Dual pressures of climate change and wildfire hazard on wine-growing potential in California

Abstract

This study presents an integrated analysis to assess climate change impacts on wine grape cultivation in California by jointly considering spatial suitability, wine quality, and wildfire weather. Species distribution modeling based on MaxEnt algorithm (AUC > 0.93) indicates substantial declines in climatic viticulture suitability across major regions such as Napa and Sonoma, alongside notable increases in northern and coastal areas under high emission scenario in future climate. While many regions show concurrent increases in suitability and extreme fire-weather days, areas such as Mendocino and Monterey exhibit increasing suitability coupled with decreasing extreme fire-weather days, identifying them as comparatively favorable expansion zones. Wine quality projections using a Random Forest classification model (F1 score > 0.94) show that responses to fire-weather conditions differ by variety. In Sonoma, Chardonnay tends to achieve higher quality in years without extreme fire weather but declines in high fire-danger years, whereas Pinot Noir shows the opposite tendency. These contrasting responses suggest that varietal selection will be critical for adaptation under increasing wildfire danger. Overall, the results highlight the importance of integrating disturbance regimes into climate impact assessments for sustainable viticulture.

1 Introduction

Climate change poses a fundamental challenge to the global wine sector (Jones et al., 2005). Wine production is exceptionally climate-sensitive because grape yield and quality are tightly constrained by environmental conditions, and because wine value is inseparable from place. This sensitivity is commonly framed through the concept of terroir, which reflects the combined influence of climate, soil, topography, and human practices on wine character (Van Leeuwen and Seguin, 2006). Consequently, climatic shifts can propagate through grape development and winemaking, altering wine style, typicity and economic value (Van Leeuwen et al., 2024).

A substantial body of research has examined how climate change alters the geographical suitability of wine grape cultivation. Global and regional studies consistently project a decline in suitability across many traditional wine-producing regions during the twenty-first century, alongside expansions

toward higher latitudes and elevations (Jones et al., 2010; Hannah et al., 2013; Fraga et al., 2016; Roehrdanz and Hannah, 2016; Tóth and Végvári, 2016; Irimia et al., 2018; Cardell et al., 2019; del Río et al., 2021; Hewer and Gough, 2021; Bai et al., 2022; Gaitán and Pino-Otín, 2023). These shifts are largely driven by rising temperatures that modify growing season heat accumulation and increase exposure to heat stress. At the same time, regional analyses reveal pronounced local heterogeneity, shaped by topography, precipitation regimes and interacting climatic variables (Hiraga and Matsumoto, 2025). Together, these findings indicate that climate change will restructure viticultural suitability in spatially complex and region-specific ways.

Climate change also directly affects grape and wine quality (De Orduna, 2010; Van Leeuwen and Darriet, 2016; Drappier et al., 2019; Wood et al., 2024). Warmer conditions advance phenological development, shifting ripening into hotter periods of the growing season. This alters berry metabolism, often increasing sugar concentrations while reducing organic acid levels, with downstream effects on alcohol content, acidity and aroma (Keller, 2023; Martínez-Lüscher et al., 2025). Increased water stress further modifies yield and fruit composition, while extreme heat can constrain vine metabolism, limiting the accumulation of phenolic and aromatic compounds (Hewitt et al., 2023; Ramos et al., 2024). These changes extend into winemaking, where elevated sugar levels and higher pH can disrupt fermentation, increase microbial risks and compromise wine stability (Mangas et al., 2023).

Despite substantial research progress shown above, analyses of climate change impacts on wine grape cultivation have largely treated climatic change in isolation, with only limited integration of climate-driven natural hazards. Climate change is, however, altering not only mean climatic conditions but also the frequency and intensity of extreme events. Among these hazards, wildfires have emerged as a particularly consequential threat to wine-producing regions, especially those with Mediterranean-type climates (Fried et al., 2004; Ruffault et al., 2020). In recent decades, major viticultural countries such as southern Europe, California, southern Australia, and Chile have experienced increasing wildfire activity (Turco et al., 2014; Williams et al., 2019; Dupuy et al., 2020; Ciocca et al., 2023). Wildfires affect vineyards through multiple pathways, including direct vine destruction, heat-induced mortality, soil degradation and smoke exposure (Madhusoodanan, 2021; Zakowski et al., 2023; Greene et al., 2025). In Napa County, California, a single wildfire burned more than 6,800 acres of land, including extensive vineyard areas, damaging approximately 1,500 acres of agricultural land and causing 10 of millions of dollars in losses, largely attributable to wine grapes (CBS News, 2025; San Francisco Chronicle, 2025). Similarly, during the extreme 2025 fire season in Mediterranean Europe, 1,000 of wildfires affected coastal regions, including a large-scale event in southern France that burned over 16,000 hectares in and around the Aude region (Energy News, 2025; Insurance Journal, 2025). The associated economic losses have been substantial, with wine grapes often accounting for a large share of agricultural damage.

Yet wildfire is rarely incorporated into quantitative projections of viticultural suitability or wine quality. Most modeling frameworks treat climate change as a smooth driver of gradual shifts, rather than as a process that also reshapes disturbance regimes. This omission represents a critical gap. Climate change may simultaneously improve certain climatic conditions for grape ripening while sharply increasing exposure to wildfire hazard, potentially offsetting or reversing anticipated benefits. From the perspective of agricultural resilience, true adaptive capacity depends not only on finding optimal thermal niches but also on minimizing exposure to extremes that can cause catastrophic losses. Impact assessment frameworks without accounting for shifting disturbance regimes risk encouraging maladaptation by identifying potential expansion zones based on favorable mean conditions while neglecting their vulnerability to extreme hazards.

Here, we address this gap by developing an integrated assessment of climate change impacts on wine grape cultivation that explicitly accounts for wildfire weather. We focus on California, a globally prominent wine-producing region that is also among the most wildfire-prone landscapes worldwide. By jointly analyzing climatic suitability for wine grape cultivation, wine quality indicators and spatiotemporal wildfire weather, we move beyond conventional assessments that consider these processes in isolation. Specifically, we (i) model present and future climatic suitability for wine grape cultivation, (ii) project climate-driven changes in wine quality, and (iii) analyse how incorporating wildfire aspects modifies these projections. By integrating gradual climatic trends with climate-driven disturbance processes, this study provides a more realistic framework for assessing the future of wine production in fire-prone regions.

2 Study area and dataset

2.1 Study area

This study focuses on the state of California in the US (Figure 1). California is characterized predominantly by a Mediterranean-type climate, with cool, wet winters and long, dry summers (Peel et al., 2007). This pronounced precipitation seasonality is governed by large-scale atmospheric circulation: wintertime westerlies facilitate the intrusion of moist air masses, whereas the summertime expansion of subtropical high-pressure systems over the northeastern Pacific suppresses rainfall. Along the coast, the cold California Current maintains low sea surface temperatures, cooling the lower atmosphere and promoting marine fog and sea breezes (Checkley and Barth, 2009). Much of California experiences large diurnal temperature ranges, with cool nights moderated by marine influence in coastal areas and strong radiative heating and cooling in inland regions (Daly et al., 2002).

Figure 1

These climatic conditions have made California one of the world's most important wine-producing regions (Parker et al., 2024). The state accounts for approximately 80% of total wine production in the United States, with viticulture distributed across diverse environments (Lapsley et al., 2019; Wine Institute, 2025). Premium wine regions such as Napa Valley and Sonoma Valley are concentrated in coastal and northern areas where marine influences are strong, whereas large-scale grape production is centered in inland regions such as the Central Valley, which underpins the state's overall production volume. Wine production in California dates back to the late eighteenth century and has since developed into a globally influential industry through continued technological and agronomic advances.

The same climatic features that favor viticulture also contribute to high wildfire activity. Prolonged summer dryness and elevated temperatures reduce vegetation moisture, while winter-dominated precipitation leads to critically low soil moisture and atmospheric humidity during the fire season. California has experienced a high frequency of severe wildfires in recent years (Brown et al., 2023), with large events in 2017, 2018, 2020, 2021, and 2025 affecting major wine-growing regions, including Napa, Sonoma, and Mendocino, and burning extensive areas that included vineyards (Higuera and Abatzoglou, 2021; Hiraga and Kavvas, 2021; Safford et al., 2022; Vinall, 2025). These wildfire impacts have resulted in substantial economic losses for the wine sector, underscoring California as a region where climatic suitability for viticulture and wildfire risk are tightly intertwined.

2.2 Dataset

2.2.1 Vineyard locations and vintage ratings

Geographic coordinates of vineyards associated with operating wineries were obtained from the California Wine Institute (California Wine Institute, 2026), yielding a total of 379 locations used as presence data in the species distribution modeling (SDM) framework, adapted for cultivated viticulture suitability (Figure 1). The majority of sites are concentrated in the North Coast region, including Napa Valley and Sonoma Valley. These areas are dominated by premium varieties such as Cabernet Sauvignon and Chardonnay. Additional vineyards are distributed across the Central Coast, including San Luis Obispo, Santa Barbara and Monterey, where strong marine influence favors varieties such as Pinot Noir and Riesling in coastal zones, while larger diurnal temperature ranges in inland areas support cultivars such as Merlot and Zinfandel (California Wine Institute, 2026).

As a proxy for wine quality at the level of grape variety and viticultural region, we used the wine ratings provided by the Vintage Charts published by Wine Spectator (Wine Spectator, 2025). These ratings are compiled by professional tasters and reflect an integrated evaluation of sensory attributes, including balance, complexity, aroma and flavor profile. Annual scores are reported on a 100-point scale and grouped into qualitative rating categories. The rating categories include five quality classes: 95–100 (classic), 90–94 (outstanding), 85–89 (very good), 80–84 (good), and 75–79 (mediocre). These categorical quality classes were used as the target variable in the wine quality modeling framework. Such vintage ratings have been widely used in previous studies as a surrogate indicator of wine quality, particularly in analyses examining relationships between climate, quality and market outcomes (Davis et al., 2019; Kumar et al., 2023). Although multiple wine rating systems have been developed by different critics and institutions over time, previous studies have shown that these independently compiled evaluations are highly correlated with one another (Jones, 1997; Jones et al., 2005). Among them, Wine Spectator's Vintage Charts is one of the most commonly used metrics (Nemani et al., 2001; Jones and Goodrich, 2008; Ramirez, 2008; Charlin and Cifuentes, 2025; Whitnall and Alston, 2025). This study used annual vintage rating categories for Chardonnay and Pinot Noir in Sonoma from 1996 to 2023, obtained from the Wine Spectator Vintage Charts, as a proxy for wine quality (Table 1). Chardonnay and Pinot Noir were selected not only for their global prominence as premium varieties but also for their contrasting physiological sensitivities to climatic and fire-weather stressors. Sonoma was chosen as the primary focus for the quality analysis due to its distinctive local climate and its standing as one of the most wildfire-prone premium wine regions in the world.

2.2.2 Climatic data

Climate data representing present and future conditions over the study domain were obtained from the Multivariate Adaptive Constructed Analogs (MACA) dataset (Abatzoglou and Brown, 2012). MACA provides statistically downscaled climate projections for the conterminous United States on a regular grid with a spatial resolution of approximately 4 km × 4 km and a daily temporal resolution. MACA applies a multivariate statistical downscaling framework that combines bias correction with a constructed analogs approach, whereby large-scale climate patterns simulated by GCMs are matched to observed historical patterns and their associated fine-scale spatial structures. This method has been shown to perform well in regions with complex topography, such as the western United States, by preserving observed spatial variability while maintaining consistency with GCM-projected climate change signals (Abatzoglou and Brown, 2012). MACA has been widely used to assess climate change impacts on agriculture, including viticulture, supporting its suitability for the present analysis (Parker et al., 2024; Wu et al., 2025).

Climatic variables were obtained for three time periods: a present climate baseline (1976–2005, 30-year mean), a mid-century period (2040–2069), and a late-century period (2070–2099). Future climatic variables were computed under two Representative Concentration Pathway scenarios, RCP4.5 and RCP8.5. Climate projections for the mid- and late-century periods were based on five GCMs: MRI-CGCM3, CNRM-CM5, HadGEM2-ES, IPSL-CM5A-MR, and CanESM2. These models were shown to exhibit differing sensitivities and response characteristics to future changes in temperature and precipitation. By using multiple GCMs, we aimed to reduce uncertainty associated with any single model and to enhance the robustness of projected climate impacts. For all subsequent analyses, we focused on the multi-model ensemble mean. This approach is consistent with established practices in regional viticultural assessments, as it highlights the consensus climate change signal across multiple GCMs while providing a stable and integrated baseline for evaluating agricultural impacts.

3 Methods

3.1 Climatic variables and metrics

This study used the climatic variables and metrics listed in Table 2 to model viticulture suitability and wine quality. The selection of climatic variables is based on the previous studies modeling viticultural suitability in similar regions (Hannah et al., 2013; Roehrdanz and Hannah, 2016). This selection aligns with an extensive body of regional literature focused on California and other wine producing regions, where heat accumulation, extreme temperature thresholds, and moisture availability have been consistently identified as the primary drivers of spatial suitability and grape quality (Jones, 2004; Van Leeuwen and Darriet, 2016; Whitnall and Alston, 2025). The growing season was defined as the period from 1 April to 31 October, corresponding to the typical grapevine growing period in the Northern Hemisphere. All variables and derived metrics were calculated as ensemble means of the five GCMs under present and future climate conditions. Detailed computations for growing degree days (GDD), vapor pressure deficit (VPD), and precipitation seasonality can be found in Supplementary material. Figure 2 shows the spatial distributions of five GCM-averaged climatic variables and metrics over California under present climate.

Figure 2

3.2 Fire Weather Index (FWI)

To characterize wildfire hazard, we employed the Fire Weather Index (FWI), a widely used metric of potential fire occurrence and intensity. FWI forms part of the Canadian Forest Fire Danger Rating System, originally developed to quantify fire behavior in Canadian boreal forests (van Wagner, 1987). The FWI combines information on fuel conditions and fire spread potential into a single metric. Fuel availability is represented through a build-up component that integrates moisture conditions across multiple fuel layers. In contrast, the rate of fire spread is primarily controlled by short-term weather variability, particularly wind speed and other daily atmospheric conditions. Together, these components allow FWI to reflect both antecedent moisture status and immediate meteorological drivers of fire behavior. Details of the FWI calculation are provided in the Supplementary Information. Previous studies have demonstrated its applicability across a broad range of vegetation types and climatic regions. In particular, FWI has been shown to capture intra-seasonal to interannual variability in fire activity at regional to global scales, especially in environments where fire occurrence is primarily constrained by fuel flammability (Abatzoglou et al., 2018, 2019). Flannigan et al. (2010) demonstrated that higher FWI values were linked to increased fire frequency and burned area in boreal forests during warm, dry years. He et al. (2025) assessed changes in fire danger across western North America by calculating FWI under future climate scenarios, suggesting elevated wildfire danger under future climates. Given the challenges in reliably projecting future wildfire occurrence and spread, and the demonstrated ability of FWI to represent fire danger, we used FWI as a proxy for wildfire hazard. It is important to emphasize that the FWI characterizes the atmospheric and fuel moisture conditions conducive to fire ignition and spread; it does not, however, represent a deterministic prediction of actual fire occurrence, which also depends on non-meteorological factors such as ignition sources and human activity.

This study evaluated changes in the number of days exceeding the 99th percentile of the Fire Weather Index (FWI99), as fire growth tends to occur preferentially under extreme fire danger conditions (Barbero et al., 2014). Similar percentile-based approaches have been used to assess changes in extreme wildfire danger under climate change (Abatzoglou et al., 2019; Touma et al., 2022; Barik and Baidya Roy, 2023). First, for each 4 km × 4 km grid cell, the 99th percentile of FWI under the present climate was calculated. For the mid-century (2040–2069) and late-century (2070–2099) periods, the number of days exceeding this present-climate FWI99 threshold was counted for each grid cell. The baseline exceedance frequency under the present climate was then used to compute the relative change in future periods. These calculations were conducted separately for each of the five GCMs, and ensemble means were derived for each emission scenario and time slice. The derived relative changes were used in the geographical viticulture suitability analysis. The spatial distribution of FWI99 under the present climate is shown in Figure 2f.

Furthermore, we calculated the spatially-averaged number of days exceeding present-climate FWI99 for each viticultural region (e.g., Sonoma in Figure 1). First, the present-climate FWI99 threshold was calculated as the spatial average across the five GCMs within the viticultural region. We then computed annual counts of days exceeding this threshold for the future period (2040–2099). Years in which the annual number of exceedance days surpassed the present-climate FWI99 threshold were classified as “FWI99 exceeding years,” whereas those below the threshold were defined as “non-exceeding years.” This classification was used for the subsequent wine score analysis.

3.3 Species distribution modeling for spatial viticultural suitability assessment

Viticultural suitability was modeled using MaxEnt algorithm, a machine-learning algorithm widely applied to estimate species distributions from presence-only data (Phillips et al., 2006; Phillips and Dudík, 2008). MaxEnt has been extensively used in previous assessments of climate-driven changes in viticultural suitability across multiple regions, supporting its application in this study (Tóth and Végvári, 2016; del Río et al., 2021, 2024; Bai et al., 2022; Hiraga and Matsumoto, 2025). MaxEnt estimates the relative suitability of each grid cell by identifying the probability distribution of maximum entropy constrained by environmental conditions at known occurrence locations (Phillips, 2017). Formally, the model corresponds to a Gibbs (exponential) distribution in which environmental predictors are weighted to maximize a penalized log-likelihood of the presence data. To prevent overfitting, MaxEnt applies L1 regularization, which shrinks model coefficients by imposing a penalty on their absolute values, thereby balancing model fit and complexity (Tibshirani et al., 2005). A detailed description of the underlying mathematical formulation is provided in Phillips et al. (2006). Because model performance can be sensitive to the choice of regularization strength (Radosavljevic and Anderson, 2014), we evaluated a range of regularization multipliers (β = 0.1, 1, and 10; Furlong et al., 2022) and generated an ensemble of parameter settings.

The model was developed using present-climate conditions and vineyard occurrence (i.e., presence) data, and subsequently projected onto future climate scenarios. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC). Presence data were randomly partitioned into 80% for training and 20% for testing, and the model was run with 100 replicates and a maximum of 1,000 iterations to ensure convergence. Suitability outputs were expressed as logistic values ranging from 0 to 1. Variable importance was assessed using percent contribution, permutation importance, and jackknife tests, with values averaged across replicate runs. This modeling framework produced three suitability estimates for the present climate (corresponding to different regularization settings) and 15 future projections per emission scenario (three regularization settings × five GCMs), enabling the assessment of both parameter and climate-model uncertainty.

3.4 Multiclass classification modeling for wine quality assessment

Wine quality was modeled using a Random Forest–based multiclass classification algorithm implemented in the scikit-learn library (Pal, 2005; Pedregosa et al., 2011), specifically using the RandomForestClassifier module. Vintage rating categories were treated as the response variable. The model was trained to predict annual ratings for Chardonnay and Pinot Noir in Sonoma region during 1996–2023, using regional-average climatic variables and metrics listed in Table 1 as explanatory variables. Model hyperparameters, including the number of trees, maximum depth, the number of variables considered at each split, and the minimum samples per leaf node, were tuned to maximize the F1 score. This approach ensured parameter settings that prioritized generalization performance. The trained model was subsequently applied to project future annual vintage ratings for 2040–2099 in each grape variety, using regional-average future climatic variables and metrics averaged across the five GCMs. Prior to selecting the Random Forest architecture, we evaluated the performance of other models including multiple linear regression and LightGBM. Such other models tended to underestimate year-to-year variations in vintage rating categories. The RF classifier was ultimately selected for its ability to balance the representation of interannual score fluctuations with the non-linear influences of climate and fire-weather stressors.

4 Results

4.1 Spatial viticultural suitability

4.1.1 Model development and present-climate suitability

The MaxEnt model developed in this study showed strong predictive performance in identifying viticultural locations across all regularization settings. Mean AUC values from 100 replicate runs were 0.969, 0.965, and 0.934 for β values of 0.1, 1, and 10, respectively. Although all models performed well (AUC > 0.93), lower regularization parameters yielded higher AUC values, suggesting a greater risk of overfitting. ROC curves are provided in the Supplementary material.

The spatial distribution of viticultural suitability under the present climate, estimated using different regularization parameters, is shown in Figure 3. As the regularization parameter decreased, suitability patterns became increasingly localized, with stronger spatial contrasts reflecting tighter model fit. High suitability values (≥ 0.8) were identified in major wine-producing regions with high vineyard density, including Napa, Sonoma, Santa Barbara, and San Luis Obispo. In addition, relatively high suitability (≥ 0.7) was estimated in areas that are not traditionally considered primary production centers, such as the San Francisco Bay Area, Ramona Valley, and Yucaipa Valley. Notably, Contra Costa County—part of the San Francisco Bay Area—and Yucaipa Valley were designated as American Viticultural Areas (AVAs) after 2024, and these regions have recently attracted increasing attention within the wine industry (TTB, 2024a,b).

Figure 3

Table 3 summarizes the percent contribution and permutation importance of climatic variables in the suitability models constructed under different regularization settings. Overall, the models showed strong dependence on annual total precipitation and annual mean vapor pressure deficit (VPD). Precipitation seasonality and growing season cumulative temperature also contributed substantially to model performance. The jackknife test on test gain showed that growing season cumulative temperature and the minimum temperature of the coldest month during the growing season consistently showed high test gain when used in isolation across all regularization settings (Supplementary material). Overall, four variables—annual total precipitation, annual mean vapor pressure deficit, growing season cumulative temperature, and the minimum temperature of the coldest month during the growing season—were identified as key contributors to model generalization, demonstrating strong predictive performance on independent test data. The response curves for Annual PRECIP, VPD, GDD, and TMIN show the ranges of these variables associated with increasing viticultural suitability Supplementary material.

4.1.2 Suitability changes with climate change

Figure 4 illustrates ensemble-mean projections of changes in viticultural suitability under each RCP scenario for the mid- and late-century periods. Each panel reflects the average of 15 simulations derived from combinations of three regularization settings and five GCMs. Substantial declines (exceeding 0.5) were projected across all scenarios in Napa and Sonoma, currently the most intensively cultivated wine regions in California. Similar pronounced decreases were also identified in other major wine-producing areas, including San Luis Obispo and Santa Barbara. In contrast, suitability was projected to increase over time in Mendocino, located north of Napa and Sonoma. Increases were also identified in Monterey and in mid to south coastal areas, in some cases reaching up to 0.5 under the mid-century RCP8.5 scenario. Overall, suitability tended to increase in northern and coastal regions, while declining in lowland areas. Under the late-century RCP8.5 scenario, where climate change impacts were most pronounced, projected increases in Mendocino and Monterey became more moderate (up to 0.3), whereas larger gains were projected further north in the Redwood region. The suitability changes in these key regions are summarized in boxplots under the RCP4.5 and RCP8.5 scenarios (Figure 5). Climate change produces a clear spatial divergence between regions of increasing and decreasing suitability, and emissions mitigation slightly attenuates this pattern.

Figure 4

Figure 5

These projected shifts were largely driven by changes in temperature-related variables. Figure 6 shows density histograms illustrating the climatic niches currently occupied by vineyards and their projected shifts under future scenarios. Under the present climate, vineyards are primarily located in areas with relatively high annual precipitation and low annual mean vapor pressure deficit (VPD), indicating concentration in comparatively humid zones of California. Thermal conditions at vineyard locations are also narrowly constrained, with growing season cumulative temperature (GDD) and minimum temperature of the coldest month distributed within a limited range compared to the broader study region. Comparing present and future climates at vineyard locations shows that precipitation and VPD remain largely within current ranges, although precipitation increases by 200–300 mm under late-century RCP8.5. In contrast, temperature-related variables shift substantially upward. Under late-century RCP8.5, GDD increases by approximately 600–700 degree-days and minimum temperature by 4–5 °C, exceeding the present suitability range. These results suggest that projected shifts in viticultural suitability are primarily driven by temperature increases rather than changes in moisture conditions. Such temperature increases were particularly pronounced in northern regions, inland high-elevation areas, and along the coast (Figure 7).

Figure 6

Figure 7

4.1.3 Incorporating fire weather changes

Figure 8 illustrates the projected relative change in the number of days exceeding the present-climate FWI99 threshold (i.e., FWI99 exceedance days), comparing future and present climate conditions. The results shown represent the ensemble mean across the five GCMs. Overall, an increasing trend in FWI99 exceedance days was projected across large parts of California. The increase was particularly pronounced in northern regions and inland high-elevation areas. These regions broadly coincide with areas exhibiting substantial increases in growing season cumulative temperature and minimum temperature of the coldest month (Figure 7), while marked increases along the northern coastal region correspond closely with projected rises in annual mean vapor pressure deficit (Figure 7), suggesting that rising temperatures and intensified atmospheric moisture stress together drive the projected increase in high fire-weather days. In contrast, some inland areas north of Napa showed a decreasing trend in FWI99 exceedance days. Compared with regions experiencing increases, these areas exhibited smaller temperature rises and projected increases in precipitation, which may help maintain higher fuel moisture and thereby reduce fire-spread potential.

Figure 8

Figure 9 presents the integrated spatial distribution of projected changes in viticultural suitability and FWI99 exceedance days, shown for the mid- and late-century RCP8.5 scenarios. Changes in both viticultural suitability and FWI99 exceedance days were averaged within each administrative region of California. Overall, regions exhibiting larger increases in viticultural suitability also tend to show greater increases in FWI99 exceedance days. This pattern is particularly evident in northern coastal areas and inland high-elevation regions of California, where projected gains in suitability coincide with relatively large increases in FWI99 exceedance days. These findings suggest that areas becoming more climatically suitable for viticulture under warming may simultaneously face heightened wildfire hazard. Napa and Sonoma—currently dominant wine-producing regions—show projected suitability declines and relatively modest increases in FWI99 exceedance days, indicating that while warming may exceed optimal thermal thresholds, fire-weather intensification is less pronounced than in northern and higher-elevation regions.

Figure 9

To jointly evaluate projected shifts in climatic suitability and wildfire hazard, we synthesized changes in viticultural suitability and FWI99 exceedance days using a four-quadrant matrix framework (Figure 10). For the late-century RCP8.5 scenario, which exhibits more pronounced pattern, nearly half of all cells (1,252; 48.5%) exhibited simultaneous increases in suitability and FWI99 exceedance days. These cells were concentrated primarily in Redwood (55.6%), indicating that regions gaining climatic suitability are also likely to face heightened wildfire hazard. Notably, a smaller subset (195 cells) showed increasing suitability alongside decreasing fire-weather hazard, largely concentrated in Mendocino and Monterey. These areas may represent comparatively favorable expansion zones where improved climatic suitability coincides with reduced wildfire constraints. Conversely, 547 cells exhibited concurrent declines in both suitability and FWI99 exceedance days, primarily within current major production regions. In these areas, excessive warming may reduce climatic suitability, while projected increases in precipitation could partially suppress fire-weather conditions. This integrative framework enables identification of future viticultural lands not only on the basis of projected suitability changes, but also by accounting for evolving wildfire hazard. By jointly evaluating climatic potential and fire-weather danger, the analysis provides a more robust basis for land selection, adaptation strategies, and mitigation planning under climate change. In particular, areas in Mendocino and Monterey characterized by increasing suitability and declining fire-weather hazard emerge as especially promising candidates for future viticulture. These regions may represent comparatively resilient expansion zones under a warming climate.

Figure 10

4.2 Wine quality

4.2.1 Model development and predictions of vintage rating categories

The Random Forest–based multiclass classification algorithm showed strong performance in predicting vintage rating categories for 1996–2023, with F1 scores of 0.94 and 0.99 for Chardonnay and Pinot Noir, respectively. Previous studies have modeled wine quality using market price as a proxy and applied multivariate linear regression with climate variables as predictors, achieving coefficients of determination around 0.8 (Corsi and Ashenfelter, 2019). However, the linear assumption may limit the ability to capture non-linear responses, threshold effects, and complex climatic interactions. In contrast, this study employed a tree-based algorithm to account for such non-linearities, resulting in strong predictive performance.

The developed model was applied to project vintage rating categories for Chardonnay and Pinot Noir in the Sonoma region over the period 2040–2099 (Figures 11a–d). Even within the same region (Sonoma), predicted changes in quality categories differed among grape varieties, highlighting the potential importance of selecting varieties suited to regional conditions under future climate scenarios. Moreover, the frequency of the highest category was projected to decrease under the RCP8.5 scenario compared with the RCP4.5 scenario. These findings suggest that future changes in wine quality will not be uniform, but will vary depending on the combination of grape variety and region.

Figure 11

4.2.2 Incorporating fire weather changes

The future period was categorized into two groups based on the Sonoma-averaged FWI predictions (Figure 11e and f): “exceeding years,” which surpassed present-climate FWI99, and “non-exceeding years,” which did not. The predicted number of years exceeding the present-climate FWI99 threshold is projected to be substantially greater under the RCP8.5 scenario (21 years) compared to the RCP4.5 scenario (13 years). Years with relatively high FWI99 tended to correspond to years with low annual cumulative precipitation (Supplementary material), suggesting that interannual variations in moisture conditions are one of the primary factors controlling variability in fire danger.

Figure 12 summarizes the frequency of predicted rating categories during 2040–2099 for Chardonnay and Pinot Noir under RCP4.5 and RCP8.5 scenarios. The average rating category for 2040–2099 was predicted to be higher under the RCP4.5 scenario than under RCP8.5 for both Chardonnay and Pinot Noir in Sonoma, suggesting that higher greenhouse gas emissions may lead to a decline in wine quality. For Chardonnay, the average rating category is predicted to be higher in non-exceedance years than in FWI99 exceedance years, suggesting that wine quality tends to be higher under lower fire danger conditions (Figures 12a, b). In contrast, Pinot Noir showed the opposite pattern, with higher wine quality tending to co-occur with high fire danger conditions (Figures 12c, d). Although the differences in predicted rating categories between FWI99 exceedance and non-exceedance years were relatively small, incorporating a fire danger perspective indeed altered the distribution and frequency of the predicted scores. For instance, the relative frequency patterns of certain rating categories differed between FWI99 exceedance and non-exceedance years for both Chardonnay (e.g., the relative occurrence of categories 4 and 5) and Pinot Noir (e.g., the relative occurrence of categories 2 and 3 under the RCP8.5 scenario). Such information is important when considering varietal selection for long-term cultivation. As such, a framework that jointly analyzes wine quality indicators and wildfire danger is important, moving beyond conventional assessments that treat these processes in isolation.

Figure 12

5 Discussion and concluding remarks

This study provides an integrated assessment of how climate change and wildfire weather may jointly reshape wine grape cultivation in California. To our knowledge, few studies have explicitly integrated wildfire weather into projections of both viticultural suitability and wine quality, highlighting the importance of considering disturbance regimes alongside gradual climatic trends when evaluating future wine-growing potential. Integrating these processes may help guide regional planning, varietal selection, and climate adaptation strategies for wine production in fire-prone regions.

5.1 Summary and discussion of the findings

• - Projected changes in climatic suitability indicate a spatial redistribution of viticulture across California. Suitability declines were projected in major wine-producing regions such as Napa and Sonoma, primarily driven by increasing growing season temperatures. In contrast, increases in suitability were projected in northern coastal regions and some higher-elevation areas. These patterns are broadly consistent with previous studies projecting poleward or upslope shifts in viticultural suitability under climate warming (Jones et al., 2010; Hannah et al., 2013; Fraga et al., 2016). However, the present results demonstrate that many newly suitable regions may simultaneously experience increases in extreme fire-weather conditions, introducing an additional dimension of climate risk rarely considered in previous assessments.

• - Meanwhile, we found comparatively favorable expansion zones (e.g., Mendocino and Monterey) increasing suitability alongside decreasing fire-weather hazard in future climate. While future changes in viticulture suitability (Roehrdanz and Hannah, 2016) and fire weather (Goss et al., 2020; Rother et al., 2022) have been examined independently, our integrative spatial analysis highlights the spatially heterogeneous nature of these dual pressures. This encompasses not only the trade-offs where increased suitability overlaps with heightened fire danger but also the identification of resilient expansion zones—such as parts of Mendocino and Monterey—where improved climatic conditions coincide with decreasing wildfire hazard. These findings offer a robust framework for developing adaptation strategies that pinpoint favorable opportunities while proactively avoiding maladaptation in fire-prone landscapes. The wine quality analysis further indicates that climate change may affect grape varieties differently even within the same region. Chardonnay tended to achieve higher predicted scores in years with lower fire danger conditions, whereas Pinot Noir showed the opposite tendency, with higher scores occurring more frequently in years associated with elevated fire-weather conditions. Although the differences in predicted scores between fire-danger categories were relatively small, incorporating wildfire weather altered the distribution of predicted quality categories. This suggests that wildfire-related climatic conditions may influence the probability of certain quality outcomes. From a viticultural perspective, such information may be relevant when considering long-term varietal selection and adaptation strategies in fire-prone regions.

• - For policymakers and land-use planners, these results emphasize the need for integrated climate-hazard zoning. Our integrated framework suggests that future insurance models and agricultural subsidies should transition from single-variable assessments to compound evaluations that account for both gradual yield changes and extreme disturbance losses.

5.2 Limitations and future research directions

Some limitations should be acknowledged. First, wildfire hazard was represented using the Fire Weather Index rather than explicit modeling of wildfire occurrence, spread, or smoke exposure. While FWI is widely used as a proxy for fire danger and correlates with fire activity, actual wildfire impacts depend on additional factors including ignition sources, vegetation structure, fuel conditions, and fire suppression efforts. Therefore, our use of FWI primarily characterizes the meteorological potential for fire behavior rather than providing a deterministic prediction of fire occurrences. This choice was necessitated by the focus on long-term climate projections. Consequently, the favorable areas identified in our analysis should be interpreted as zones where the meteorological hazard is relatively lower, though they may still remain vulnerable if ignitions occur within fuel-dense landscapes. Furthermore, as FWI does not consider the spatial complexities of wildfire spread and smoke transport, which are critical for viticulture, these findings should be viewed as a baseline hazard assessment to be refined by future studies incorporating fire spread or smoke transport analyses.

Second, wine quality was represented using vintage rating categories derived from Wine Spectator. Although such ratings are widely used in empirical studies and correlate with other rating systems, they remain subjective indicators and may not fully capture biochemical or sensory characteristics of wines. In addition, the analysis focused on two grape varieties and a limited number of regions and years due to data availability. The reliance on these proxies raises important questions regarding the robustness of our projections. The exceptionally high F1 scores reported in this study should be interpreted as high-fidelity reproductions of historical patterns within the specific environmental niche of Sonoma, rather than as an indicator of broader predictive performance in diverse viticultural contexts. By utilizing the full dataset for training, we optimized the model to learn the historical sensitivities of premium varieties to local climate and fire-weather conditions. However, this carries an inherent risk of overfitting to the historical period. Our future projections for 2040–2099 are thus based on the assumption that these established regional archetypes of climate-quality relationships will persist, though the potential for future physiological tipping points remains an area for further investigation. Ideally, the findings should be cross-validated by establishing empirical relationships between expert ratings and objective chemical profiles, such as phenolic composition or sugar-acid balances (e.g., Cortez et al., 2009; Sáenz-Navajas et al., 2015). Such an approach would confirm whether the projected quality shifts align with fundamental biophysical changes in the grapes. However, at present, the availability of longitudinal, spatially explicit chemical data is insufficient for regional-scale modeling, necessitating the use of professional ratings as a pragmatic surrogate (Jones and Goodrich, 2008; Kumar et al., 2023; Kopsacheilis et al., 2024). This limitation also affects the generalizability of our results; while the model is robust for the premium archetypes of Sonoma, it may not account for emerging regions where quality benchmarks are still evolving. Future research should prioritize the integration of objective chemical metrics to enhance the scientific robustness and broader applicability of viticultural quality assessments.

A notable limitation of our modeling framework is the omission of non-climatic components of terroir, such as soil physical properties, topography (slope and aspect), and human management practices. While climate is the primary driver of regional viticultural shifts, local-scale suitability may be ultimately refined by soil water-holding capacity and site-specific drainage. Furthermore, human adaptations—including irrigation management, canopy manipulation, and rootstock selection—can significantly buffer climatic stress. Our results should therefore be interpreted as a projection of climatic suitability potential rather than a comprehensive assessment of future terroir. Future research incorporating high-resolution pedological and topographic layers alongside dynamic management scenarios would provide a more granular understanding of viticultural resilience.

Finally, the modeling framework assumes that historical relationships between climate variables and viticultural outcomes remain valid under future climates. Extrapolation beyond the historical climate range may introduce uncertainty, and vineyard management adaptations such as irrigation, canopy management, or cultivar substitution were not explicitly considered. This reliance on stationarity means that our model may not fully capture potential physiological tipping points—such as extreme heat-induced metabolic shifts—that could occur once historical thermal thresholds are breached. Future studies may consider coupling climate projections with dynamic plant-physiology models that explicitly incorporate adaptive management scenarios. Integrating empirical modeling and such a dynamic modeling will move beyond simple extrapolation, allowing for a more robust assessment of climate change impacts in an increasingly non-stationary environment.

Our findings have direct implications for the long-term resilience of the viticulture industry. The identification of resilient expansion zones—where increased suitability coincides with stable or decreasing wildfire hazard—provides a spatial blueprint for land-use planning. Conversely, the quantified increase in fire hazard within newly suitable areas highlights a trade-off. For practitioners, this suggests that the benefits of expanding into climatically favorable niches needs to be balanced against rising insurance costs and the increased necessity for fire-resilient vineyard designs. By providing a spatial template of these trade-offs, our results are expected to serve as a strategic guideline for adaptation, helping producers and policymakers avoid maladaptive investments in high-vulnerability landscapes. Future research could extend this framework in several ways. Integrating wildfire models that incorporate ignition probability, fuel dynamics, and fire spread processes would improve representation of disturbance risk. Incorporating smoke exposure and smoke taint perspectives would also be valuable, as smoke represents one of the most significant wildfire-related threats to wine production. Expanding analyses to additional grape varieties and viticultural regions, while considering adaptation strategies such as vineyard relocation and cultivar substitution, would further improve understanding of the resilience of wine production systems under climate change.

References

• 1

  Abatzoglou J. T Brown T. J. A. , (2012). comparison of statistical downscaling methods suited for wildfire applications. Int. J. Climatol.32, 772–780 doi: 10.1002/joc.2312

  • CrossRef
  • Google Scholar
  • View reference in article

• 2

  AbatzoglouJ. T.WilliamsA. P.BarberoR. (2019). Global emergence of anthropogenic climate change in fire weather indices. Geophys. Res. Lett.46, 326–336. doi: 10.1029/2018GL080959

  • CrossRef
  • Google Scholar
  • View reference in article

• 3

  AbatzoglouJ. T.WilliamsA. P.BoschettiL.ZubkovaM.KoldenC. A. (2018). Global patterns of interannual climate-fire relationships. Glob. Change Biol.24, 5164–5175. doi: 10.1111/gcb.14405

  • CrossRef
  • Google Scholar
  • View reference in article

• 4

  BaiH.SunZ.YaoX.KongJ.WangY.ZhangX.et al. (2022). Viticultural suitability analysis based on multi-source data highlights climate-change-induced decrease in potential suitable areas: a case analysis in Ningxia, China. Remote Sens.14:3717. doi: 10.3390/rs14153717

  • CrossRef
  • Google Scholar
  • View reference in article

• 5

  BarberoR.AbatzoglouJ. T.SteelE. A.LarkinN. K. (2014). Modeling very large-fire occurrences over the continental United States from weather and climate forcing. Environ. Res. Lett.9:124009. doi: 10.1088/1748-9326/9/12/124009

  • CrossRef
  • Google Scholar
  • View reference in article

• 6

  BarikA.Baidya RoyS. (2023). Climate change strongly affects future fire weather danger in Indian forests. Commun. Earth Environ.4:452. doi: 10.1038/s43247-023-01112-w

  • CrossRef
  • Google Scholar
  • View reference in article

• 7

  BrownP. T.HanleyH.MaheshA.ReedC.StrenfelS. J.DavisS. J.et al. (2023). Climate warming increases extreme daily wildfire growth risk in California. Nature621, 760–766. doi: 10.1038/s41586-023-06444-3

  • CrossRef
  • Google Scholar
  • View reference in article

• 8

  California Wine Institute. (2026). Navigate the state: explore California's wine regions and AVAs. Available online at: https://discovercaliforniawines.com/wine-map-winery-directory/ (Accessed February 2026).

  • Google Scholar

• 9

  CardellM. F.AmengualA.RomeroR. (2019). Future effects of climate change on the suitability of wine grape production across Europe. Reg. Environ. Change19, 2299–2310. doi: 10.1007/s10113-019-01502-x

  • CrossRef
  • Google Scholar
  • View reference in article

• 10

  CBS News (2025). Early estimates of Pickett Fire's economic impact to Napa wine industry around $65M, Available online at: https://www.cbsnews.com/sanfrancisco/news/pickett-fire-napa-wine-industry-economic-impact/ (Accessed February 2026).

  • Google Scholar
  • View reference in article

• 11

  CharlinV.CifuentesA. (2025). The Effects of climate change on the quality of Chile's Maipo Valley Cabernet Sauvignon wines. J. Wine Res.36, 39–53. doi: 10.1080/09571264.2025.2463097

  • CrossRef
  • Google Scholar
  • View reference in article

• 12

  CheckleyD. M.BarthJ. A. (2009). Patterns and processes in the California current system. Prog. Oceanogr.83, 49–64. doi: 10.1016/j.pocean.2009.07.028

  • CrossRef
  • Google Scholar
  • View reference in article

• 13

  CioccaI.FernándezA.JaqueE.JustinoF.ShumacherV.da SilvaA. S.et al. (2023). Increased wildfire hazard along South-Central Chile under the RCP8. 5 scenario as revealed by high-resolution modeling. Environ. Res. Lett.18:034023. doi: 10.1088/1748-9326/acba33

  • CrossRef
  • Google Scholar
  • View reference in article

• 14

  CorsiA.AshenfelterO. (2019). Predicting Italian wine quality from weather data and expert ratings. J. Wine Econ.14, 234–251. doi: 10.1017/jwe.2019.41

  • CrossRef
  • Google Scholar
  • View reference in article

• 15

  CortezP.CerdeiraA.AlmeidaF.MatosT.ReisJ. (2009). Modeling wine preferences by data mining from physicochemical properties. Decis. Support Syst.47, 547–553. doi: 10.1016/j.dss.2009.05.016

  • CrossRef
  • Google Scholar
  • View reference in article

• 16

  DalyC.GibsonW. P.TaylorG. H.JohnsonG. L.PasterisP. P. (2002). High-resolution spatial modeling of daily weather elements for a catchment in the western United States. J. Appl. Meteorol.41, 1007–1024.

  • Google Scholar
  • View reference in article

• 17

  DavisR. E.DimonR. A.JonesG. V.BoisB. (2019). The effect of climate on Burgundy vintage quality rankings. OENO One 53. doi: 10.20870/oeno-one.2019.53.1.2359

  • CrossRef
  • Google Scholar
  • View reference in article

• 18

  De OrdunaR. M. (2010). Climate change associated effects on grape and wine quality and production. Food Res. Int. 43, 1844–1855. doi: 10.1016/j.foodres.2010.05.001

  • CrossRef
  • Google Scholar
  • View reference in article

• 19

  del RíoS.Álvarez-EstebanR.Alonso-RedondoR.ÁlvarezR.Rodríguez-FernándezM. P.González-PérezA.et al. (2024). Applications of bioclimatology to assess effects of climate change on viticultural suitability in the DO León (Spain). Theor. Appl. Climatol.155, 3387–3404. doi: 10.1007/s00704-024-04831-y

  • CrossRef
  • Google Scholar
  • View reference in article

• 20

  del RíoS.Àlvarez-EstebanR.Alonso-RedondoR.HidalgoC.PenasÀ. (2021). A new integrated methodology for characterizing and assessing suitable areas for viticulture: a case study in Northwest Spain. Eur. J. Agron. 131, 126391. doi: 10.1016/j.eja.2021.126391

  • CrossRef
  • Google Scholar
  • View reference in article

• 21

  DrappierJ.ThibonC.RabotA.Geny-DenisL. (2019). Relationship between wine composition and temperature: Impact on Bordeaux wine typicity in the context of global warming. Crit. Rev. Food Sci. Nutr.59, 14–30. doi: 10.1080/10408398.2017.1355776

  • CrossRef
  • Google Scholar
  • View reference in article

• 22

  DupuyJ. L.FargeonH.Martin-StPaulN.PimontF.RuffaultJ.GuijarroM.et al. (2020). Climate change impact on future wildfire danger and activity in southern Europe: a review. Ann. For. Sci.77:35. doi: 10.1007/s13595-020-00933-5

  • CrossRef
  • Google Scholar
  • View reference in article

• 23

  Energy News (2025). French winemakers regret loss of “firebreak” vineyards as blazes spread, Available online at: https://energynews.oedigital.com/climate-change/2025/08/07/french-winemakers-regret-loss-of-firebreak-vineyards-as-blazes-spread (Accessed February 2026)

  • Google Scholar
  • View reference in article

• 24

  FlanniganM. D.StocksB. J.WottonB. M.MartellD. L.SkinnerW. R.MasonJ. A. (2010). Trends in global wildfire potential in a changing climate. For. Ecol. Manag.259, 685–697. doi: 10.1016/j.foreco.2009.09.002

  • CrossRef
  • Google Scholar
  • View reference in article

• 25

  FragaH.SantosJ. A.MalheiroA. C.OliveiraA. A.Moutinho-PereiraJ.JonesG. V. (2016). Climatic suitability of Portuguese grapevine varieties and climate change adaptation. Int. J. Climatol. 36. doi: 10.1002/joc.4325

  • CrossRef
  • Google Scholar
  • View reference in article

• 26

  FriedJ. S.TornM. S.MillsE. (2004). The impact of climate change on wildfire severity: a regional forecast for northern California. Clim. Change64, 169–191. doi: 10.1023/B:CLIM.0000024667.89579.ed

  • CrossRef
  • Google Scholar

• 27

  FurlongM.AdamuA.HicksonR. I.HorwoodP.GolchinM.HoskinsA.RussellT. (2022). Estimating the distribution of Japanese encephalitis vectors in Australia using ecological niche modelling. Trop. Med. Infect. Dis.7:393. doi: 10.3390/tropicalmed7120393

  • CrossRef
  • Google Scholar
  • View reference in article

• 28

  GaitánE.Pino-OtínM. R. (2023). Using bioclimatic indicators to assess climate change impacts on the Spanish wine sector. Atmos. Res.286:106660. doi: 10.1016/j.atmosres.2023.106660

  • CrossRef
  • Google Scholar
  • View reference in article

• 29

  GossM.SwainD. L.AbatzoglouJ. T.SarhadiA.KoldenC. A.WilliamsA. P.et al. (2020). Climate change is increasing the likelihood of extreme autumn wildfire conditions across California. Environ. Res. Lett.15:094016. doi: 10.1088/1748-9326/ab83a7

  • CrossRef
  • Google Scholar
  • View reference in article

• 30

  GreeneA. L.SandhuP. K.HallB.O'BrienR.HaywardR.MurchS. J. (2025). Metabolomics to understand the impacts of wildfires on wines. ACS Food Sci. Technol.5, 3759–3773. doi: 10.1021/acsfoodscitech.5c00386

  • CrossRef
  • Google Scholar
  • View reference in article

• 31

  HannahL.RoehrdanzP. R.IkegamiM.ShepardA. V.ShawM. R.TaborG.et al. (2013). Climate change, wine, and conservation. Proc. Natl. Acad. Sci.110, 6907–6912. doi: 10.1073/pnas.1210127110

  • CrossRef
  • Google Scholar
  • View reference in article

• 32

  HeY.ZhouZ.ImE.-O.KwonH. (2025). Wildfire risk in a changing climate: evaluating fire weather indices and their global patterns with CMIP6 multi-model projections. Weather Clim. Extrem.48:100751. doi: 10.1016/j.wace.2025.100751

  • CrossRef
  • Google Scholar
  • View reference in article

• 33

  HewerM. J.GoughW. A. (2021). Climate change impact assessment on grape growth and wine production in the Okanagan Valley (Canada). Clim. Risk Manag.33:100343. doi: 10.1016/j.crm.2021.100343

  • CrossRef
  • Google Scholar
  • View reference in article

• 34

  HewittS.Hernández-MontesE.DhingraA.KellerM. (2023). Impact of heat stress, water stress, and their combined effects on the metabolism and transcriptome of grape berries. Sci. Rep.13:9907. doi: 10.1038/s41598-023-36160-x

  • CrossRef
  • Google Scholar
  • View reference in article

• 35

  HigueraP. E.AbatzoglouJ. T. (2021). Record-setting climate enabled the extraordinary 2020 fire season in the western United States. Global Change Biol. 27, 1–2. doi: 10.1111/gcb.15388

  • CrossRef
  • Google Scholar
  • View reference in article

• 36

  HiragaY.KavvasM. L. (2021). Hydrological and meteorological controls on large wildfire ignition and burned area in Northern California during 2017–2020. Fire4:90. doi: 10.3390/fire4040090

  • CrossRef
  • Google Scholar
  • View reference in article

• 37

  HiragaY.MatsumotoT. (2025). Viticulture suitability estimates under climate change in Japan. Theor. Appl. Climatol.156:671. doi: 10.1007/s00704-025-05889-y

  • CrossRef
  • Google Scholar
  • View reference in article

• 38

  Insurance Journal (2025). Europe wildfires scorching vineyards, homes as heat stays relentless, Available online at: https://www.insurancejournal.com/news/international/2025/08/08/834994.htm (Accessed February 2026)

  • Google Scholar

• 39

  IrimiaL. M.PatricheC. V.Ro?caB. (2018). Climate change impact on climate suitability for wine production in Romania. Theor. Appl. Climatol.133, 1–14. doi: 10.1007/s00704-017-2156-z

  • CrossRef
  • Google Scholar
  • View reference in article

• 40

  JonesG. V. (1997). A Synoptic Climatological Assessment of Viticultural Phenology Dissertation. University of Virginia, 410

  • Google Scholar

• 41

  JonesG. V. (2004). “Climate change in the western United States grape growing regions,” in VII International Symposium on Grapevine Physiology and Biotechnology. 41–60. doi: 10.17660/ActaHortic.2005.689.2

  • CrossRef
  • Google Scholar
  • View reference in article

• 42

  JonesG. V.DuffA. A.HallA.MyersJ. W. (2010). Spatial analysis of climate in winegrape growing regions in the western United States. Am. J. Enol. Vitic.61, 313–326. doi: 10.5344/ajev.2010.61.3.313

  • CrossRef
  • Google Scholar
  • View reference in article

• 43

  JonesG. V.GoodrichG. B. (2008). Influence of climate variability on wine regions in the western USA and on wine quality in the Napa Valley. Clim. Res.35, 241–254. doi: 10.3354/cr00708

  • CrossRef
  • Google Scholar
  • View reference in article

• 44

  JonesG. V.WhiteM. A.CooperO. R.StorchmannK. (2005). Climate change and global wine quality. Clim. Change73, 319–343. doi: 10.1007/s10584-005-4704-2

  • CrossRef
  • Google Scholar
  • View reference in article

• 45

  KellerM. (2023). Climate change impacts on Vineyards in warm and dry areas. Am. J. Enol. Vitic.74:0740033. doi: 10.5344/ajev.2023.23024

  • CrossRef
  • Google Scholar
  • View reference in article

• 46

  KopsacheilisO.AnalytisP. P.KaushikK.HerzogS. M.BahramiB.DeroyO. (2024). Crowdsourcing the assessment of wine quality: vivino ratings, professional critics, and the weather. J. Wine Econ.19, 285–304. doi: 10.1017/jwe.2024.20

  • CrossRef
  • Google Scholar

• 47

  KumarA. P.MeriluotoL.WattR. (2023). New evidence on using expert ratings to proxy for wine quality in climate change research. J. Wine Res.34, 101–121. doi: 10.1080/09571264.2023.2205117

  • CrossRef
  • Google Scholar

• 48

  LapsleyJ. T.AlstonJ. M.SambucciO. (2019). “The US wine industry,” in The Palgrave Handbook of Wine Industry Economics” (Cham: Springer International Publishing), 105–129. doi: 10.1007/978-3-319-98633-3_5

  • CrossRef
  • Google Scholar

• 49

  MadhusoodananJ. (2021). Wildfires pose a burning problem for wines and winemakers. Proc. Natl. Acad. Sci.118:e2113327118. doi: 10.1073/pnas.2113327118

  • CrossRef
  • Google Scholar

• 50

  MangasR.GonzálezM. R.MartínP.Rodríguez-NogalesJ. M. (2023). Impact of glucose oxidase treatment in high sugar and pH musts on volatile composition of white wines. LWT184:114975. doi: 10.1016/j.lwt.2023.114975

  • CrossRef
  • Google Scholar

• 51

  Martínez-LüscherJ.MatusJ. T.GomèsE.PascualI. (2025). Toward understanding grapevine responses to climate change. J. Exp. Bot.76, 2949–2969. doi: 10.1093/jxb/erae482

  • CrossRef
  • Google Scholar

• 52

  NemaniR. R.WhiteM. A.CayanD. R.JonesG. V.RunningS. W.CoughlanJ. C.et al. (2001). Asymmetric warming over coastal California and its impact on the premium wine industry. Climate Res. 19, 25–34. doi: 10.3354/cr019025

  • CrossRef
  • Google Scholar

• 53

  PalM. (2005). Random forest classifier for remote sensing classification. Int. J. Remote Sens.26, 217–222. doi: 10.1080/01431160412331269698

  • CrossRef
  • Google Scholar

• 54

  ParkerL. E.ZhangN.AbatzoglouJ. T.KisekkaI.McElroneA. J.OstojaS. M. (2024). A variety-specific analysis of climate change effects on California winegrapes. Int. J. Biometeorol.68, 1559–1571. doi: 10.1007/s00484-024-02684-8

  • CrossRef
  • Google Scholar

• 55

  PedregosaF.VaroquauxG.GramfortA.MichelV.ThirionB.GriselO.et al. (2011). Scikit-learn: machine learning in Python. The J. Mach. Learn. Res.12, 2825–2830.

  • Google Scholar

• 56

  PeelM. C.FinlaysonB. L.McMahonT. A. (2007). Updated world map of the Köppen–Geiger climate classification. Hydrol. Earth Syst. Sci.11, 1633–1644. doi: 10.5194/hess-11-1633-2007

  • CrossRef
  • Google Scholar

• 57

  PhillipsS. J. (2017). A brief tutorial on maxent. Available online at from url: http://biodiversityinformatics.amnh.org/open_source/maxent/ (Accessed August 2024).

  • Google Scholar

• 58

  Phillips S. J. Anderson R. P. Schapire R. E. (2006) Maximum entropy modeling of species geographic distributions. Ecol. Model. 190, 231–259. 10.1016/j.ecolmodel.2005.03.026

• 59

  Phillips S. J. Dudík M. (2008) Modeling of species distributions with Maxent: new extensions a comprehensive evaluation. Ecography 31, 161–175. 10.1111/j.0906-7590.2008.5203.x

• 60

  RadosavljevicA.AndersonR. P. (2014). Making better Maxent models of species distributions. J. Biogeogr.41, 629–643. doi: 10.1111/jbi.12227

  • CrossRef
  • Google Scholar

• 61

  RamirezC. D. (2008). Wine quality, wine prices, and the weather: is Napa “different”?. J. Wine Econ. 3, 114–131. doi: 10.1017/S193143610000057X

  • CrossRef
  • Google Scholar

• 62

  RamosM. C.JaraM. A. I.RosilloL.SalinasM. R. (2024). Effect of temperature and water availability on grape phenolic compounds. Sci. Hortic.337:113475. doi: 10.1016/j.scienta.2024.113475

  • CrossRef
  • Google Scholar

• 63

  RoehrdanzP. R.HannahL. (2016). Climate change, California wine, and wildlife habitat. J. Wine Econ.11, 69–87. doi: 10.1017/jwe.2014.31

  • CrossRef
  • Google Scholar

• 64

  RotherD. E.De SalesF.StowD.McFaddenJ. P. (2022). Summer and fall extreme fire weather projected to occur more often and affect a growing portion of California throughout the 21st century. Fire5:177. doi: 10.3390/fire5060177

  • CrossRef
  • Google Scholar

• 65

  RuffaultJ.CurtT.MoronV.TrigoR. M.MouillotF.KoutsiasN.et al. (2020). Increased likelihood of heat-induced large wildfires. Sci. Rep.10:13790. doi: 10.1038/s41598-020-70069-z

  • CrossRef
  • Google Scholar

• 66

  Sáenz-NavajasM. P.AvizcuriJ. M.BallesterJ.Fernández-ZurbanoP.FerreiraV.PeyronD.et al. (2015). Sensory-active compounds influencing wine experts“ and consumers” perception of red wine intrinsic quality. LWT-Food Sci. Technol.60, 400–411. doi: 10.1016/j.lwt.2014.09.026

  • CrossRef
  • Google Scholar

• 67

  SaffordH. D.PaulsonA. K.SteelZ. L.YoungD. J.WaymanR. B. (2022). The 2020 California fire season. Glob. Ecol. Biogeogr.31, 2005–2025. doi: 10.1111/geb.13498

  • CrossRef
  • Google Scholar

• 68

  San Francisco Chronicle (2025). Napa Valley's harvest just began. Available online at: https://www.sfchronicle.com/bayarea/article/pickett-fire-causes-65m-agricultural-damages-21018984.php (Accessed February 2026).

  • Google Scholar

• 69

  TibshiraniR.SaundersM.RossetS.ZhuJ.KnightK. (2005). Sparsity and smoothness via the fused lasso. J. R. Stat. Soc. Ser. B67, 91–108. doi: 10.1111/j.1467-9868.2005.00490.x

  • CrossRef
  • Google Scholar

• 70

  TóthJ. P.VégváriZ. (2016). Future of winegrape growing regions in Europe. Aust. J. Grape Wine Res.22, 64–72. doi: 10.1111/ajgw.12168

  • CrossRef
  • Google Scholar

• 71

  ToumaD.StevensonS.SwainD. L.SinghD.KalashnikovD. A.HuangX. (2022). Climate change increases risk of extreme rainfall following wildfire. Sci. Adv.8:eabm0320. doi: 10.1126/sciadv.abm0320

  • CrossRef
  • Google Scholar

• 72

  TTB (2024a). Establishment of the Yucaipa Valley Viticultural Area (Final Rule, FR Doc No. 2024-08868, 27 CFR Part 9). Fed. Regist. 89, 31632–31636.

  • Google Scholar

• 73

  TTB (2024b). Alcohol and Tobacco Tax and Trade Bureau: Establishment of the Contra Costa Viticultural Area and modification of the San Francisco Bay and Central Coast viticultural areas (Final Rule, FR Doc. 2024-05476, 27 CFR Part 9). Fed. Regist. 89, 18797–18802.

  • Google Scholar

• 74

  TurcoM.LlasatM. C.von HardenbergJ.ProvenzaleA. (2014). Climate change impacts on wildfires in a Mediterranean environment. Clim. Change125, 369–380. doi: 10.1007/s10584-014-1183-3

  • CrossRef
  • Google Scholar

• 75

  Van LeeuwenC.DarrietP. (2016). The impact of climate change on viticulture and wine quality. J. Wine Econ.11, 150–167. doi: 10.1017/jwe.2015.21

  • CrossRef
  • Google Scholar

• 76

  Van LeeuwenC.SeguinG. (2006). The concept of terroir in viticulture. J. Wine Res.17, 1–10. doi: 10.1080/09571260600633135

  • CrossRef
  • Google Scholar

• 77

  Van LeeuwenC.SgubinG.BoisB.OllatN.SwingedouwD.ZitoS.et al. (2024). Climate change impacts and adaptations of wine production. Nat. Rev. Earth Environ.5, 258–275. doi: 10.1038/s43017-024-00521-5

  • CrossRef
  • Google Scholar

• 78

  van WagnerC. V. (1987). Development and structure of the Canadian forest fire weather index system, Forestry Technical Report. 35. Chalk River, ON: Canadian Forest Service, viii+−37.

  • Google Scholar

• 79

  VinallF. (2025). Pickett Fire prompts evacuation orders in California's Napa County. The Washington Post.

  • Google Scholar

• 80

  WhitnallS.AlstonJ. (2025). Climate, weather, and collective reputation. J. Wine Econ.20, 122–167. doi: 10.1017/jwe.2025.8

  • CrossRef
  • Google Scholar

• 81

  WilliamsA. P.AbatzoglouJ. T.GershunovA.Guzman-MoralesJ.BishopD. A.BalchJ. K.et al. (2019). Observed impacts of anthropogenic climate change on wildfire in California. Earth's Future7, 892–910. doi: 10.1029/2019EF001210

  • CrossRef
  • Google Scholar

• 82

  Wine Institute (2025). California and US wine production. Available online at: https://wineinstitute.org/our-industry/statistics/california-us-wine-production (Accessed December 2025).

  • Google Scholar

• 83

  Wine Spectator (2025). Vintage Charts, Available online at: https://www.winespectator.com/vintage-charts (Accessed December 2025).

  • Google Scholar

• 84

  WoodA.SotoS. V.GambettaG.SchoberD.JeffersE.CoulsonT. (2024). Modelling the climate changing concentrations of key red wine grape quality molecules. OENO One 58. doi: 10.20870/oeno-one.2024.58.4.8086

  • CrossRef
  • Google Scholar

• 85

  WuS.ZikalalaP. G.AlbaS.Jarvis-SheanK. S.KisekkaI.SegaranM.et al. (2025). Advancing the modeling of future climate and innovation impacts on perennial crops to support adaptation: a case study of California almonds. Earths Fut.13, e2024EF005033. doi: 10.1029/2024EF005033

  • CrossRef
  • Google Scholar

• 86

  ZakowskiE.ParkerL. E.JohnsonD.AguirreJ.OstojaS. M. (2023). California wine grape growers need support to manage risks from wildfire and smoke. Calif. Agric.77, 40–48. doi: 10.3733/ca.2023a0006

  • CrossRef
  • Google Scholar