulf Coast, and the Southeast—regions that show low activity under the fixed threshold. Still, the SWUS exhibits elevated frequencies, but the spatial coherence is reduced compared to the fixed-threshold map. The frequency scale is lower overall, with a maximum of around 3%, due to the thresholds being adapted to local norms. Despite spatial differences (Fig. 1 a vs. 1d), the temporal evolution of fire weather extreme frequencies across the SWUS shows a broadly consistent pattern under both fixed and varying thresholds, characterized by a V-shaped trajectory centered around 1980 (Figs. 1 b and e). Under the fixed-threshold definition (Fig. 1 b), the frequency of events lasting ≥ 5 days (black line) declines slightly from the 1940s to the late 1970s, followed by a pronounced increase through the 2010s, indicating a general increase in the frequency of fire weather extremes in recent decades. Although the relative frequency decreases with increasing event duration (grey lines), the temporal variability remains consistent across durations. Similarly, under the varying threshold (Fig. 1 e), the relative frequency of fire weather extremes (black line) exhibits a comparable V-shaped pattern, with post-1980 increases closely tracking the rise in annual burned area across California (red) and the US (orange). The temporal correlation between fire weather extremes in the SWUS and California burned area is strong during 1981–2022 ( r = 0.76; 0.68 after detrending), but weaker over the full 1940–2022 period ( r = 0.58), likely due to limited pre-satellite observations for wildfire burned areas. Nationwide, the correlation between fire weather extremes and total burned area from 1983–2022 is r = 0.67 ( p < 0.05). Since the early 1980s, California has experienced an average annual increase of approximately 30,000 burned acres, while the US as a whole has seen a rise of about 166,000 acres per year since 1983 — equivalent to the area of Rhode Island being burned every five years. In addition, we account for gap days 30 — defined as non-extreme periods of two days or fewer between extreme fire weather events—by merging them into a single continuous extreme event (see Methods and Ref. 30 ). Although Hobday et al. 30 is not a wildfire-specific study, it provides a methodological framework that is well-suited for application in this context. This approach better captures the persistence of fire-conducive conditions that may be briefly interrupted but still pose sustained risk. As expected, this method slightly increases the overall frequency due to the inclusion of non-extreme days, but it does not alter the broader temporal variability of fire weather extremes (compare the cyan and black lines in Fig. 1 e). For consistency, subsequent analyses exclude gap days unless otherwise noted. Fire risk in the SWUS exhibits strong seasonality (Figs. 1 c and f). Seasonal breakdowns show that while summer (June, July, and August; JJA) remains the peak season for fire weather extremes, both winter (December, January, and February; DJF) and spring (March, April, and May; MAM) have experienced increased fire risk in recent decades, particularly under the varying threshold (Fig. 1 f). Although the increase is statistically significant ( p < 0.05) only in spring under the varying threshold, it might reflect a tendency toward a lengthening of the fire season or at least an early onset. The fixed threshold (Fig. 1 c) also reflects this feature, though more subtly, likely due to its insensitivity to regional and seasonal variability. Owing to their similar behavior, winter and spring are grouped as the extended winter season (December to May), and summer and fall as the extended summer season (June to November) in this study. Meteorological drivers of fire risk in the SWUS To better understand the distinct seasonal behavior of fire risk between extended winter and summer, we examine the underlying meteorological drivers by decomposing daily FWI into its key components: temperature, relative humidity, and wind speed. Under both fixed and varying thresholds (Fig. 2 a), the recent increase in extended winter fire risk is primarily driven by rising temperatures and drier atmospheric conditions, as indicated by the increased frequency of low-RH extremes (i.e., days below the 10th percentile). In contrast, wind extremes have declined in recent decades, a trend that may be linked to ‘terrestrial stilling’ 31–33 — an observed long-term reduction in near-surface wind speeds over land, potentially driven by factors such as increased surface roughness from snowpack melting 34 or urbanization, and changes in atmospheric circulation. However, the exact causes of ‘terrestrial stilling’ remain uncertain. This decomposition illustrates that the observed increase in extended winter fire risk is largely driven by warming and drying trends rather than changes in wind, consistent with findings from Jain et al. 13 . Given the consistent behavior across both threshold definitions, the following analyses will focus on results derived using the varying threshold. The increase in temperature and decline in wind speed are well captured by the 30-member SPEAR prescribed-SST simulations (Fig. 2 b). However, the model struggles to reproduce the observed increase in low-RH extremes, showing only subtle change between the pre- and post-1980 periods instead. To tackle this discrepancy, we compare the observed and simulated time series of temperature and RH extremes across annual, extended winter, and extended summer periods in the SWUS (Figs. 2 c and d). The model (dashed lines and shaded spread) reasonably captures the increasing trend in temperature-related extremes, which largely reflects the global warming signal. Although the ensemble mean (dashed lines) slightly underestimates the rate of warming compared to observations (solid lines), the temporal evolution is broadly consistent across seasons. In contrast, the model struggles to reproduce the observed variability in low-RH extremes. Observations reveal a pronounced V-shaped pattern in low-RH extremes around the early 1980s, particularly during the extended summer season. This pattern is strongly correlated with the evolution of fire weather extremes (Fig. 1 e), with a temporal correlation of 0.91 for the annual time series, highlighting the close linkage between FWI and low-RH extremes. However, the simulations show only subtle variations and lack clear V-shaped evolutions (dashed lines in Fig. 2 c). This discrepancy is further supported by the spectral analysis (Fig. 2 d), where the model struggles to capture the observed interdecadal variability in RH extremes, despite closely matching the spectral characteristics of temperature extremes. These findings indicate a key limitation of the model in representing moisture-related processes that are critical for fire risk, even though it performs reasonably well in capturing temperature-driven trends. This discrepancy aligns with Simpson et al. 14 , who similarly reported substantial model-observation differences in annual humidity trends using multi-model ensembles. Seasonal differences in humidity extremes Low-RH extremes exhibit distinct seasonality (Fig. 3 ). Summertime RH extremes display a pronounced V-shaped trajectory around 1980 (Fig. 3 b), whereas wintertime RH extremes have increased continuously over the entire record (Fig. 3 a), with a notably faster rate of increase in recent decades (Fig. 2 c), indicating an acceleration in cold-season drying. Collectively, while summertime dryness in the SWUS lessened prior to 1980, wintertime conditions continued to dry during that same period, highlighting a remarkable seasonal contrast in RH trends. This contrast diminishes after 1980 as both seasons exhibit a more synchronized increase in dryness (Fig. 2 c), reflecting a growing seasonal coherence. Seasonal differences in low-RH extremes appear to be shaped by varying underlying mechanisms. Figure 3 a shows that wintertime RH is closely linked to temperature, with a temporal correlation of 0.61 in observations and 0.64 in model simulations. Interestingly, when the linear trend is removed from both variables, the RH-temperature correlation drops to 0.47 in observations but increases to 0.78 in the model, indicating that the model might overestimate synchronized co-variability between wintertime RH and temperature, while underestimating longer-term trends–especially for RH, as reflected in the relatively flat simulated RH trend (dashed cyan line in Fig. 3 a). In contrast, summertime RH exhibits a distinct V-shaped trend centered around 1980 (Fig. 2 c and Fig. 3 b), with RH extremes decreasing before 1980 and increasing afterward. This fluctuation-like evolution indicates that climate change alone cannot fully explain the observed changes. Instead, the timing of these changes in summertime RH extremes aligns well with the Pacific Decadal Oscillation (PDO), the dominant mode of SST variability in the North Pacific, which exhibits strong decadal variability. Summertime RH is significantly correlated with the PDO across annual, summer, and winter timescales, with the strongest correlation found with the preceding winter's PDO ( r = 0.60 in observations and 0.48 in model simulations). This suggests that the evolution of the wintertime PDO may establish favorable preconditions for summertime RH conditions to develop. Since RH is influenced not only by humidity but also by temperature, we further decompose RH into two components: the temperature-controlled saturation vapor pressure and the humidity-controlled actual vapor pressure. Consistent with earlier results, the model can reproduce the temperature-controlled component with temporal correlations reaching 0.78 for the annual time series (Fig. 3 c). However, it struggles to reproduce the pronounced decadal variation in actual vapor pressure—particularly the V-shaped pattern centered around 1980. Instead, the simulated actual vapor pressure trends remain relatively flat across all seasons before and after 1980, suggesting that the model underestimates decadal-scale humidity variability during summer. It becomes clear that the exacerbated fire risk in recent decades (post-1980) is primarily driven by the superposed effects of rising temperatures and declining RH across seasons. To further investigate the individual contributions of saturation and actual vapor pressure to overall RH evolution, we examine which linear weighting of saturation and actual vapor pressure best explains observed RH changes. Practically, we obtain the weights via multiple linear regression. Figure 3 e shows the temporal correlations between RH and various linear combinations of saturation and actual vapor pressures. The results indicate that summertime RH variability is largely controlled by the humidity-driven actual vapor pressure, accounting for approximately 70% of the overall RH variability. In contrast, wintertime RH is influenced more equally by both saturation and actual vapor pressures. The model reproduces the summer dominance of actual vapor pressure reasonably well (75%, as indicated by the vertical dashed line in Fig. 3 e), but overestimates its contribution during winter. This misrepresentation is effectively compensated by the model’s tendency to overestimate the synchronized co-variability between RH and temperature in winter (Fig. 3 a), leading to a seemingly reasonable simulation of overall wintertime RH despite inaccuracies in the underlying component contributions—an example of 'getting the right answer for the wrong reasons.' To further pinpoint the model discrepancies, we examine the joint probability density between daily saturation and actual vapor pressures. While the model captures the overall shape of the observed distribution (Supplementary Fig. 1), it displays notable seasonal biases (Fig. 4 ). As shown in Fig. 4 a, the wintertime biases organize into two oppositely signed dipoles with lobes elongated along the RH isolines, which is consistent with a Clausius-Clapeyron-like shift that operates at nearly constant RH, so percent errors in actual vapor pressure track those in saturation vapor pressure. Because the differences are largely confined along RH lines, positive and negative contributions within the low-RH wedge mostly cancel, helping to explain SPEAR’s relatively good performance for winter low-RH extremes. In contrast, extended summer (Fig. 4 b) shows a dipole structure that straddles the ~ 50% RH line: a negative lobe on the lower-RH (drier) side and a positive lobe on the higher-RH (more humid) side over a similar range of saturation vapor pressure. This across-RH structure indicates that SPEAR underrepresents dry conditions and overrepresents humid conditions, yielding an overall moist bias in summer. This discrepancy is not unique to the SPEAR model. To further evaluate model performance, we analyze a 10-member ensemble of prescribed-SST simulations from the Community Earth System Model version 2 (CESM2) 35 . CESM2’s joint-density difference during the extended winter is broader than SPEAR’s and shows a clear across-RH structure (Fig. 4 c): deficits on low-RH lines (≤ 30%) paired with excess on mid-RH lines (~ 30–60%) over a similar range of saturation vapor pressure. The overall underrepresentation of very dry states weakens cancellation within the low-RH wedge and increases the likelihood of underestimating winter low-RH extremes relative to SPEAR. For the extended summer (Fig. 4 d), CESM2 exhibits a broad negative swath across most RH values when saturation vapor pressure is less than 3 kPa, indicating an overall underrepresentation of those states. At warmer conditions, the distribution shows a clear overestimation of mid-RH states. This places more weight on “less-dry” hot days and, as a result, reduces the occurrence of very hot, dry states. These biases underscore persistent challenges in simulating joint moisture–temperature extremes. In common, both models underestimate dry conditions during the extended summer, while the differences highlight that SPEAR generally exhibits smaller and more spatially localized biases, indicating closer agreement with observations, particularly in winter. Moreover, both CESM2 and SPEAR can accurately simulate saturation vapor pressure, with CESM2 showing a temporal correlation of 0.82 with observations. However, CESM2 also markedly underestimates interdecadal variability of actual vapor pressure (Supplementary Fig. 2). Linking summer RH extremes to internal climate variability Both the frequency (Fig. 3 ) and aggregated intensity (Fig. 5 ) of summertime RH extremes correspond to the wintertime PDO. It is expected that aggregated RH intensity (i.e., the normalized sum of RH anomalies during extreme days) is highly correlated with its frequency ( r = 0.99 in Fig. 5 a). Although the model struggles to reproduce the observed V-shaped evolution, it still exhibits strong internal consistency, with correlations between low-RH frequency and its intensity generally exceeding 0.98 across ensemble members. Moreover, the aggregated intensity of summertime RH extremes is strongly correlated with the area-averaged summer RH anomalies over the SWUS (computed from daily values without restricting to extreme conditions; purple in Fig. 5 a), with a correlation of r = 0.89. Spatial patterns (Figs. 5 b–f) reveal that the positive phase of the preceding winter PDO—characterized by warmer-than-normal SST anomalies in the tropical Pacific and cooler-than-normal SST anomalies in the western and central North Pacific—is associated with reduced summer RH intensity. This phase coincides with a deepened Aleutian Low, a semi-permanent low-pressure system over the Aleutian Islands, which extends toward the western US at the surface (Fig. 5 b). In the mid- to upper troposphere, geopotential height anomalies exhibit a Pacific–North American (PNA)-like wave train pattern, featuring a weak positive node over western North America. The Aleutian Low remains evident across the pressure levels, with its core centered over the North Pacific. The vertical cross-section across western North America (Figs. 5 d-f) shows that the SWUS lies within a pronounced pressure gradient zone—situated between a deep high-pressure system over higher latitudes and a shallow, weak low-pressure anomaly over the SWUS and subtropics. This configuration is associated with enhanced easterly and northerly flow. The presence of a weak baroclinic structure—characterized by low surface pressure beneath an upper-level heat dome—creates a dynamically favorable environment for precipitation development, potentially preconditioning against summertime RH extremes. Similar spatial patterns are observed using the seasonal mean RH anomalies (Supplementary Fig. 3). Despite exhibiting much weaker spatial loadings, the model ensemble mean captures the general patterns of the PDO and Aleutian Low (Fig. 5 g). This weaker relationship may again be linked to the absence of interdecadal humidity variability in the model simulations. The difference maps (Figs. 5 h and i) further illustrate this underestimation of the winter PDO/PNA and summer RH relationships in the model. Although the model appears to capture the high-pressure system at higher latitudes and reproduces the associated anomalous easterly and northerly winds (Figs. 5 j–l), these features are noticeably weaker compared to observations. Collectively, the summer RH biases can be seen through the teleconnection patterns. With the absence of decadal humidity variability, wintertime preconditioning signal appears much weaker in the model. Discussion, summary, and conclusions Consensus holds that anthropogenic forcing has exacerbated fire risk in the western US in recent decades 1 – 3 , 6 , 36 . Our results are consistent with this view: climate change–driven global warming, combined with decreasing humidity, has contributed to hotter, drier, and more fire-prone conditions in the SWUS, particularly since the early 1980s. However, how fire risk may evolve across different seasons remains unclear. To address this, we analyzed daily FWI across seasons (Fig. 1 ) and found that potential fire risk has continued to increase during the non-traditional fire season (December through May), while showing a slight reduction during the warm seasons (June through November). These findings motivate further investigation into the underlying drivers of these seasonal shifts. Decomposing the FWI into its key meteorological components, we identified that temperature and RH are the main drivers of fire weather conditions across the SWUS, consistent with Jain et al. 13 . While temperature has exhibited an accelerated upward trend, RH, particularly in summer, follows a pronounced V-shaped trajectory centered around 1980. Specifically, the frequency of low-RH extremes declined before 1980 but began increasing thereafter, with the post-1980 upward trend nearly matching the magnitude of the earlier decline (Fig. 2 ). Meanwhile, RH extremes exhibit distinct seasonality, particularly before 1980. Summertime RH extremes show a continued decline, whereas wintertime RH displays a slight upward trend, implying the presence of different underlying mechanisms across seasons. Figure 6 summarizes the contributing factors underlying the distinct seasonal behavior of fire risk. The consistent upward trends in wintertime RH and temperature suggest that anthropogenic global warming is likely exacerbating fire risk outside the traditional fire season by increasing temperatures and enhancing atmospheric dryness. In contrast, internal climate variability—particularly the PDO—plays a dominant role in modulating summertime RH, resulting in pronounced decadal fluctuations in summertime fire risk. ESMs continue to exhibit substantial limitations in reproducing humidity-related variables 14 , 25 . Specifically, Simpson et al. 14 reported that most climate models fail to capture the observed trends in humidity over the past 40 years from 1980–2020. Even when driven by observed SSTs, these models do not fully resolve the discrepancy in humidity trends. Lou et al. 25 further argued that although climate models can reasonably reproduce the spatial patterns of VPD in the SWUS on seasonal time scales, they struggle to represent the decadal component of the leading VPD mode of variability. This deficiency primarily stems from biases in simulating water vapor, rather than temperature. Due to Clausius–Clapeyron scaling, the atmosphere's capacity to hold moisture is substantially lower in winter than in summer. During winter, both temperature and humidity contribute comparably to fire risk (Figs. 3 and 6 ). Although the model exhibits limited skill in capturing humidity variability, its relatively strong performance in simulating temperature partially offsets this deficiency, thereby improving overall model fidelity. As a result, wintertime relative humidity (RH) extremes are reasonably well represented. In contrast, during summer, actual vapor pressure—which is largely governed by humidity—emerges as the dominant driver of RH and fire weather conditions (Figs. 3 and 6 ). Because the model performs poorly in simulating moisture-related variables, this shortcoming becomes more critical and directly constrains model fidelity. In this case, the limiting factor—moisture—is also the most essential, and its deficiency cannot be compensated by the relatively well-simulated temperature field. This leads to a clear misrepresentation of summertime low-RH extremes, manifested in the absence of observed decadal humidity variability (i.e., the V-shaped trajectory in Figs. 2 and 3 ) and overly weak teleconnections with the wintertime PDO/PNA patterns (Fig. 4 ). Comparing observations and model simulations (Fig. 6 and Supplementary Fig. 4), we identify at least three key limitations that warrant further investigation. First, the model underestimates the relationship between summer humidity extremes and winter PDO variability ( r = 0.62 in observations vs. r = 0.38 in the model). Second, the model’s relative frequency of low-RH extremes is over-synchronized across seasons: the winter–summer relationship is much stronger than observed, with a correlation of r = 0.62 in the model versus r = 0.29 in observations (for summer leading). The full set of temporal correlations is shown in Fig. 6 and Supplementary Fig. 4. Third, the model exhibits an overly dominant role of humidity during winter, although the impact on RH is partially offset by the model’s strong performance in temperature-related quantities. Moreover, these model discrepancies are not unique to the model examined in this study. Similar biases—such as the underestimation of decadal humidity variability and the misrepresentation of long-term vapor pressure trend—have been documented across other ESMs as well 14 , 23 , 25 (also see Fig. 4 and Supplementary Fig. 2). This highlights a broader challenge in accurately simulating moisture-related processes and their associated extremes, particularly in regions sensitive to both internal variability and long-term climate change. Last, it is worth noting that this study does not explicitly examine whether the PDO itself is internally generated 37 or externally forced 38 . Resolving this question lies beyond our scope and remains an open direction for future work. Methods Observations and model simulations We use daily data from the ECMWF Reanalysis v5 (ERA5 27 ), including the Canadian Fire Weather Index (FWI 28 , 39 ; see Supplementary Table 1 for details), 2-meter dew point temperature, 2-meter air temperature, and 10-meter wind speed. In addition, we analyze large-scale oceanic and atmospheric conditions using monthly ERA5 data, including sea surface temperature (SST), sea level pressure (SLP), geopotential heights, and zonal and meridional wind components at various pressure levels. All daily variables are remapped to a regular 0.5° × 0.5° grid. To focus on fire weather extremes over land in the US, we apply a land mask to exclude fire weather conditions (e.g., temperature, humidity, and wind speed) over the ocean. This mask is derived from the ERA5 SST field, which is also remapped to a 0.5° × 0.5° grid. Leap days (i.e., Feb 29) are excluded from the daily data. Monthly variables are remapped to a regular 2° × 2° grid. The model simulations are taken from the Seamless System for Prediction and Earth System Research (SPEAR; Delworth et al. 29 ), a general circulation model designed by NOAA’s Geophysical Fluid Dynamics Laboratory (GFDL) to support both seasonal prediction and Earth system research. We use prescribed-SST simulations, in which observed SSTs are used as surface forcing. The prescribed-SST simulations have 30 ensemble members, with horizontal grid spacing of ~ 50 km for the atmosphere and land, and ~ 100 km for the ocean and sea ice. Daily outputs of 2-meter temperature, RH, and wind speed are used, with the same preprocessing methods applied as for the reanalysis data. Monthly outputs of SST, SLP, geopotential height, and zonal and meridional wind components at multiple pressure levels are obtained from the SPEAR prescribed-SST simulations, and processed similarly to the observational dataset. For comparison, we also analyze a 10-member ensemble of prescribed-SST simulations from the Community Earth System Model Version 2 (CESM2 35 ). Daily RH and temperature outputs from CESM2 are used and preprocessed using the same methodology as for the other datasets. Estimation of vapor pressure quantities Daily 2-meter dew point \(\:{T}_{d}\) and air temperature \(\:T\) from ERA5 reanalysis are used to calculate saturation vapor pressure \(\:{\varvec{e}}_{\varvec{s}}\left(T\right)\) and actual vapor pressure \(\:{\varvec{e}}_{\varvec{a}}\left({T}_{d}\right)\) . These are then used to derive RH, defined as the ratio \(\:RH={\varvec{e}}_{\varvec{a}}/{\varvec{e}}_{\varvec{s}}\) , and VPD, defined as the difference \(\:{VPD=\varvec{e}}_{\varvec{s}}-{\varvec{e}}_{\varvec{a}}\) . Here, \(\:{\varvec{e}}_{\varvec{s}}\) and \(\:{\varvec{e}}_{\varvec{a}}\) are computed following ref 40 . Daily outputs of RH and T from the models are used. Then, \(\:{\varvec{e}}_{\varvec{a}}\) is calculated as \(\:{e}_{s}\left(T\right)*RH/100\) . Here, Clausius–Clapeyron (CC) scaling describes how the saturation vapor pressure \(\:{e}_{s}\left(T\right)\) increases exponentially with temperature. In our previous study 25 , we compared different methods for calculating \(\:{\varvec{e}}_{\varvec{s}}\left(T\right)\) : one is based on the average of \(\:{\varvec{e}}_{\varvec{s}}\left({T}_{max}\right)\:and\:{\varvec{e}}_{\varvec{s}}\left({T}_{min}\right)\:\) , where \(\:{T}_{max}\) and \(\:{T}_{min}\) are the daily maximum and minimum temperatures, and another based on \(\:{\varvec{e}}_{\varvec{s}}\left({T}_{mean}\right)\) , where \(\:{T}_{mean}\) is the daily mean temperature. From a temporal correlation perspective, both methods yielded comparable results, with no notable differences reported ( r = 0.99 for monthly \(\:{\varvec{e}}_{\varvec{s}}\left(T\right)\) time series in the SWUS). Here, \(\:{\varvec{e}}_{\varvec{s}}\left(T\right)\) is computed using \(\:{\varvec{e}}_{\varvec{s}}\left({T}_{mean}\right)\) . The relative PDO index The relative PDO index is calculated as the area-averaged monthly relative SST anomalies over the North Pacific region (25°N–45°N, 140°E–145°W; Region 1 in Henley et al. 41 ). The term ‘relative’ refers to subtracting the near-global (60°S–60°N) mean SST anomalies from the traditional PDO time series. As noted by L’Heureux et al. 42 , relative Niño3.4 index is less sensitive to the choice of baseline period used to define climatology, and is more closely tied to anomalous tropical convection. Similarly, Tan et al. 43 adopted the relative SST framework to define a relative Atlantic Niño3.4 index. Supplementary Fig. 5 illustrates that both the near-global mean SST anomalies and the traditional PDO index are sensitive to various baseline choices — whether using a 30-year sliding climatology 44 , 45 , or fixed climatologies based on the first 30 years, last 30 years, or the full record. These fixed baseline approaches become especially problematic in a changing climate, where the background mean state itself is varying 42 . In contrast, the relative PDO index — by removing the near-global mean SST anomalies — shows much lower sensitivity to baseline selection. For this reason, we adopt the relative PDO index in this study. Extreme thresholds We employ two complementary approaches to define fire weather extremes: a fixed (or absolute) threshold and a varying threshold. The fixed threshold approach identifies values that exceed the 90th percentile (such as FWI, temperature, wind speed) or fall below the 10th percentile (such as RH, actual vapor pressure) of the entire dataset, calculated across all time steps and all geographical locations—specifically over land areas within our study domain (24°N–50°N, 126°W–64°W). This threshold is universally applicable and not dependent on specific seasons or locations. It allows us to understand the climatology and seasonal cycle of fire weather conditions, as well as to identify fire-prone regions. However, fixed thresholds do not account for regional climatological differences. Fire vulnerability and management capacity vary widely—what constitutes a moderate fire risk in fire-prone regions may be considered extreme in typically humid areas. To address this, we also analyze fire weather extremes using a varying threshold scheme, defined relative to season- and location-specific 90th or 10th percentile thresholds. Specifically, we adopt the threshold calculation from Hobday et al. 30 , in which thresholds are determined based on the 90th or 10th percentiles of daily values within an 11-day centered window, aggregated across all years. This varying threshold approach allows us to identify extremes that are unusual or disruptive relative to regional norms, regardless of their absolute magnitude. Although we refer to it as a varying threshold approach, it incorporates a fixed component: for example, fire weather extremes are initially identified using season- and location-specific 90th percentile FWI thresholds, but any values falling below the fixed lower tercile (33.3%) of FWI across the full dataset are excluded. Similarly, low-RH extremes are first defined using varying 10th percentile thresholds, but any values exceeding the fixed upper tercile (66.6%) of RH across the entire record are excluded. This hybrid approach avoids false positives in low-variability or low-risk regions. Although we describe our season- and location-specific threshold as a “varying threshold,” it should not be confused with the “shifting threshold” of Amaya et al. 46 , which explicitly removes the long-term background state (which is ocean warming in that case) before defining extremes. Our approach varies the threshold across season and space but does not subtract a long-term background, as the reference period is based on the full record. Effective fire management strategies often integrate both fixed- and varying- threshold fire risk assessments in the synergistic planning. This integration allows for robust baseline preparedness while maintaining the flexibility to respond to unexpected changes in risk levels. Climatology We follow Hobday et al. 30 to compute the daily climatology. First, we apply an 11-day centered moving average to smooth the daily values for each calendar day. Then, the smoothed values are averaged across the baseline years (1940–2022 in this study) to produce the climatological mean for each day of the year. For monthly outputs, we apply a 30-year moving baseline to account for changes in the background mean state. Differences resulting from the choice of baseline periods are discussed in the previous section (see Supplementary Fig. 5). Characteristics of fire weather extremes Relative frequency is defined as the ratio of extreme day counts to the total number of days within the respective period. Aggregated intensity is calculated as the sum of daily anomalies during extreme conditions. Unlike averaged intensity, aggregated intensity accounts for the spatial extent and cumulative impact of fire weather extremes. Duration is defined as the number of consecutive days meeting the extreme threshold. Following Ref. 30 , we also account for short breaks—referred to as 'gap days'—between extreme events. Specifically, if two or fewer non-extreme days separate periods of fire weather extremes with durations ≥ 5 days, these gap days are included in the event duration. For example, a sequence of five fire weather days, followed by two non-extreme days, and then another seven fire weather days is considered a single 14-day event. Since including gap days does not notably alert the extreme characteristics, we exclude them for most of our analysis. Extreme events are defined as periods of extreme conditions lasting five days or more. Declarations COMPETING INTERESTS The authors declare no competing interests. Author contributions J.L. performed the analyses and led the writing, with input from all authors. Y.J. and T.L.D. helped shape the scientific ideas and contributed to discussions and writing. A.T.W. refined the presentation of the results and interpretations. D.S. cross-validated some results and identified and helped correct a coding error that was important for interpreting the findings. Acknowledgement The authors thank Drs. Liwei Jia and John Krasting for their valuable comments and discussions that helped improve this manuscript. We also acknowledge William Cooke for generating the SPEAR model outputs. J.L. thanks Nathaniel Johnson for the discussion of the relative SST indices. This work was conducted by J.L. under award NA22OAR4050664d from the National Oceanic and Atmospheric Administration (NOAA), U.S. Department of Commerce. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the views of NOAA or the U.S. Department of Commerce. J.L. also acknowledges the use of NOAA’s Research and Development High-Performance Computing System (RDHPCS) and the NSF NCAR high-performance computing and storage resources, managed by the Computational and Information Systems Laboratory (CISL; doi: 10.5065/qx9a-pg09 ). Data availability This study utilizes multiple observational and model-based datasets relevant to fire weather conditions and wildfire activity. Daily Canadian Forest Service Fire Weather Index (FWI) values are obtained from ERA5 through the Copernicus Emergency Management Service Fire Danger System historical product, available at: https://cds.climate.copernicus.eu/cdsapp#!/dataset/cems-fire-historical?tab=overview . National wildfire statistics, including total wildland fires and acres burned in the U.S. from 1983 to 2022, are compiled by the National Interagency Fire Center and available at: https://www.nifc.gov/fire-information/statistics/wildfires . Spatial data for burned area in California are accessed through the California State Geoportal: https://gis.data.ca.gov/ . Additionally, CESM2 prescribed-SST simulations used in this study are publicly available through NCAR's CESM Climate Simulation Working Group at: https://www.cesm.ucar.edu/working-groups/climate/simulations/cam6-prescribed-sst . Code availability All R scripts used for analysis and figure generation will be made publicly available through a data repository upon acceptance of the manuscript. References Abatzoglou, J. T. & Williams, A. P. Impact of anthropogenic climate change on wildfire across western US forests. Proc. Natl. Acad. Sci. 113 , 11770–11775 (2016). Brown, P. T. et al. Climate warming increases extreme daily wildfire growth risk in California. Nature 621 , 760–766 (2023). Burke, M. et al. The changing risk and burden of wildfire in the United States. Proc. Natl. Acad. Sci. 118 , e2011048118 (2021). Dennison, P. E., Brewer, S. C., Arnold, J. D. & Moritz, M. A. Large wildfire trends in the western United States, 1984–2011. Geophys. Res. Lett. 41 , 2928–2933 (2014). Littell, J. S., McKenzie, D., Peterson, D. L. & Westerling, A. L. Climate and wildfire area burned in western U.S. ecoprovinces, 1916–2003. Ecol. Appl. 19 , 1003–1021 (2009). Williams, A. P. et al. Observed Impacts of Anthropogenic Climate Change on Wildfire in California. Earths Future 7 , 892–910 (2019). Zou, Y., Rasch, P. J., Wang, H., Xie, Z. & Zhang, R. Increasing large wildfires over the western United States linked to diminishing sea ice in the Arctic. Nat. Commun. 12 , 6048 (2021). Barbero, R., Abatzoglou, J. T., Larkin, S., Kolden, C. A. & Stocks, B. Climate change presents increased potential for very large fires in the contiguous United States. Int. J. Wildland Fire 24 , 892–899 (2015). Moritz, M. A. et al. Climate change and disruptions to global fire activity. Ecosphere 3 , art49 (2012). Jolly, W. M. et al. Climate-induced variations in global wildfire danger from 1979 to 2013. Nat. Commun. 6 , 7537 (2015). Jain, P., Wang, X. & Flannigan, M. D. Trend analysis of fire season length and extreme fire weather in North America between 1979 and 2015. Int. J. Wildland Fire 26 , 1009–1020 (2017). Weather and Climate Extreme Events in a Changing Climate. in Climate Change 2021 – The Physical Science Basis: Working Group I Contribution to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change (ed. Intergovernmental Panel on Climate Change (IPCC)) 1513–1766 (Cambridge University Press, Cambridge, 2023). doi:10.1017/9781009157896.013. Jain, P., Castellanos-Acuna, D., Coogan, S. C. P., Abatzoglou, J. T. & Flannigan, M. D. Observed increases in extreme fire weather driven by atmospheric humidity and temperature. Nat. Clim. Change 12 , 63–70 (2022). Simpson, I. R. et al. Observed humidity trends in dry regions contradict climate models. Proc. Natl. Acad. Sci. 121 , e2302480120 (2024). Pérez-Invernón, F. J., Gordillo-Vázquez, F. J., Huntrieser, H. & Jöckel, P. Variation of lightning-ignited wildfire patterns under climate change. Nat. Commun. 14 , 739 (2023). Oliveras Menor, I. et al. Integrated fire management as an adaptation and mitigation strategy to altered fire regimes. Commun. Earth Environ. 6 , 202 (2025). Mouillot, F. & Field, C. B. Fire history and the global carbon budget: a 1°× 1° fire history reconstruction for the 20th century. Glob. Change Biol. 11 , 398–420 (2005). Wagner, V. & E, C. Development and Structure of the Canadian Forest Fire Weather Index System . (1987). Fire behaviour in eucalypt forests / [by] A. G. McArthur - Trove. https://trove.nla.gov.au/work/21914760. Quilcaille, Y., Batibeniz, F., Ribeiro, A. F. S., Padrón, R. S. & Seneviratne, S. I. Fire weather index data under historical and shared socioeconomic pathway projections in the 6th phase of the Coupled Model Intercomparison Project from 1850 to 2100. Earth Syst. Sci. Data 15 , 2153–2177 (2023). Williams, A. P. et al. Growing impact of wildfire on western US water supply. Proc. Natl. Acad. Sci. U. S. A. 119 , e2114069119 (2022). Williams, A. P. et al. Causes and Implications of Extreme Atmospheric Moisture Demand during the Record-Breaking 2011 Wildfire Season in the Southwestern United States. J. Appl. Meteorol. Climatol. 53 , 2671–2684 (2014). Simpson, I. R. et al. Confronting Earth System Model trends with observations. Sci. Adv. 11 , eadt8035 (2025). Shaw, T. A. et al. Regional climate change: consensus, discrepancies, and ways forward. Front. Clim. 6 , (2024). Lou, J., Joh, Y. & Delworth, T. L. Identifying common precursors of decadal vapor pressure deficit variability in the Southwestern United States. (2025) doi:10.1175/JCLI-D-24-0296.1. van Oldenborgh, G. J. et al. Pathways and pitfalls in extreme event attribution. Clim. Change 166 , 1–27 (2021). Hersbach, H. et al. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc. 146 , 1999–2049 (2020). Vitolo, C. et al. ERA5-based global meteorological wildfire danger maps. Sci. Data 7 , 216 (2020). Delworth, T. L. et al. SPEAR: The Next Generation GFDL Modeling System for Seasonal to Multidecadal Prediction and Projection. J. Adv. Model. Earth Syst. 12 , e2019MS001895 (2020). Hobday, A. J. et al. A hierarchical approach to defining marine heatwaves. Prog. Oceanogr. 141 , 227–238 (2016). Deng, K. et al. Terrestrial Stilling Projected to Continue in the Northern Hemisphere Mid-Latitudes. Earths Future 10 , e2021EF002448 (2022). Klink, K. Trends in mean monthly maximum and minimum surface wind speeds in the coterminous United States, 1961 to 1990. Clim. Res. 13 , 193–205 (1999). McVicar, T. R. et al. Global review and synthesis of trends in observed terrestrial near-surface wind speeds: Implications for evaporation. J. Hydrol. 416–417 , 182–205 (2012). Vautard, R., Cattiaux, J., Yiou, P., Thépaut, J.-N. & Ciais, P. Northern Hemisphere atmospheric stilling partly attributed to an increase in surface roughness. Nat. Geosci. 3 , 756–761 (2010). Danabasoglu, G. et al. The Community Earth System Model Version 2 (CESM2). J. Adv. Model. Earth Syst. 12 , e2019MS001916 (2020). Kuo, Y.-N. et al. Recent southwestern US drought exacerbated by anthropogenic aerosols and tropical ocean warming. Nat. Geosci. 18 , 578–585 (2025). Newman, M. et al. The Pacific Decadal Oscillation, Revisited. J. Clim. 29 , 4399–4427 (2016). Klavans, J. M. et al. Human emissions drive recent trends in North Pacific climate variations. Nature 1–9 (2025) doi:10.1038/s41586-025-09368-2. Development and structure of the Canadian Forest Fire Weather Index System | Fire Research and Management Exchange System. https://www.frames.gov/catalog/6293. Bolton, D. The Computation of Equivalent Potential Temperature. (1980). Henley, B. J. et al. A Tripole Index for the Interdecadal Pacific Oscillation. Clim. Dyn. 45 , 3077–3090 (2015). L’Heureux, M. L. et al. A Relative Sea Surface Temperature Index for Classifying ENSO Events in a Changing Climate. J. Clim. 37 , 1197–1211 (2024). Tan, W., Liu, Y., Li, X., Johnson, N. C. & Hu, Z.-Z. Multi-Time Scale Variations in Atlantic Niño and a Relative Atlantic Niño Index. Geophys. Res. Lett. 50 , e2023GL106511 (2023). Lou, J., Newman, M. & Hoell, A. Multi-decadal variation of ENSO forecast skill since the late 1800s. Npj Clim. Atmospheric Sci. 6 , 1–14 (2023). Lou, J., Joh, Y., Delworth, T. L. & Jia, L. Identifying source of predictability for vapor pressure deficit variability in the southwestern United States. Npj Clim. Atmospheric Sci. 8 , 1–14 (2025). Amaya, D. J. et al. Marine heatwaves need clear definitions so coastal communities can adapt. Nature 616 , 29–32 (2023). Additional Declarations There is NO Competing Interest. Supplementary Files Supplementarymaterials.docx Supplementary materials Cite Share Download PDF Status: Under Review 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-7621598","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":516973890,"identity":"b0959d43-a8d3-427c-ba00-20f44dab072c","order_by":0,"name":"Jiale Lou","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA+ElEQVRIie2PsWrDMBCGTwjsDkrmdGjzChcyFVz6KhEpmRzw2NFgUBanXdOpr9AumSUO0ofQkqmzspnioVaGQiAoHgvVN4j/4D5+HUAk8gdhFTANBUCSVlo78OGyAhoQYCh20my8wns0HZXbUT4l4cdLCl/xvXaYSTXKkbLv7XjIgblDHvpYgmaDC6nEV0HLZztRHPj16zZ4C5JAkiqdv9OytqxTEj4IKqmj1iswQ7qr7UMPpasAr1w9dqGxso9SmBoXUyV2M7Mu7VxxVgVvmbx8fuybp+zmbVWRa1p73wXjDiGlPOlUx7c8t/nL+HRsg8uRSCTyT/kBLhdXI5Wh5iYAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0002-4014-5242","institution":"Princeton University","correspondingAuthor":true,"prefix":"","firstName":"Jiale","middleName":"","lastName":"Lou","suffix":""},{"id":516973891,"identity":"aa65dfc6-8c79-4ec0-b387-aa75049e1d68","order_by":1,"name":"Youngji Joh","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Youngji","middleName":"","lastName":"Joh","suffix":""},{"id":516973892,"identity":"6b94d03d-6ad9-430d-a9db-40fe5c6d2b59","order_by":2,"name":"Darri Stubber","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Darri","middleName":"","lastName":"Stubber","suffix":""},{"id":516973893,"identity":"e65ee671-da4f-47bf-bef3-e834f746a5e1","order_by":3,"name":"Thomas Delworth","email":"","orcid":"https://orcid.org/0000-0003-4865-5391","institution":"GFDL/NOAA","correspondingAuthor":false,"prefix":"","firstName":"Thomas","middleName":"","lastName":"Delworth","suffix":""},{"id":516973894,"identity":"053aa99d-41f6-41df-924a-431a163a8e57","order_by":4,"name":"Andrew Wittenberg","email":"","orcid":"https://orcid.org/0000-0003-1680-8963","institution":"NOAA GFDL","correspondingAuthor":false,"prefix":"","firstName":"Andrew","middleName":"","lastName":"Wittenberg","suffix":""}],"badges":[],"createdAt":"2025-09-15 14:21:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7621598/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7621598/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":93371516,"identity":"00c83e7d-301f-46d3-a8f9-57f5de76a670","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"jpg","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":907417,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/7de8833847a2e6cff8c51741.jpg"},{"id":93370732,"identity":"b8bf3e4a-ee5d-4260-96be-e84f40c3c7c1","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3107924,"visible":true,"origin":"","legend":"","description":"","filename":"Louetalfireweatherextremeclean.docx","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/c4e00c8ed23462903354eb61.docx"},{"id":93371518,"identity":"d0846b2d-dee9-47f9-aa80-0f2310bfd79f","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"jpg","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1354211,"visible":true,"origin":"","legend":"","description":"","filename":"Figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/8933ec18c91c33e659d36aa7.jpg"},{"id":93370734,"identity":"21992d1b-835f-4612-9412-8c88a38be182","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpg","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2289387,"visible":true,"origin":"","legend":"","description":"","filename":"Figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/2e90f3e3eb916fc427f83462.jpg"},{"id":93370731,"identity":"cd98529a-a377-471b-805f-ce5789ef9acc","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpg","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":463475,"visible":true,"origin":"","legend":"","description":"","filename":"Figure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/9dfa4901c8df8d2b534e35b1.jpg"},{"id":93371520,"identity":"7f99e428-c6e3-4377-a177-85b2b95b69c6","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"jpg","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1440201,"visible":true,"origin":"","legend":"","description":"","filename":"Figure5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/eeb9b7cc6d99ba49b0507a2f.jpg"},{"id":93372466,"identity":"093ba73c-9e33-48da-bb80-5848633a51b0","added_by":"auto","created_at":"2025-10-13 06:57:43","extension":"jpg","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1435221,"visible":true,"origin":"","legend":"","description":"","filename":"Figure6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/7d1bf4507a0cdddd56a6754c.jpg"},{"id":93370735,"identity":"ec4f1f0b-8a1a-43a8-9804-e08e5ececb0d","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"json","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6138,"visible":true,"origin":"","legend":"","description":"","filename":"COMMSENV254316.json","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/e73fdc80791f9268d2ee2996.json"},{"id":93370741,"identity":"5f3c351e-2a69-4c4d-b470-f2bd742ff1bf","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"docx","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7039835,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/e85da89de014896b95cb6f1f.docx"},{"id":93371767,"identity":"5097ee28-4a59-4728-b9db-d94221b655c3","added_by":"auto","created_at":"2025-10-13 06:49:43","extension":"xml","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":129857,"visible":true,"origin":"","legend":"","description":"","filename":"COMMSENV2543160enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/965bce4eb2560c9fa458b50a.xml"},{"id":93371523,"identity":"62842d75-b7f5-4f6c-99cc-83fbec749bba","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"jpg","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":907417,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/a24f94827ab574c1c3c37087.jpg"},{"id":93370740,"identity":"8757684b-9b2d-4197-a289-54897713055d","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpg","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1354211,"visible":true,"origin":"","legend":"","description":"","filename":"Figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/a9ed61515967095ebca01054.jpg"},{"id":93371526,"identity":"1bdd7f1e-594f-4dc3-84e7-6441294c11fe","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"jpg","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2289387,"visible":true,"origin":"","legend":"","description":"","filename":"Figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/5dc7e87d358a8e637a1533bb.jpg"},{"id":93370746,"identity":"59294830-3ef8-4713-a940-da88159d6fd1","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpg","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":463475,"visible":true,"origin":"","legend":"","description":"","filename":"Figure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/8c1923e096e172614eca6a2d.jpg"},{"id":93371524,"identity":"2ee96461-3034-4b5a-bbd2-844e3c0e83bf","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"jpg","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1440201,"visible":true,"origin":"","legend":"","description":"","filename":"Figure5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/1bf413e9e3f63b3702202592.jpg"},{"id":93370761,"identity":"a67169bc-2e7d-4bdf-aa42-a7ecc4ca6c9a","added_by":"auto","created_at":"2025-10-13 06:33:44","extension":"jpg","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1435221,"visible":true,"origin":"","legend":"","description":"","filename":"Figure6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/8fcc7edc8ab984074eb467ad.jpg"},{"id":93370739,"identity":"74dbd2e2-cef9-4a93-a699-64b304a37dc4","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpeg","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":369432,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/b504f03454d4c9e49a6d220f.jpeg"},{"id":93370749,"identity":"5a291fbe-fa0c-40b1-9313-407f8ee4d920","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpeg","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":453031,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/b9795d6e817ae27899634f01.jpeg"},{"id":93372467,"identity":"7c715bf0-00a2-461a-8327-f943fe0be558","added_by":"auto","created_at":"2025-10-13 06:57:43","extension":"jpeg","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":521538,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/88a68dd8572a7e8a763495e8.jpeg"},{"id":93370751,"identity":"0b922b9b-de65-4af9-b675-c33c3f419211","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpeg","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":463475,"visible":true,"origin":"","legend":"","description":"","filename":"Figure4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/e601a7a491949f61b0585441.jpeg"},{"id":93371528,"identity":"ca770bab-9bd2-47af-838f-f925bb35ba5b","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"jpeg","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":703684,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/1b1d6165c696c24b4601fe87.jpeg"},{"id":93370743,"identity":"8e6d4004-54b4-4fb5-962c-5f1c784014dd","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpeg","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":471057,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/818c9c71ab93932d8b421e86.jpeg"},{"id":93371769,"identity":"4ad085b6-53a1-4460-9e99-fe96da6c4648","added_by":"auto","created_at":"2025-10-13 06:49:43","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":154845,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/6d79ed56b0e46b2ffa251eda.png"},{"id":93370754,"identity":"58458e5b-9133-40ea-ad15-73b17cffff79","added_by":"auto","created_at":"2025-10-13 06:33:44","extension":"png","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":261494,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/03f5aff1ecef314157c64be2.png"},{"id":93371531,"identity":"66dc084c-3968-45c8-8566-7157c46c1e55","added_by":"auto","created_at":"2025-10-13 06:41:44","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":366901,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/da149247522f6b0958221429.png"},{"id":93371529,"identity":"8dd53147-0ff9-4499-bc5b-3c028a9377e1","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"png","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":87275,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/ed7e0731438fa95d195ae988.png"},{"id":93370752,"identity":"6a73bf2f-5445-4aed-9a3a-5eebe26b6eba","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":260251,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/484114794fde8db3f16dbbf9.png"},{"id":93371530,"identity":"b816fa02-8557-4014-9d3a-36da7b8956b3","added_by":"auto","created_at":"2025-10-13 06:41:44","extension":"png","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":178804,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure6.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/e61b4077f81b331d51e8d18b.png"},{"id":93370757,"identity":"cc7ce60e-5779-40f8-af84-ad7483833d8b","added_by":"auto","created_at":"2025-10-13 06:33:44","extension":"png","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":99696,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/84de52832fcc019bc813a786.png"},{"id":93371533,"identity":"4d58a5d9-75e8-4d40-b0b8-d3ffb825fd7d","added_by":"auto","created_at":"2025-10-13 06:41:44","extension":"png","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":145560,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/32bae423bd1434fcd991fd39.png"},{"id":93371532,"identity":"d62a6a81-fb37-4d93-84cc-010e7cc6e7c4","added_by":"auto","created_at":"2025-10-13 06:41:44","extension":"png","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":171075,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/9d37b9011dcb09504d28f682.png"},{"id":93370755,"identity":"202775ee-54d5-4ca1-8249-d73d9b0d0360","added_by":"auto","created_at":"2025-10-13 06:33:44","extension":"png","order_by":31,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":87275,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/2047c2d3fa8ff69cb697a961.png"},{"id":93370764,"identity":"a85729d2-738b-40ca-8ac9-0ad1be6fa860","added_by":"auto","created_at":"2025-10-13 06:33:44","extension":"png","order_by":32,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":233050,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/78bc371bcfe1f47fb8afcb57.png"},{"id":93370765,"identity":"1b7b5b18-c763-415a-ba0b-c5f19f15e3cb","added_by":"auto","created_at":"2025-10-13 06:33:44","extension":"png","order_by":33,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":113903,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/fe403a1006da4dc6071957de.png"},{"id":93370756,"identity":"f67bdb98-82c0-4d67-9107-79b723d6c62e","added_by":"auto","created_at":"2025-10-13 06:33:44","extension":"xml","order_by":34,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":128447,"visible":true,"origin":"","legend":"","description":"","filename":"COMMSENV2543160structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/44a5743eba4843f6530b4745.xml"},{"id":93370759,"identity":"fb4612fe-ab8d-4851-a221-21f753f62bb4","added_by":"auto","created_at":"2025-10-13 06:33:44","extension":"html","order_by":35,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":142935,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/8da59753e5c9e23fe05fa296.html"},{"id":93370725,"identity":"89a07255-b85a-4d9e-8d0d-389044669d72","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":907417,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRelative frequency of extreme fire weather conditions\u003c/strong\u003e. Panels (\u003cstrong\u003ea\u003c/strong\u003e–\u003cstrong\u003ec\u003c/strong\u003e) show fire weather extremes defined using a fixed threshold, where daily fire weather index (FWI) values exceed the 90th percentile (FWI₉₀ = 44). This universal threshold is independent of location and season, allowing consistent identification of climatologically fire-prone regions. Panels (\u003cstrong\u003ed\u003c/strong\u003e–\u003cstrong\u003ef\u003c/strong\u003e) use varying thresholds, where the 90th percentile is calculated separately for each location and season (see text for details). (\u003cstrong\u003ea\u003c/strong\u003e) Spatial distribution of the relative frequency of extreme fire weather events lasting longer than 5 days, based on the fixed threshold. The cyan box indicates the SWUS, including California, Nevada, Arizona, Utah, Colorado, and New Mexico. (\u003cstrong\u003eb\u003c/strong\u003e) Annual time series of the relative frequency of extreme fire weather events in the SWUS. Thin grey lines represent events of varying durations (from ≥1 to ≥10 days), with lighter shades indicating longer durations. The thick black line highlights events lasting longer than 5 days. (\u003cstrong\u003ec\u003c/strong\u003e) Seasonal relative frequency of fire weather extremes before 1980 (blue) and after 1980 (red). (\u003cstrong\u003ed\u003c/strong\u003e–\u003cstrong\u003ef\u003c/strong\u003e) Same as (\u003cstrong\u003ea\u003c/strong\u003e–\u003cstrong\u003ec\u003c/strong\u003e), but based on varying thresholds. In (\u003cstrong\u003ee\u003c/strong\u003e), red and orange lines show annual burned area in California and the entire US, respectively. The cyan line indicates the relative frequency of fire weather extremes accounting for gap days—non-extreme days (≤2 days) separating consecutive extreme events. These are included in a continuous sequence following the approach of Hobday et al. (2016). For example, a sequence of 5 extreme days, followed by 2 non-extreme days, and then 8 more extreme days is treated as a single 15-day extreme event. The relative frequency is calculated as the number of extreme days divided by the total number of calendar days in each respective period. In panel \u003cstrong\u003eb\u003c/strong\u003e, the decreasing trend prior to 1980 is not statistically significant (\u003cem\u003ep\u003c/em\u003e = 0.13), whereas the increasing trend after 1980 is statistically significant (\u003cem\u003ep\u003c/em\u003e\u0026lt; 0.05). In panel \u003cstrong\u003ec\u003c/strong\u003e, the linear trends are all statistically significant (\u003cem\u003ep\u003c/em\u003e\u0026lt; 0.05). A two-proportion \u003cem\u003ez\u003c/em\u003e-test was applied in panels \u003cstrong\u003ec\u003c/strong\u003e and \u003cstrong\u003ef\u003c/strong\u003e to assess whether the changes in occurrence rates before and after 1980 are statistically significant. Under the varying threshold definition, the increase is statistically significant (\u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05) in spring.\u003c/p\u003e","description":"","filename":"Figure1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/1fa7e8d2d19d7034fed20408.jpg"},{"id":93371515,"identity":"a75b535f-8a16-4afa-9ef7-e39be5bb91cb","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1354211,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eObserved and simulated changes in daily fire weather conditions across the SWUS\u003c/strong\u003e. (\u003cstrong\u003ea\u003c/strong\u003e) Observed relative frequency of extreme fire weather conditions associated with high maximum temperature, high minimum temperature, low relative humidity (RH), and strong wind speed, shown for the periods before 1980 (blue) and after 1980 (red). Bars above the zero-reference line represent extremes computed using fixed thresholds (threshold values labeled above each category), while bars below the zero line represent extremes based on varying thresholds. See text for the detailed description of the definition of fixed and varying thresholds. (\u003cstrong\u003eb\u003c/strong\u003e) Simulated relative frequency of extreme fire weather conditions associated with high temperature, low RH, and strong wind speed pre- and post-1980 using varying thresholds. The daily outputs were obtained from the 30-member SPEAR prescribed-SST simulations. The error bars denote the spread of the model simulations. (\u003cstrong\u003ec\u003c/strong\u003e) Temporal evolution of extreme temperature and RH conditions. Annual-mean and seasonal-mean relative frequencies are shown, based on the occurrence of extremes during the full year (black), extended winter (December–May; sky blue), and extended summer (June–November; deep pink). Solid lines denote observations; dashed lines indicate model simulations. Shaded areas represent the spread across the model ensembles. A quadratic polynomial trend is fitted to the temperature extremes over the full period, while linear trends are applied to RH extremes for the pre- and post-1980 subperiods. (\u003cstrong\u003ed\u003c/strong\u003e) Spectral analysis of the annual-mean and seasonal-mean relative frequencies shown in panel (\u003cstrong\u003ec\u003c/strong\u003e). Solid lines represent observations; dashed lines represent model simulations. Both axes are displayed on a linear–linear scale. Line colors correspond to the same seasons as in panel (\u003cstrong\u003ec\u003c/strong\u003e).\u003c/p\u003e","description":"","filename":"Figure2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/cff7b6caefdff1342c8b426e.jpg"},{"id":93370723,"identity":"70a6fbe2-04d4-41b2-b94a-434d9c26f0ac","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2289387,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTemporal features of extreme fire weather conditions in the SWUS\u003c/strong\u003e. (\u003cstrong\u003ea\u003c/strong\u003e) Normalized relative frequency of relative humidity (RH; cyan) and temperature (sky blue) extremes during the extended winter season (December–May). Solid lines indicate observations; dashed lines represent SPEAR prescribed-SST simulations. Shading denotes the spread across the 30-member SPEAR prescribed-SST ensembles. Linear trends are shown for both observed and ensemble-mean frequencies. (\u003cstrong\u003eb\u003c/strong\u003e) Normalized relative frequency of observed RH extremes (thick solid) during the extended summer season (June–November). Shading shows the ensemble spread from SPEAR prescribed-SST simulations; dashed line indicates the ensemble mean. Thin solid lines show normalized relative Pacific Decadal Oscillation (PDO) indices, where ‘relative’ refers to the removal of near-global mean (60°S–60°N) SST anomalies from the traditional PDO index (see main text). Linear fits are applied for the subperiods before and after 1980. (\u003cstrong\u003ec\u003c/strong\u003e–\u003cstrong\u003ed\u003c/strong\u003e) Relative frequency of saturation vapor pressure (\u003cstrong\u003ec\u003c/strong\u003e) and actual vapor pressure (\u003cstrong\u003ed\u003c/strong\u003e). Solid lines represent observations; dashed lines represent model simulations. Annual, extended summer, and extended winter values are shown in black, sky blue, and deep pink, respectively. Corresponding shadings denote the ensemble spread. A quadratic polynomial trend is fitted for the full period in panel (\u003cstrong\u003ec\u003c/strong\u003e), while linear trends are applied for the pre- and post-1980 subperiods in panel (\u003cstrong\u003ed\u003c/strong\u003e). (\u003cstrong\u003ee \u003c/strong\u003eand\u003cstrong\u003e f\u003c/strong\u003e) Sensitivity test showing the relative contributions of saturation and actual vapor pressure to the evolution of RH extremes (i.e., in panels \u003cstrong\u003ea\u003c/strong\u003e and \u003cstrong\u003eb\u003c/strong\u003e) during extended winter and summer. The x-axis denotes the fractional contribution from saturation vapor pressure; the y-axis indicates the correlation between the linearly combined vapor pressures and RH. Grey shading on the left marks the tercile where actual vapor pressure dominates and shading on the right marks the tercile where saturation vapor pressure dominates. Thin curves represent individual ensemble members from the 30-member SPEAR simulations. Vertical lines indicate the linear combinations at which maximum correlations occur for observations (solid) and model simulations (dashed).\u003c/p\u003e","description":"","filename":"Figure3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/80fbf7fbb96d6a8ead508594.jpg"},{"id":93370727,"identity":"7f77949f-7c40-4413-b277-4bec72f7ba9e","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":463475,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eKernel density differences between observations and model simulations\u003c/strong\u003e. Panels (\u003cstrong\u003ea\u003c/strong\u003e–\u003cstrong\u003ed\u003c/strong\u003e) show the differences in the joint probability distribution of daily saturation vapor pressure and actual vapor pressure between observations and model outputs over the Southwestern US: (\u003cstrong\u003ea\u003c/strong\u003e, \u003cstrong\u003eb\u003c/strong\u003e) for SPEAR and (\u003cstrong\u003ec\u003c/strong\u003e, \u003cstrong\u003ed\u003c/strong\u003e) for CESM2, during extended winter (\u003cstrong\u003ea\u003c/strong\u003e, \u003cstrong\u003ec\u003c/strong\u003e) and extended summer (\u003cstrong\u003eb\u003c/strong\u003e, \u003cstrong\u003ed\u003c/strong\u003e). \u0026nbsp;Radial dashed lines from the origin [0, 0] indicate constant relative humidity levels. Negative values (indicating model underestimation) are shown with dashed contours, while positive values (indicating overestimation) are shown with solid contours. Contour values range from –0.25 to 0.25 at intervals of 0.05 (excluding zero). The density units are expressed in 1/kPa².\u003c/p\u003e","description":"","filename":"Figure4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/53b4e2a2280aa4d2f0a8a3ed.jpg"},{"id":93371517,"identity":"e3140087-c24e-412b-8729-7c14dc8e023b","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1440201,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSpatiotemporal relationship between summer relative humidity (RH) extremes in the SWUS and preceding winter atmospheric and oceanic conditions\u003c/strong\u003e. (\u003cstrong\u003ea\u003c/strong\u003e) Normalized aggregated intensity (pink) and relative frequency (green) of RH extremes during boreal summer (June–August; JJA). Shaded areas denote the spread across the 30-member SPEAR prescribed-SST ensemble. Linear trends are fitted to the observed RH intensity (pink) for the subperiods before and after 1980. The temporal correlation between the sign-flipped observed intensity and frequency reaches 0.99, as shown in the bottom right corner. (\u003cstrong\u003eb\u003c/strong\u003e–\u003cstrong\u003ef\u003c/strong\u003e) Regression maps showing the relationship between normalized aggregated RH intensity in JJA and large-scale climatic conditions in preceding winter (December–February; DJF). (\u003cstrong\u003eb\u003c/strong\u003e) Winter sea surface temperature (SST; shading) and sea level pressure (SLP; contours); (c) Winter 500 hPa geopotential height (Z500; shading) and 200 hPa geopotential height (Z200; contours). (\u003cstrong\u003ed\u003c/strong\u003e–\u003cstrong\u003ef\u003c/strong\u003e) Longitudinal mean cross-sections showing regressions onto (\u003cstrong\u003ed\u003c/strong\u003e) geopotential heights, (\u003cstrong\u003ee\u003c/strong\u003e) zonal wind (U), and (\u003cstrong\u003ef\u003c/strong\u003e) meridional wind (V). Black boxes in (\u003cstrong\u003eb\u003c/strong\u003e) and (\u003cstrong\u003ec\u003c/strong\u003e) indicate the region used for cross-sections, while those in (\u003cstrong\u003ed\u003c/strong\u003e–\u003cstrong\u003ef\u003c/strong\u003e) show the latitudinal extent of the SWUS. (\u003cstrong\u003eg\u003c/strong\u003e) Same as (\u003cstrong\u003eb\u003c/strong\u003e), but based on the ensemble mean of the SPEAR prescribed-SST simulations. (\u003cstrong\u003eh\u003c/strong\u003e–\u003cstrong\u003ei\u003c/strong\u003e) Difference maps between the ensemble mean model simulations and observations (model minus observation) for (\u003cstrong\u003eh\u003c/strong\u003e) SST (shading) and SLP (contours), and (\u003cstrong\u003ei\u003c/strong\u003e) Z500 (shading) and Z200 (contours). (\u003cstrong\u003ej\u003c/strong\u003e–\u003cstrong\u003el\u003c/strong\u003e) Same as (\u003cstrong\u003ed\u003c/strong\u003e–\u003cstrong\u003ef\u003c/strong\u003e), but for the model ensemble mean (shading) and model–observation differences (contours). Thickened contour lines denote zero reference levels. All variables are normalized and therefore unitless.\u003c/p\u003e","description":"","filename":"Figure5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/b49f159f83e7661fd756be15.jpg"},{"id":93371519,"identity":"a80f38ed-b621-484c-a84e-4dac72cb00f4","added_by":"auto","created_at":"2025-10-13 06:41:43","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1435221,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSchematic of seasonal fire risk and its contributing factors.\u003c/strong\u003e The bar plot displays global mean air temperature anomalies relative to the 1901–2000 average, based on NOAA NCEI data. The spatial map shows the regression of wintertime SST anomalies onto the relative PDO time series for the period 1940–2022. Arrows indicate temporal relationships, with correlation values labeled. Key variables are enclosed in boxes. Blue elements represent extended winter, while red elements denote extended summer. Here, r\u003csub\u003ew \u003c/sub\u003edenotes extended winter (DJFMAM) and r\u003csub\u003es\u003c/sub\u003e denotes extended summer (JJASON). \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003ew1\u003c/em\u003e\u003c/sub\u003e and \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003es1\u003c/em\u003e\u003c/sub\u003e are the temporal correlations between global-mean air temperature (the bar plot) and the relative frequency of saturation vapor pressure over the SWUS, as illustrated in Fig. 3c, for winter and summer, respectively. \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003ew2\u003c/em\u003e\u003c/sub\u003e and \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003es2\u003c/em\u003e\u003c/sub\u003e are the temporal correlations between the annual PDO index (Fig. 3b) and the relative frequency of actual vapor pressure over the SWUS, as illustrated in Fig. 3d, for winter and summer, respectively. \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003ew3\u003c/em\u003e\u003c/sub\u003e and \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003es3\u003c/em\u003e\u003c/sub\u003e denote the temporal correlations between the relative frequency of saturation vapor pressure and the relative frequency of RH over the SWUS (Fig. 2c). \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003ew4\u003c/em\u003e\u003c/sub\u003e and \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003ew4 \u003c/em\u003e\u003c/sub\u003edenote the temporal correlations between the relative frequency of actual vapor pressure and the relative frequency of RH over the SWUS (Fig. 2c), for winter and summer, respectively. Finally, \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003ews\u003c/em\u003e\u003c/sub\u003e and \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003esw\u003c/em\u003e\u003c/sub\u003e (shown along the green dashed arrows) indicate cross-season correlations between low-RH extremes, where \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003ews \u003c/em\u003e\u003c/sub\u003euses the preceding winter to relate to summer extremes and \u003cem\u003er\u003c/em\u003e\u003csub\u003e\u003cem\u003esw\u003c/em\u003e\u003c/sub\u003e uses the preceding summer to relate to winter extremes.\u003c/p\u003e","description":"","filename":"Figure6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/7c1ad7a8e54840d75db9274b.jpg"},{"id":93372633,"identity":"a547d78d-03ed-44f4-8afc-fd34aaf7e494","added_by":"auto","created_at":"2025-10-13 07:05:47","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":8968663,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/9aef9d40-32f1-40c0-9db1-ec0fe1d99754.pdf"},{"id":93370729,"identity":"87874420-4825-4324-bb66-726112770e83","added_by":"auto","created_at":"2025-10-13 06:33:43","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":7039835,"visible":true,"origin":"","legend":"Supplementary materials","description":"","filename":"Supplementarymaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-7621598/v1/1031b52146a70e59f6897c52.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Global warming and the Pacific Decadal Oscillation drive seasonally varying increases in extreme fire weather over the southwestern US","fulltext":[{"header":"Introduction","content":"\u003cp\u003eNotable increase in wildfire incidents in the western United States (US) over the past few decades \u003csup\u003e\u003cspan additionalcitationids=\"CR2 CR3 CR4 CR5 CR6 CR7\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e highlight the urgent need to better understand wildfire behavior and to improve seasonal-to-interannual fire weather predictions. Wildfire occurrence and spread depend on three fundamental factors: ignition sources, vegetation availability, and conducive fire weather conditions \u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. Western US is one of many around the world that have been experiencing longer dry seasons \u003csup\u003e\u003cspan additionalcitationids=\"CR11\" citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e, increased temperatures \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e, decreased humidity \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e, and more extreme weather events such as dry lightning\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e, all of which exacerbate fire-prone conditions and make wildfires both more likely and more destructive.\u003c/p\u003e\u003cp\u003eIn addition to these climate-driven factors, non-climatic influences such as population and infrastructure growth in fire-prone areas have increased ignition sources and amplified wildfire impacts. Moreover, a century of fire suppression policies \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e may have led to the accumulation of combustible vegetation, further intensifying fire behavior. These anthropogenic and land-management factors may also contribute to the observed trends in wildfire frequency and intensity. However, current Earth system models do not explicitly simulate wildfire processes, raising important questions about the extent to which these non-climatic drivers are accounted for in projections of future fire risk.\u003c/p\u003e\u003cp\u003eExtreme fire weather is typically assessed using indices of daily meteorological variables relevant to fuel moisture content \u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e and fire behavior. A wide range of indices are used globally \u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e,\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e, as summarized concisely in Table A1 of ref \u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. In this study, we focus on the Canadian Forest Fire Weather Index (FWI) System (see a summary in Supplementary Table\u0026nbsp;1) \u0026mdash; one of the most widely used fire danger rating systems worldwide \u0026mdash; to investigate fire weather extremes in the Southwestern US (SWUS). Jain et al.\u003csup\u003e13\u003c/sup\u003e show that the observed trend in FWI during the fire season is mainly driven by changes in atmospheric temperature and humidity. Other studies \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e have also highlighted the strong connection between temperature-controlled saturation vapor pressure, humidity-controlled actual vapor pressure, and forest burned area in the western US.\u003c/p\u003e\u003cp\u003eDespite their strong relevance to wildfire activity, hydroclimatic processes\u0026mdash;particularly humidity-related variables\u0026mdash;are often misrepresented in Earth system models (ESM) \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan additionalcitationids=\"CR24\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. For example, Simpson et al.\u003csup\u003e14\u003c/sup\u003e noted that the observed decline in water vapor over the SWUS regions during the past four decades (1980\u0026ndash;2020) is not simulated in ESMs, which instead show an increasing trend consistent with theoretical expectations in accordance with Clausius\u0026ndash;Clapeyron scaling. Additionally, Lou et al.\u003csup\u003e25\u003c/sup\u003e reported that while state-of-the-art ESMs can reasonably capture the month-to-month spatial patterns of hydroclimatic variables, such as vapor pressure deficit (VPD), over the SWUS, they struggle to represent the interdecadal variability of VPD. This limitation stems primarily from deficiencies in simulating actual vapor pressure, which is governed by humidity, as opposed to saturation vapor pressure, which is primarily temperature-dependent.\u003c/p\u003e\u003cp\u003eVarious thresholds have been used to define fire weather extremes, leading to different interpretations of similarevents \u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. For example, two common approaches to defining hot or warm days include: (1) a relative threshold, such as the 90th percentile of daily maximum temperature for a given calendar day over a baseline period\u0026mdash;allowing such events to occur year-round with seasonally varying impacts; and (2) an absolute threshold, such as exceeding 35\u0026deg;C, which is often linked to potential health risks. In the context of fire weather extremes, existing literature highlights the use of either absolute (see Supplementary Table\u0026nbsp;1) or relative measures across both research and operational applications. Here, we begin our study by comparing the different threshold definitions to examine whether any consensus emerges, regardless of the definition used.\u003c/p\u003e\u003cp\u003eAlthough extreme fire weather conditions during the fire season have been extensively studied, their seasonal evolution remains unclear. This study investigates the spatiotemporal characteristics and seasonal behavior of fire weather extremes from 1940 to 2022 (83 years), with a focus on identifying the key meteorological drivers underlying these changes. In addition, we examine the potential contributions of anthropogenic forcing and internal climate variability to the distinct seasonal patterns observed in fire weather extremes. Specifically, we address four key questions: (1) How have fire weather extremes evolved over time? (2) How do they vary across seasons? (3) What are the primary meteorological drivers of these seasonal changes? and (4) To what extent are these changes captured by ESMs? We use the ERA5 reanalysis \u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e to analyze the frequency, intensity, and duration of daily fire weather extremes (Methods). We also examine the daily outputs from historical prescribed sea surface temperature (SST) simulations of the GFDL Seamless System for Prediction and EArth System Research (SPEAR) model \u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e (Methods) to assess the model\u0026rsquo;s performance. The prescribed SST experiments are conducted using historical forcing and observation-based SST. This experimental design is crucial, as prescribing SST ensures that the models' seasonal and decadal variability remains in phase with real-world conditions. Nevertheless, similar model biases have also been identified in fully coupled historical and pre-industrial control simulations \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\n\u003ch3\u003eChanges in observed fire weather extremes\u003c/h3\u003e\n\u003cp\u003eWe begin by comparing the spatiotemporal characteristics of extreme fire weather conditions across the US using both fixed and varying FWI thresholds. Figures\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e (a\u0026ndash;c) present results based on a fixed threshold (FWI₉₀ = 44; see Methods), applied uniformly across space and time. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea shows the spatial distribution of the relative frequency of extreme fire weather events lasting\u0026thinsp;\u0026ge;\u0026thinsp;5 days. The SWUS emerges as a prominent fire-prone region, with more than 20% of periods exceeding the extreme threshold (FWI\u0026thinsp;\u0026gt;\u0026thinsp;30, based on Canadian Wildland Fire Information System rating scale; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cwfis.cfs.nrcan.gc.ca/\u003c/span\u003e\u003cspan address=\"https://cwfis.cfs.nrcan.gc.ca/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Notably, the California\u0026ndash;Arizona border consistently appears as a hotspot, with over half of the years classified as extreme. In contrast, when extremes are defined using varying thresholds (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed), they appear more evenly distributed across the southern US, including Texas, the Gulf Coast, and the Southeast\u0026mdash;regions that show low activity under the fixed threshold. Still, the SWUS exhibits elevated frequencies, but the spatial coherence is reduced compared to the fixed-threshold map. The frequency scale is lower overall, with a maximum of around 3%, due to the thresholds being adapted to local norms.\u003c/p\u003e\u003cp\u003eDespite spatial differences (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea vs. 1d), the temporal evolution of fire weather extreme frequencies across the SWUS shows a broadly consistent pattern under both fixed and varying thresholds, characterized by a V-shaped trajectory centered around 1980 (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb and e). Under the fixed-threshold definition (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb), the frequency of events lasting\u0026thinsp;\u0026ge;\u0026thinsp;5 days (black line) declines slightly from the 1940s to the late 1970s, followed by a pronounced increase through the 2010s, indicating a general increase in the frequency of fire weather extremes in recent decades. Although the relative frequency decreases with increasing event duration (grey lines), the temporal variability remains consistent across durations.\u003c/p\u003e\u003cp\u003eSimilarly, under the varying threshold (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ee), the relative frequency of fire weather extremes (black line) exhibits a comparable V-shaped pattern, with post-1980 increases closely tracking the rise in annual burned area across California (red) and the US (orange). The temporal correlation between fire weather extremes in the SWUS and California burned area is strong during 1981\u0026ndash;2022 (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.76; 0.68 after detrending), but weaker over the full 1940\u0026ndash;2022 period (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.58), likely due to limited pre-satellite observations for wildfire burned areas. Nationwide, the correlation between fire weather extremes and total burned area from 1983\u0026ndash;2022 is \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.67 (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Since the early 1980s, California has experienced an average annual increase of approximately 30,000 burned acres, while the US as a whole has seen a rise of about 166,000 acres per year since 1983 \u0026mdash; equivalent to the area of Rhode Island being burned every five years. In addition, we account for gap days \u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e \u0026mdash; defined as non-extreme periods of two days or fewer between extreme fire weather events\u0026mdash;by merging them into a single continuous extreme event (see Methods and Ref. \u003csup\u003e30\u003c/sup\u003e). Although Hobday et al. \u003csup\u003e30\u003c/sup\u003e is not a wildfire-specific study, it provides a methodological framework that is well-suited for application in this context. This approach better captures the persistence of fire-conducive conditions that may be briefly interrupted but still pose sustained risk. As expected, this method slightly increases the overall frequency due to the inclusion of non-extreme days, but it does not alter the broader temporal variability of fire weather extremes (compare the cyan and black lines in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ee). For consistency, subsequent analyses exclude gap days unless otherwise noted.\u003c/p\u003e\u003cp\u003eFire risk in the SWUS exhibits strong seasonality (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec and f). Seasonal breakdowns show that while summer (June, July, and August; JJA) remains the peak season for fire weather extremes, both winter (December, January, and February; DJF) and spring (March, April, and May; MAM) have experienced increased fire risk in recent decades, particularly under the varying threshold (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ef). Although the increase is statistically significant (\u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05) only in spring under the varying threshold, it might reflect a tendency toward a lengthening of the fire season or at least an early onset. The fixed threshold (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec) also reflects this feature, though more subtly, likely due to its insensitivity to regional and seasonal variability. Owing to their similar behavior, winter and spring are grouped as the extended winter season (December to May), and summer and fall as the extended summer season (June to November) in this study.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eMeteorological drivers of fire risk in the SWUS\u003c/h2\u003e\u003cp\u003eTo better understand the distinct seasonal behavior of fire risk between extended winter and summer, we examine the underlying meteorological drivers by decomposing daily FWI into its key components: temperature, relative humidity, and wind speed. Under both fixed and varying thresholds (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea), the recent increase in extended winter fire risk is primarily driven by rising temperatures and drier atmospheric conditions, as indicated by the increased frequency of low-RH extremes (i.e., days below the 10th percentile). In contrast, wind extremes have declined in recent decades, a trend that may be linked to \u0026lsquo;terrestrial stilling\u0026rsquo; \u003csup\u003e31\u0026ndash;33\u003c/sup\u003e \u0026mdash; an observed long-term reduction in near-surface wind speeds over land, potentially driven by factors such as increased surface roughness from snowpack melting \u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e or urbanization, and changes in atmospheric circulation. However, the exact causes of \u0026lsquo;terrestrial stilling\u0026rsquo; remain uncertain. This decomposition illustrates that the observed increase in extended winter fire risk is largely driven by warming and drying trends rather than changes in wind, consistent with findings from Jain et al.\u003csup\u003e13\u003c/sup\u003e. Given the consistent behavior across both threshold definitions, the following analyses will focus on results derived using the varying threshold.\u003c/p\u003e\u003cp\u003eThe increase in temperature and decline in wind speed are well captured by the 30-member SPEAR prescribed-SST simulations (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb). However, the model struggles to reproduce the observed increase in low-RH extremes, showing only subtle change between the pre- and post-1980 periods instead. To tackle this discrepancy, we compare the observed and simulated time series of temperature and RH extremes across annual, extended winter, and extended summer periods in the SWUS (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec and d). The model (dashed lines and shaded spread) reasonably captures the increasing trend in temperature-related extremes, which largely reflects the global warming signal. Although the ensemble mean (dashed lines) slightly underestimates the rate of warming compared to observations (solid lines), the temporal evolution is broadly consistent across seasons.\u003c/p\u003e\u003cp\u003eIn contrast, the model struggles to reproduce the observed variability in low-RH extremes. Observations reveal a pronounced V-shaped pattern in low-RH extremes around the early 1980s, particularly during the extended summer season. This pattern is strongly correlated with the evolution of fire weather extremes (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ee), with a temporal correlation of 0.91 for the annual time series, highlighting the close linkage between FWI and low-RH extremes. However, the simulations show only subtle variations and lack clear V-shaped evolutions (dashed lines in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec). This discrepancy is further supported by the spectral analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed), where the model struggles to capture the observed interdecadal variability in RH extremes, despite closely matching the spectral characteristics of temperature extremes. These findings indicate a key limitation of the model in representing moisture-related processes that are critical for fire risk, even though it performs reasonably well in capturing temperature-driven trends. This discrepancy aligns with Simpson et al.\u003csup\u003e14\u003c/sup\u003e, who similarly reported substantial model-observation differences in annual humidity trends using multi-model ensembles.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eSeasonal differences in humidity extremes\u003c/h3\u003e\n\u003cp\u003eLow-RH extremes exhibit distinct seasonality (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Summertime RH extremes display a pronounced V-shaped trajectory around 1980 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb), whereas wintertime RH extremes have increased continuously over the entire record (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea), with a notably faster rate of increase in recent decades (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec), indicating an acceleration in cold-season drying. Collectively, while summertime dryness in the SWUS lessened prior to 1980, wintertime conditions continued to dry during that same period, highlighting a remarkable seasonal contrast in RH trends. This contrast diminishes after 1980 as both seasons exhibit a more synchronized increase in dryness (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec), reflecting a growing seasonal coherence.\u003c/p\u003e\u003cp\u003eSeasonal differences in low-RH extremes appear to be shaped by varying underlying mechanisms. Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea shows that wintertime RH is closely linked to temperature, with a temporal correlation of 0.61 in observations and 0.64 in model simulations. Interestingly, when the linear trend is removed from both variables, the RH-temperature correlation drops to 0.47 in observations but increases to 0.78 in the model, indicating that the model might overestimate synchronized co-variability between wintertime RH and temperature, while underestimating longer-term trends\u0026ndash;especially for RH, as reflected in the relatively flat simulated RH trend (dashed cyan line in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea).\u003c/p\u003e\u003cp\u003eIn contrast, summertime RH exhibits a distinct V-shaped trend centered around 1980 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec and Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb), with RH extremes decreasing before 1980 and increasing afterward. This fluctuation-like evolution indicates that climate change alone cannot fully explain the observed changes. Instead, the timing of these changes in summertime RH extremes aligns well with the Pacific Decadal Oscillation (PDO), the dominant mode of SST variability in the North Pacific, which exhibits strong decadal variability. Summertime RH is significantly correlated with the PDO across annual, summer, and winter timescales, with the strongest correlation found with the preceding winter's PDO (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.60 in observations and 0.48 in model simulations). This suggests that the evolution of the wintertime PDO may establish favorable preconditions for summertime RH conditions to develop.\u003c/p\u003e\u003cp\u003eSince RH is influenced not only by humidity but also by temperature, we further decompose RH into two components: the temperature-controlled saturation vapor pressure and the humidity-controlled actual vapor pressure. Consistent with earlier results, the model can reproduce the temperature-controlled component with temporal correlations reaching 0.78 for the annual time series (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). However, it struggles to reproduce the pronounced decadal variation in actual vapor pressure\u0026mdash;particularly the V-shaped pattern centered around 1980. Instead, the simulated actual vapor pressure trends remain relatively flat across all seasons before and after 1980, suggesting that the model underestimates decadal-scale humidity variability during summer.\u003c/p\u003e\u003cp\u003eIt becomes clear that the exacerbated fire risk in recent decades (post-1980) is primarily driven by the superposed effects of rising temperatures and declining RH across seasons. To further investigate the individual contributions of saturation and actual vapor pressure to overall RH evolution, we examine which linear weighting of saturation and actual vapor pressure best explains observed RH changes. Practically, we obtain the weights via multiple linear regression. Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee shows the temporal correlations between RH and various linear combinations of saturation and actual vapor pressures. The results indicate that summertime RH variability is largely controlled by the humidity-driven actual vapor pressure, accounting for approximately 70% of the overall RH variability. In contrast, wintertime RH is influenced more equally by both saturation and actual vapor pressures. The model reproduces the summer dominance of actual vapor pressure reasonably well (75%, as indicated by the vertical dashed line in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee), but overestimates its contribution during winter. This misrepresentation is effectively compensated by the model\u0026rsquo;s tendency to overestimate the synchronized co-variability between RH and temperature in winter (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea), leading to a seemingly reasonable simulation of overall wintertime RH despite inaccuracies in the underlying component contributions\u0026mdash;an example of 'getting the right answer for the wrong reasons.'\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo further pinpoint the model discrepancies, we examine the joint probability density between daily saturation and actual vapor pressures. While the model captures the overall shape of the observed distribution (Supplementary Fig.\u0026nbsp;1), it displays notable seasonal biases (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea, the wintertime biases organize into two oppositely signed dipoles with lobes elongated along the RH isolines, which is consistent with a Clausius-Clapeyron-like shift that operates at nearly constant RH, so percent errors in actual vapor pressure track those in saturation vapor pressure. Because the differences are largely confined along RH lines, positive and negative contributions within the low-RH wedge mostly cancel, helping to explain SPEAR\u0026rsquo;s relatively good performance for winter low-RH extremes. In contrast, extended summer (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb) shows a dipole structure that straddles the ~\u0026thinsp;50% RH line: a negative lobe on the lower-RH (drier) side and a positive lobe on the higher-RH (more humid) side over a similar range of saturation vapor pressure. This across-RH structure indicates that SPEAR underrepresents dry conditions and overrepresents humid conditions, yielding an overall moist bias in summer.\u003c/p\u003e\u003cp\u003eThis discrepancy is not unique to the SPEAR model. To further evaluate model performance, we analyze a 10-member ensemble of prescribed-SST simulations from the Community Earth System Model version 2 (CESM2) \u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. CESM2\u0026rsquo;s joint-density difference during the extended winter is broader than SPEAR\u0026rsquo;s and shows a clear across-RH structure (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec): deficits on low-RH lines (\u0026le;\u0026thinsp;30%) paired with excess on mid-RH lines (~\u0026thinsp;30\u0026ndash;60%) over a similar range of saturation vapor pressure. The overall underrepresentation of very dry states weakens cancellation within the low-RH wedge and increases the likelihood of underestimating winter low-RH extremes relative to SPEAR. For the extended summer (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed), CESM2 exhibits a broad negative swath across most RH values when saturation vapor pressure is less than 3 kPa, indicating an overall underrepresentation of those states. At warmer conditions, the distribution shows a clear overestimation of mid-RH states. This places more weight on \u0026ldquo;less-dry\u0026rdquo; hot days and, as a result, reduces the occurrence of very hot, dry states. These biases underscore persistent challenges in simulating joint moisture\u0026ndash;temperature extremes. In common, both models underestimate dry conditions during the extended summer, while the differences highlight that SPEAR generally exhibits smaller and more spatially localized biases, indicating closer agreement with observations, particularly in winter. Moreover, both CESM2 and SPEAR can accurately simulate saturation vapor pressure, with CESM2 showing a temporal correlation of 0.82 with observations. However, CESM2 also markedly underestimates interdecadal variability of actual vapor pressure (Supplementary Fig.\u0026nbsp;2).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\n\u003ch3\u003eLinking summer RH extremes to internal climate variability\u003c/h3\u003e\n\u003cp\u003eBoth the frequency (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) and aggregated intensity (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e) of summertime RH extremes correspond to the wintertime PDO. It is expected that aggregated RH intensity (i.e., the normalized sum of RH anomalies during extreme days) is highly correlated with its frequency (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.99 in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea). Although the model struggles to reproduce the observed V-shaped evolution, it still exhibits strong internal consistency, with correlations between low-RH frequency and its intensity generally exceeding 0.98 across ensemble members. Moreover, the aggregated intensity of summertime RH extremes is strongly correlated with the area-averaged summer RH anomalies over the SWUS (computed from daily values without restricting to extreme conditions; purple in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea), with a correlation of \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.89.\u003c/p\u003e\u003cp\u003eSpatial patterns (Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb\u0026ndash;f) reveal that the positive phase of the preceding winter PDO\u0026mdash;characterized by warmer-than-normal SST anomalies in the tropical Pacific and cooler-than-normal SST anomalies in the western and central North Pacific\u0026mdash;is associated with reduced summer RH intensity. This phase coincides with a deepened Aleutian Low, a semi-permanent low-pressure system over the Aleutian Islands, which extends toward the western US at the surface (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb). In the mid- to upper troposphere, geopotential height anomalies exhibit a Pacific\u0026ndash;North American (PNA)-like wave train pattern, featuring a weak positive node over western North America. The Aleutian Low remains evident across the pressure levels, with its core centered over the North Pacific.\u003c/p\u003e\u003cp\u003eThe vertical cross-section across western North America (Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ed-f) shows that the SWUS lies within a pronounced pressure gradient zone\u0026mdash;situated between a deep high-pressure system over higher latitudes and a shallow, weak low-pressure anomaly over the SWUS and subtropics. This configuration is associated with enhanced easterly and northerly flow. The presence of a weak baroclinic structure\u0026mdash;characterized by low surface pressure beneath an upper-level heat dome\u0026mdash;creates a dynamically favorable environment for precipitation development, potentially preconditioning against summertime RH extremes. Similar spatial patterns are observed using the seasonal mean RH anomalies (Supplementary Fig.\u0026nbsp;3).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eDespite exhibiting much weaker spatial loadings, the model ensemble mean captures the general patterns of the PDO and Aleutian Low (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eg). This weaker relationship may again be linked to the absence of interdecadal humidity variability in the model simulations. The difference maps (Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eh and i) further illustrate this underestimation of the winter PDO/PNA and summer RH relationships in the model. Although the model appears to capture the high-pressure system at higher latitudes and reproduces the associated anomalous easterly and northerly winds (Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ej\u0026ndash;l), these features are noticeably weaker compared to observations. Collectively, the summer RH biases can be seen through the teleconnection patterns. With the absence of decadal humidity variability, wintertime preconditioning signal appears much weaker in the model.\u003c/p\u003e"},{"header":"Discussion, summary, and conclusions","content":"\u003cp\u003eConsensus holds that anthropogenic forcing has exacerbated fire risk in the western US in recent decades \u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e,\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. Our results are consistent with this view: climate change\u0026ndash;driven global warming, combined with decreasing humidity, has contributed to hotter, drier, and more fire-prone conditions in the SWUS, particularly since the early 1980s. However, how fire risk may evolve across different seasons remains unclear. To address this, we analyzed daily FWI across seasons (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) and found that potential fire risk has continued to increase during the non-traditional fire season (December through May), while showing a slight reduction during the warm seasons (June through November). These findings motivate further investigation into the underlying drivers of these seasonal shifts.\u003c/p\u003e\u003cp\u003eDecomposing the FWI into its key meteorological components, we identified that temperature and RH are the main drivers of fire weather conditions across the SWUS, consistent with Jain et al.\u003csup\u003e13\u003c/sup\u003e. While temperature has exhibited an accelerated upward trend, RH, particularly in summer, follows a pronounced V-shaped trajectory centered around 1980. Specifically, the frequency of low-RH extremes declined before 1980 but began increasing thereafter, with the post-1980 upward trend nearly matching the magnitude of the earlier decline (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Meanwhile, RH extremes exhibit distinct seasonality, particularly before 1980. Summertime RH extremes show a continued decline, whereas wintertime RH displays a slight upward trend, implying the presence of different underlying mechanisms across seasons.\u003c/p\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e summarizes the contributing factors underlying the distinct seasonal behavior of fire risk. The consistent upward trends in wintertime RH and temperature suggest that anthropogenic global warming is likely exacerbating fire risk outside the traditional fire season by increasing temperatures and enhancing atmospheric dryness. In contrast, internal climate variability\u0026mdash;particularly the PDO\u0026mdash;plays a dominant role in modulating summertime RH, resulting in pronounced decadal fluctuations in summertime fire risk.\u003c/p\u003e\u003cp\u003eESMs continue to exhibit substantial limitations in reproducing humidity-related variables\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. Specifically, Simpson et al.\u003csup\u003e14\u003c/sup\u003e reported that most climate models fail to capture the observed trends in humidity over the past 40 years from 1980\u0026ndash;2020. Even when driven by observed SSTs, these models do not fully resolve the discrepancy in humidity trends. Lou et al.\u003csup\u003e25\u003c/sup\u003e further argued that although climate models can reasonably reproduce the spatial patterns of VPD in the SWUS on seasonal time scales, they struggle to represent the decadal component of the leading VPD mode of variability. This deficiency primarily stems from biases in simulating water vapor, rather than temperature.\u003c/p\u003e\u003cp\u003eDue to Clausius\u0026ndash;Clapeyron scaling, the atmosphere's capacity to hold moisture is substantially lower in winter than in summer. During winter, both temperature and humidity contribute comparably to fire risk (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e and \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Although the model exhibits limited skill in capturing humidity variability, its relatively strong performance in simulating temperature partially offsets this deficiency, thereby improving overall model fidelity. As a result, wintertime relative humidity (RH) extremes are reasonably well represented.\u003c/p\u003e\u003cp\u003eIn contrast, during summer, actual vapor pressure\u0026mdash;which is largely governed by humidity\u0026mdash;emerges as the dominant driver of RH and fire weather conditions (Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e and \u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Because the model performs poorly in simulating moisture-related variables, this shortcoming becomes more critical and directly constrains model fidelity. In this case, the limiting factor\u0026mdash;moisture\u0026mdash;is also the most essential, and its deficiency cannot be compensated by the relatively well-simulated temperature field. This leads to a clear misrepresentation of summertime low-RH extremes, manifested in the absence of observed decadal humidity variability (i.e., the V-shaped trajectory in Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e and \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) and overly weak teleconnections with the wintertime PDO/PNA patterns (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eComparing observations and model simulations (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e and Supplementary Fig.\u0026nbsp;4), we identify at least three key limitations that warrant further investigation. First, the model underestimates the relationship between summer humidity extremes and winter PDO variability (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.62 in observations vs. \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.38 in the model). Second, the model\u0026rsquo;s relative frequency of low-RH extremes is over-synchronized across seasons: the winter\u0026ndash;summer relationship is much stronger than observed, with a correlation of \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.62 in the model versus \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.29 in observations (for summer leading). The full set of temporal correlations is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e and Supplementary Fig.\u0026nbsp;4. Third, the model exhibits an overly dominant role of humidity during winter, although the impact on RH is partially offset by the model\u0026rsquo;s strong performance in temperature-related quantities. Moreover, these model discrepancies are not unique to the model examined in this study. Similar biases\u0026mdash;such as the underestimation of decadal humidity variability and the misrepresentation of long-term vapor pressure trend\u0026mdash;have been documented across other ESMs as well \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e (also see Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e and Supplementary Fig.\u0026nbsp;2). This highlights a broader challenge in accurately simulating moisture-related processes and their associated extremes, particularly in regions sensitive to both internal variability and long-term climate change. Last, it is worth noting that this study does not explicitly examine whether the PDO itself is internally generated \u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e or externally forced \u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. Resolving this question lies beyond our scope and remains an open direction for future work.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eObservations and model simulations\u003c/h2\u003e\u003cp\u003eWe use daily data from the ECMWF Reanalysis v5 (ERA5 \u003csup\u003e27\u003c/sup\u003e), including the Canadian Fire Weather Index (FWI \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e; see Supplementary Table\u0026nbsp;1 for details), 2-meter dew point temperature, 2-meter air temperature, and 10-meter wind speed. In addition, we analyze large-scale oceanic and atmospheric conditions using monthly ERA5 data, including sea surface temperature (SST), sea level pressure (SLP), geopotential heights, and zonal and meridional wind components at various pressure levels. All daily variables are remapped to a regular 0.5\u0026deg; \u0026times; 0.5\u0026deg; grid. To focus on fire weather extremes over land in the US, we apply a land mask to exclude fire weather conditions (e.g., temperature, humidity, and wind speed) over the ocean. This mask is derived from the ERA5 SST field, which is also remapped to a 0.5\u0026deg; \u0026times; 0.5\u0026deg; grid. Leap days (i.e., Feb 29) are excluded from the daily data. Monthly variables are remapped to a regular 2\u0026deg; \u0026times; 2\u0026deg; grid.\u003c/p\u003e\u003cp\u003eThe model simulations are taken from the Seamless System for Prediction and Earth System Research (SPEAR; Delworth et al. \u003csup\u003e29\u003c/sup\u003e), a general circulation model designed by NOAA\u0026rsquo;s Geophysical Fluid Dynamics Laboratory (GFDL) to support both seasonal prediction and Earth system research. We use prescribed-SST simulations, in which observed SSTs are used as surface forcing. The prescribed-SST simulations have 30 ensemble members, with horizontal grid spacing of ~\u0026thinsp;50 km for the atmosphere and land, and ~\u0026thinsp;100 km for the ocean and sea ice. Daily outputs of 2-meter temperature, RH, and wind speed are used, with the same preprocessing methods applied as for the reanalysis data. Monthly outputs of SST, SLP, geopotential height, and zonal and meridional wind components at multiple pressure levels are obtained from the SPEAR prescribed-SST simulations, and processed similarly to the observational dataset. For comparison, we also analyze a 10-member ensemble of prescribed-SST simulations from the Community Earth System Model Version 2 (CESM2 \u003csup\u003e35\u003c/sup\u003e). Daily RH and temperature outputs from CESM2 are used and preprocessed using the same methodology as for the other datasets.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eEstimation of vapor pressure quantities\u003c/h3\u003e\n\u003cp\u003eDaily 2-meter dew point \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{d}\\)\u003c/span\u003e\u003c/span\u003e and air temperature \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:T\\)\u003c/span\u003e\u003c/span\u003e from ERA5 reanalysis are used to calculate saturation vapor pressure \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{s}}\\left(T\\right)\\)\u003c/span\u003e\u003c/span\u003e and actual vapor pressure \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{a}}\\left({T}_{d}\\right)\\)\u003c/span\u003e\u003c/span\u003e. These are then used to derive RH, defined as the ratio \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:RH={\\varvec{e}}_{\\varvec{a}}/{\\varvec{e}}_{\\varvec{s}}\\)\u003c/span\u003e\u003c/span\u003e, and VPD, defined as the difference \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{VPD=\\varvec{e}}_{\\varvec{s}}-{\\varvec{e}}_{\\varvec{a}}\\)\u003c/span\u003e\u003c/span\u003e. Here, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{s}}\\)\u003c/span\u003e\u003c/span\u003e and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{a}}\\)\u003c/span\u003e\u003c/span\u003e are computed following ref \u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. Daily outputs of \u003cem\u003eRH\u003c/em\u003e and \u003cem\u003eT\u003c/em\u003e from the models are used. Then, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{a}}\\)\u003c/span\u003e\u003c/span\u003e is calculated as \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{e}_{s}\\left(T\\right)*RH/100\\)\u003c/span\u003e\u003c/span\u003e. Here, Clausius\u0026ndash;Clapeyron (CC) scaling describes how the saturation vapor pressure \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{e}_{s}\\left(T\\right)\\)\u003c/span\u003e\u003c/span\u003e increases exponentially with temperature.\u003c/p\u003e\u003cp\u003eIn our previous study\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e, we compared different methods for calculating \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{s}}\\left(T\\right)\\)\u003c/span\u003e\u003c/span\u003e: one is based on the average of \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{s}}\\left({T}_{max}\\right)\\:and\\:{\\varvec{e}}_{\\varvec{s}}\\left({T}_{min}\\right)\\:\\)\u003c/span\u003e\u003c/span\u003e, where \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{max}\\)\u003c/span\u003e\u003c/span\u003e and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{min}\\)\u003c/span\u003e\u003c/span\u003e are the daily maximum and minimum temperatures, and another based on \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{s}}\\left({T}_{mean}\\right)\\)\u003c/span\u003e\u003c/span\u003e, where \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{mean}\\)\u003c/span\u003e\u003c/span\u003e is the daily mean temperature. From a temporal correlation perspective, both methods yielded comparable results, with no notable differences reported (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.99 for monthly \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{s}}\\left(T\\right)\\)\u003c/span\u003e\u003c/span\u003e time series in the SWUS). Here, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{s}}\\left(T\\right)\\)\u003c/span\u003e\u003c/span\u003e is computed using \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\varvec{e}}_{\\varvec{s}}\\left({T}_{mean}\\right)\\)\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e\n\u003ch3\u003eThe relative PDO index\u003c/h3\u003e\n\u003cp\u003eThe relative PDO index is calculated as the area-averaged monthly relative SST anomalies over the North Pacific region (25\u0026deg;N\u0026ndash;45\u0026deg;N, 140\u0026deg;E\u0026ndash;145\u0026deg;W; Region 1 in Henley et al. \u003csup\u003e41\u003c/sup\u003e). The term \u0026lsquo;relative\u0026rsquo; refers to subtracting the near-global (60\u0026deg;S\u0026ndash;60\u0026deg;N) mean SST anomalies from the traditional PDO time series. As noted by L\u0026rsquo;Heureux et al.\u003csup\u003e42\u003c/sup\u003e, relative Ni\u0026ntilde;o3.4 index is less sensitive to the choice of baseline period used to define climatology, and is more closely tied to anomalous tropical convection. Similarly, Tan et al. \u003csup\u003e43\u003c/sup\u003e adopted the relative SST framework to define a relative Atlantic Ni\u0026ntilde;o3.4 index. Supplementary Fig.\u0026nbsp;5 illustrates that both the near-global mean SST anomalies and the traditional PDO index are sensitive to various baseline choices \u0026mdash; whether using a 30-year sliding climatology\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e,\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e, or fixed climatologies based on the first 30 years, last 30 years, or the full record. These fixed baseline approaches become especially problematic in a changing climate, where the background mean state itself is varying \u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. In contrast, the relative PDO index \u0026mdash; by removing the near-global mean SST anomalies \u0026mdash; shows much lower sensitivity to baseline selection. For this reason, we adopt the relative PDO index in this study.\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eExtreme thresholds\u003c/h2\u003e\u003cp\u003eWe employ two complementary approaches to define fire weather extremes: a fixed (or absolute) threshold and a varying threshold. The fixed threshold approach identifies values that exceed the 90th percentile (such as FWI, temperature, wind speed) or fall below the 10th percentile (such as RH, actual vapor pressure) of the entire dataset, calculated across all time steps and all geographical locations\u0026mdash;specifically over land areas within our study domain (24\u0026deg;N\u0026ndash;50\u0026deg;N, 126\u0026deg;W\u0026ndash;64\u0026deg;W). This threshold is universally applicable and not dependent on specific seasons or locations. It allows us to understand the climatology and seasonal cycle of fire weather conditions, as well as to identify fire-prone regions.\u003c/p\u003e\u003cp\u003eHowever, fixed thresholds do not account for regional climatological differences. Fire vulnerability and management capacity vary widely\u0026mdash;what constitutes a moderate fire risk in fire-prone regions may be considered extreme in typically humid areas. To address this, we also analyze fire weather extremes using a varying threshold scheme, defined relative to season- and location-specific 90th or 10th percentile thresholds. Specifically, we adopt the threshold calculation from Hobday et al.\u003csup\u003e30\u003c/sup\u003e, in which thresholds are determined based on the 90th or 10th percentiles of daily values within an 11-day centered window, aggregated across all years. This varying threshold approach allows us to identify extremes that are unusual or disruptive relative to regional norms, regardless of their absolute magnitude. Although we refer to it as a varying threshold approach, it incorporates a fixed component: for example, fire weather extremes are initially identified using season- and location-specific 90th percentile FWI thresholds, but any values falling below the fixed lower tercile (33.3%) of FWI across the full dataset are excluded. Similarly, low-RH extremes are first defined using varying 10th percentile thresholds, but any values exceeding the fixed upper tercile (66.6%) of RH across the entire record are excluded. This hybrid approach avoids false positives in low-variability or low-risk regions. Although we describe our season- and location-specific threshold as a \u0026ldquo;varying threshold,\u0026rdquo; it should not be confused with the \u0026ldquo;shifting threshold\u0026rdquo; of Amaya et al. \u003csup\u003e46\u003c/sup\u003e, which explicitly removes the long-term background state (which is ocean warming in that case) before defining extremes. Our approach varies the threshold across season and space but does not subtract a long-term background, as the reference period is based on the full record.\u003c/p\u003e\u003cp\u003eEffective fire management strategies often integrate both fixed- and varying- threshold fire risk assessments in the synergistic planning. This integration allows for robust baseline preparedness while maintaining the flexibility to respond to unexpected changes in risk levels.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003eClimatology\u003c/h2\u003e\u003cp\u003eWe follow Hobday et al.\u003csup\u003e30\u003c/sup\u003e to compute the daily climatology. First, we apply an 11-day centered moving average to smooth the daily values for each calendar day. Then, the smoothed values are averaged across the baseline years (1940\u0026ndash;2022 in this study) to produce the climatological mean for each day of the year. For monthly outputs, we apply a 30-year moving baseline to account for changes in the background mean state. Differences resulting from the choice of baseline periods are discussed in the previous section (see Supplementary Fig.\u0026nbsp;5).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eCharacteristics of fire weather extremes\u003c/h2\u003e\u003cp\u003eRelative frequency is defined as the ratio of extreme day counts to the total number of days within the respective period. Aggregated intensity is calculated as the sum of daily anomalies during extreme conditions. Unlike averaged intensity, aggregated intensity accounts for the spatial extent and cumulative impact of fire weather extremes. Duration is defined as the number of consecutive days meeting the extreme threshold. Following Ref. \u003csup\u003e30\u003c/sup\u003e, we also account for short breaks\u0026mdash;referred to as 'gap days'\u0026mdash;between extreme events. Specifically, if two or fewer non-extreme days separate periods of fire weather extremes with durations\u0026thinsp;\u0026ge;\u0026thinsp;5 days, these gap days are included in the event duration. For example, a sequence of five fire weather days, followed by two non-extreme days, and then another seven fire weather days is considered a single 14-day event. Since including gap days does not notably alert the extreme characteristics, we exclude them for most of our analysis. Extreme events are defined as periods of extreme conditions lasting five days or more.\u003c/p\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eCOMPETING INTERESTS\u003c/h2\u003e\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\u003c/p\u003e\u003ch2\u003eAuthor contributions\u003c/h2\u003e\u003cp\u003eJ.L. performed the analyses and led the writing, with input from all authors. Y.J. and T.L.D. helped shape the scientific ideas and contributed to discussions and writing. A.T.W. refined the presentation of the results and interpretations. D.S. cross-validated some results and identified and helped correct a coding error that was important for interpreting the findings.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eThe authors thank Drs. Liwei Jia and John Krasting for their valuable comments and discussions that helped improve this manuscript. We also acknowledge William Cooke for generating the SPEAR model outputs. J.L. thanks Nathaniel Johnson for the discussion of the relative SST indices. This work was conducted by J.L. under award NA22OAR4050664d from the National Oceanic and Atmospheric Administration (NOAA), U.S. Department of Commerce. The statements, findings, conclusions, and recommendations are those of the author(s) and do not necessarily reflect the views of NOAA or the U.S. Department of Commerce. J.L. also acknowledges the use of NOAA\u0026rsquo;s Research and Development High-Performance Computing System (RDHPCS) and the NSF NCAR high-performance computing and storage resources, managed by the Computational and Information Systems Laboratory (CISL; doi:\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5065/qx9a-pg09\u003c/span\u003e\u003cspan address=\"10.5065/qx9a-pg09\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003cdiv id=\"Sec14\" class=\"Section2\"\u003e\u003ch2\u003eData availability\u003c/h2\u003e\u003cp\u003eThis study utilizes multiple observational and model-based datasets relevant to fire weather conditions and wildfire activity. Daily Canadian Forest Service Fire Weather Index (FWI) values are obtained from ERA5 through the Copernicus Emergency Management Service Fire Danger System historical product, available at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cds.climate.copernicus.eu/cdsapp#!/dataset/cems-fire-historical?tab=overview\u003c/span\u003e\u003cspan address=\"https://cds.climate.copernicus.eu/cdsapp#!/dataset/cems-fire-historical?tab=overview\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. National wildfire statistics, including total wildland fires and acres burned in the U.S. from 1983 to 2022, are compiled by the National Interagency Fire Center and available at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.nifc.gov/fire-information/statistics/wildfires\u003c/span\u003e\u003cspan address=\"https://www.nifc.gov/fire-information/statistics/wildfires\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Spatial data for burned area in California are accessed through the California State Geoportal: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://gis.data.ca.gov/\u003c/span\u003e\u003cspan address=\"https://gis.data.ca.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. Additionally, CESM2 prescribed-SST simulations used in this study are publicly available through NCAR's CESM Climate Simulation Working Group at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.cesm.ucar.edu/working-groups/climate/simulations/cam6-prescribed-sst\u003c/span\u003e\u003cspan address=\"https://www.cesm.ucar.edu/working-groups/climate/simulations/cam6-prescribed-sst\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec15\" class=\"Section2\"\u003e\u003ch2\u003eCode availability\u003c/h2\u003e\u003cp\u003eAll R scripts used for analysis and figure generation will be made publicly available through a data repository upon acceptance of the manuscript.\u003c/p\u003e\u003c/div\u003e"},{"header":"References","content":"\u003col\u003e\n \u003col\u003e\n\u003cli\u003eAbatzoglou, J. T. \u0026amp; Williams, A. P. Impact of anthropogenic climate change on wildfire across western US forests. \u003cem\u003eProc. Natl. Acad. Sci.\u003c/em\u003e \u003cstrong\u003e113\u003c/strong\u003e, 11770\u0026ndash;11775 (2016).\u003c/li\u003e\n\u003cli\u003eBrown, P. T. \u003cem\u003eet al.\u003c/em\u003e Climate warming increases extreme daily wildfire growth risk in California. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e621\u003c/strong\u003e, 760\u0026ndash;766 (2023).\u003c/li\u003e\n\u003cli\u003eBurke, M. \u003cem\u003eet al.\u003c/em\u003e The changing risk and burden of wildfire in the United States. \u003cem\u003eProc. Natl. Acad. Sci.\u003c/em\u003e \u003cstrong\u003e118\u003c/strong\u003e, e2011048118 (2021).\u003c/li\u003e\n\u003cli\u003eDennison, P. E., Brewer, S. C., Arnold, J. D. \u0026amp; Moritz, M. A. Large wildfire trends in the western United States, 1984\u0026ndash;2011. \u003cem\u003eGeophys. Res. Lett.\u003c/em\u003e \u003cstrong\u003e41\u003c/strong\u003e, 2928\u0026ndash;2933 (2014).\u003c/li\u003e\n\u003cli\u003eLittell, J. S., McKenzie, D., Peterson, D. L. \u0026amp; Westerling, A. L. Climate and wildfire area burned in western U.S. ecoprovinces, 1916\u0026ndash;2003. \u003cem\u003eEcol. Appl.\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 1003\u0026ndash;1021 (2009).\u003c/li\u003e\n\u003cli\u003eWilliams, A. P. \u003cem\u003eet al.\u003c/em\u003e Observed Impacts of Anthropogenic Climate Change on Wildfire in California. \u003cem\u003eEarths Future\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 892\u0026ndash;910 (2019).\u003c/li\u003e\n\u003cli\u003eZou, Y., Rasch, P. J., Wang, H., Xie, Z. \u0026amp; Zhang, R. Increasing large wildfires over the western United States linked to diminishing sea ice in the Arctic. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 6048 (2021).\u003c/li\u003e\n\u003cli\u003eBarbero, R., Abatzoglou, J. T., Larkin, S., Kolden, C. A. \u0026amp; Stocks, B. Climate change presents increased potential for very large fires in the contiguous United States. \u003cem\u003eInt. J. Wildland Fire\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 892\u0026ndash;899 (2015).\u003c/li\u003e\n\u003cli\u003eMoritz, M. A. \u003cem\u003eet al.\u003c/em\u003e Climate change and disruptions to global fire activity. \u003cem\u003eEcosphere\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, art49 (2012).\u003c/li\u003e\n\u003cli\u003eJolly, W. M. \u003cem\u003eet al.\u003c/em\u003e Climate-induced variations in global wildfire danger from 1979 to 2013. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 7537 (2015).\u003c/li\u003e\n\u003cli\u003eJain, P., Wang, X. \u0026amp; Flannigan, M. D. Trend analysis of fire season length and extreme fire weather in North America between 1979 and 2015. \u003cem\u003eInt. J. Wildland Fire\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 1009\u0026ndash;1020 (2017).\u003c/li\u003e\n\u003cli\u003eWeather and Climate Extreme Events in a Changing Climate. in \u003cem\u003eClimate Change 2021 \u0026ndash; The Physical Science Basis: Working Group I Contribution to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change\u003c/em\u003e (ed. Intergovernmental Panel on Climate Change (IPCC)) 1513\u0026ndash;1766 (Cambridge University Press, Cambridge, 2023). doi:10.1017/9781009157896.013.\u003c/li\u003e\n\u003cli\u003eJain, P., Castellanos-Acuna, D., Coogan, S. C. P., Abatzoglou, J. T. \u0026amp; Flannigan, M. D. Observed increases in extreme fire weather driven by atmospheric humidity and temperature. \u003cem\u003eNat. Clim. Change\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 63\u0026ndash;70 (2022).\u003c/li\u003e\n\u003cli\u003eSimpson, I. R. \u003cem\u003eet al.\u003c/em\u003e Observed humidity trends in dry regions contradict climate models. \u003cem\u003eProc. Natl. Acad. Sci.\u003c/em\u003e \u003cstrong\u003e121\u003c/strong\u003e, e2302480120 (2024).\u003c/li\u003e\n\u003cli\u003eP\u0026eacute;rez-Invern\u0026oacute;n, F. J., Gordillo-V\u0026aacute;zquez, F. J., Huntrieser, H. \u0026amp; J\u0026ouml;ckel, P. Variation of lightning-ignited wildfire patterns under climate change. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 739 (2023).\u003c/li\u003e\n\u003cli\u003eOliveras Menor, I. \u003cem\u003eet al.\u003c/em\u003e Integrated fire management as an adaptation and mitigation strategy to altered fire regimes. \u003cem\u003eCommun. Earth Environ.\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 202 (2025).\u003c/li\u003e\n\u003cli\u003eMouillot, F. \u0026amp; Field, C. B. Fire history and the global carbon budget: a 1\u0026deg;\u0026times; 1\u0026deg; fire history reconstruction for the 20th century. \u003cem\u003eGlob. Change Biol.\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 398\u0026ndash;420 (2005).\u003c/li\u003e\n\u003cli\u003eWagner, V. \u0026amp; E, C. \u003cem\u003eDevelopment and Structure of the Canadian Forest Fire Weather Index System\u003c/em\u003e. (1987).\u003c/li\u003e\n\u003cli\u003eFire behaviour in eucalypt forests / [by] A. G. McArthur - Trove. https://trove.nla.gov.au/work/21914760.\u003c/li\u003e\n\u003cli\u003eQuilcaille, Y., Batibeniz, F., Ribeiro, A. F. S., Padr\u0026oacute;n, R. S. \u0026amp; Seneviratne, S. I. Fire weather index data under historical and shared socioeconomic pathway projections in the 6th phase of the Coupled Model Intercomparison Project from 1850 to 2100. \u003cem\u003eEarth Syst. Sci. Data\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 2153\u0026ndash;2177 (2023).\u003c/li\u003e\n\u003cli\u003eWilliams, A. P. \u003cem\u003eet al.\u003c/em\u003e Growing impact of wildfire on western US water supply. \u003cem\u003eProc. Natl. Acad. Sci. U. S. A.\u003c/em\u003e \u003cstrong\u003e119\u003c/strong\u003e, e2114069119 (2022).\u003c/li\u003e\n\u003cli\u003eWilliams, A. P. \u003cem\u003eet al.\u003c/em\u003e Causes and Implications of Extreme Atmospheric Moisture Demand during the Record-Breaking 2011 Wildfire Season in the Southwestern United States. \u003cem\u003eJ. Appl. Meteorol. Climatol.\u003c/em\u003e \u003cstrong\u003e53\u003c/strong\u003e, 2671\u0026ndash;2684 (2014).\u003c/li\u003e\n\u003cli\u003eSimpson, I. R. \u003cem\u003eet al.\u003c/em\u003e Confronting Earth System Model trends with observations. \u003cem\u003eSci. Adv.\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, eadt8035 (2025).\u003c/li\u003e\n\u003cli\u003eShaw, T. A. \u003cem\u003eet al.\u003c/em\u003e Regional climate change: consensus, discrepancies, and ways forward. \u003cem\u003eFront. Clim.\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, (2024).\u003c/li\u003e\n\u003cli\u003eLou, J., Joh, Y. \u0026amp; Delworth, T. L. Identifying common precursors of decadal vapor pressure deficit variability in the Southwestern United States. (2025) doi:10.1175/JCLI-D-24-0296.1.\u003c/li\u003e\n\u003cli\u003evan Oldenborgh, G. J. \u003cem\u003eet al.\u003c/em\u003e Pathways and pitfalls in extreme event attribution. \u003cem\u003eClim. Change\u003c/em\u003e \u003cstrong\u003e166\u003c/strong\u003e, 1\u0026ndash;27 (2021).\u003c/li\u003e\n\u003cli\u003eHersbach, H. \u003cem\u003eet al.\u003c/em\u003e The ERA5 global reanalysis. \u003cem\u003eQ. J. R. Meteorol. Soc.\u003c/em\u003e \u003cstrong\u003e146\u003c/strong\u003e, 1999\u0026ndash;2049 (2020).\u003c/li\u003e\n\u003cli\u003eVitolo, C. \u003cem\u003eet al.\u003c/em\u003e ERA5-based global meteorological wildfire danger maps. \u003cem\u003eSci. Data\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 216 (2020).\u003c/li\u003e\n\u003cli\u003eDelworth, T. L. \u003cem\u003eet al.\u003c/em\u003e SPEAR: The Next Generation GFDL Modeling System for Seasonal to Multidecadal Prediction and Projection. \u003cem\u003eJ. Adv. Model. Earth Syst.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, e2019MS001895 (2020).\u003c/li\u003e\n\u003cli\u003eHobday, A. J. \u003cem\u003eet al.\u003c/em\u003e A hierarchical approach to defining marine heatwaves. \u003cem\u003eProg. Oceanogr.\u003c/em\u003e \u003cstrong\u003e141\u003c/strong\u003e, 227\u0026ndash;238 (2016).\u003c/li\u003e\n\u003cli\u003eDeng, K. \u003cem\u003eet al.\u003c/em\u003e Terrestrial Stilling Projected to Continue in the Northern Hemisphere Mid-Latitudes. \u003cem\u003eEarths Future\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, e2021EF002448 (2022).\u003c/li\u003e\n\u003cli\u003eKlink, K. Trends in mean monthly maximum and minimum surface wind speeds in the coterminous United States, 1961 to 1990. \u003cem\u003eClim. Res.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 193\u0026ndash;205 (1999).\u003c/li\u003e\n\u003cli\u003eMcVicar, T. R. \u003cem\u003eet al.\u003c/em\u003e Global review and synthesis of trends in observed terrestrial near-surface wind speeds: Implications for evaporation. \u003cem\u003eJ. Hydrol.\u003c/em\u003e \u003cstrong\u003e416\u0026ndash;417\u003c/strong\u003e, 182\u0026ndash;205 (2012).\u003c/li\u003e\n\u003cli\u003eVautard, R., Cattiaux, J., Yiou, P., Th\u0026eacute;paut, J.-N. \u0026amp; Ciais, P. Northern Hemisphere atmospheric stilling partly attributed to an increase in surface roughness. \u003cem\u003eNat. Geosci.\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 756\u0026ndash;761 (2010).\u003c/li\u003e\n\u003cli\u003eDanabasoglu, G. \u003cem\u003eet al.\u003c/em\u003e The Community Earth System Model Version 2 (CESM2). \u003cem\u003eJ. Adv. Model. Earth Syst.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, e2019MS001916 (2020).\u003c/li\u003e\n\u003cli\u003eKuo, Y.-N. \u003cem\u003eet al.\u003c/em\u003e Recent southwestern US drought exacerbated by anthropogenic aerosols and tropical ocean warming. \u003cem\u003eNat. Geosci.\u003c/em\u003e \u003cstrong\u003e18\u003c/strong\u003e, 578\u0026ndash;585 (2025).\u003c/li\u003e\n\u003cli\u003eNewman, M. \u003cem\u003eet al.\u003c/em\u003e The Pacific Decadal Oscillation, Revisited. \u003cem\u003eJ. Clim.\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e, 4399\u0026ndash;4427 (2016).\u003c/li\u003e\n\u003cli\u003eKlavans, J. M. \u003cem\u003eet al.\u003c/em\u003e Human emissions drive recent trends in North Pacific climate variations. \u003cem\u003eNature\u003c/em\u003e 1\u0026ndash;9 (2025) doi:10.1038/s41586-025-09368-2.\u003c/li\u003e\n\u003cli\u003eDevelopment and structure of the Canadian Forest Fire Weather Index System | Fire Research and Management Exchange System. https://www.frames.gov/catalog/6293.\u003c/li\u003e\n\u003cli\u003eBolton, D. The Computation of Equivalent Potential Temperature. (1980).\u003c/li\u003e\n\u003cli\u003eHenley, B. J. \u003cem\u003eet al.\u003c/em\u003e A Tripole Index for the Interdecadal Pacific Oscillation. \u003cem\u003eClim. Dyn.\u003c/em\u003e \u003cstrong\u003e45\u003c/strong\u003e, 3077\u0026ndash;3090 (2015).\u003c/li\u003e\n\u003cli\u003eL\u0026rsquo;Heureux, M. L. \u003cem\u003eet al.\u003c/em\u003e A Relative Sea Surface Temperature Index for Classifying ENSO Events in a Changing Climate. \u003cem\u003eJ. Clim.\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 1197\u0026ndash;1211 (2024).\u003c/li\u003e\n\u003cli\u003eTan, W., Liu, Y., Li, X., Johnson, N. C. \u0026amp; Hu, Z.-Z. Multi-Time Scale Variations in Atlantic Ni\u0026ntilde;o and a Relative Atlantic Ni\u0026ntilde;o Index. \u003cem\u003eGeophys. Res. Lett.\u003c/em\u003e \u003cstrong\u003e50\u003c/strong\u003e, e2023GL106511 (2023).\u003c/li\u003e\n\u003cli\u003eLou, J., Newman, M. \u0026amp; Hoell, A. Multi-decadal variation of ENSO forecast skill since the late 1800s. \u003cem\u003eNpj Clim. Atmospheric Sci.\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 1\u0026ndash;14 (2023).\u003c/li\u003e\n\u003cli\u003eLou, J., Joh, Y., Delworth, T. L. \u0026amp; Jia, L. Identifying source of predictability for vapor pressure deficit variability in the southwestern United States. \u003cem\u003eNpj Clim. Atmospheric Sci.\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 1\u0026ndash;14 (2025).\u003c/li\u003e\n\u003cli\u003eAmaya, D. J. \u003cem\u003eet al.\u003c/em\u003e Marine heatwaves need clear definitions so coastal communities can adapt. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e616\u003c/strong\u003e, 29\u0026ndash;32 (2023).\u003c/li\u003e\n\u003cli\u003e\u003c/li\u003e\n \u003c/ol\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7621598/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7621598/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eWildfire activity has increased across the western United States (US), with evidence linking these trends to anthropogenic climate forcing. Here, we analyze the frequency, intensity, and drivers of fire weather extremes in the southwestern US over the past 83 years. These extremes exhibit strong seasonality: wintertime changes align with long-term warming trends, while summertime variations are dominated by decadal fluctuations modulated by the Pacific Decadal Oscillation (PDO). Fire risk has increased steadily during December\u0026ndash;May, outside the traditional fire season, primarily due to rising temperatures and declining relative humidity (RH). In contrast, summertime fire risk\u0026mdash;largely governed by humidity\u0026mdash;follows a roughly V-shaped trajectory, decreasing from 1940\u0026ndash;1980 and then rising from 1981\u0026ndash;2022, in close alignment with decadal shifts in the PDO. Although Earth system models (ESMs) reliably capture temperature-related changes, they underestimate interdecadal variability in humidity, manifested by a weak representation of the observed PDO\u0026ndash;RH relationship and low-RH extremes in summer.\u003c/p\u003e","manuscriptTitle":"Global warming and the Pacific Decadal Oscillation drive seasonally varying increases in extreme fire weather over the southwestern US","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-10-13 06:33:38","doi":"10.21203/rs.3.rs-7621598/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"communications-earth-and-environment","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"commsenv","sideBox":"Learn more about [Communications Earth and Environment](https://www.nature.com/commsenv/)","snPcode":"","submissionUrl":"","title":"Communications Earth \u0026 Environment","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Communications Series","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"0c9791fe-fce8-4d82-8560-5d3a88c6cd8a","owner":[],"postedDate":"October 13th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":54930922,"name":"Earth and environmental sciences/Natural hazards"},{"id":54930923,"name":"Earth and environmental sciences/Climate sciences/Climate change"}],"tags":[],"updatedAt":"2026-03-28T09:11:28+00:00","versionOfRecord":[],"versionCreatedAt":"2025-10-13 06:33:38","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7621598","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7621598","identity":"rs-7621598","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}