Entropy Model Research Articles

Inflation and dark matter from the low entropy hypothesis and modeling mechanism of modified gravity

Abstract The hypothesis of low entropy in the initial state of the Universe usually explains the observed entropy increase is in only one time direction: the thermodynamic arrow of time. The Hamiltonian formalism is commonly used in the context of general relativity. The set of Lagrange multipliers are introduced in the formalism, and they are corresponding to the Hamiltonian constraints which are written in terms of ‘weak equality’—the equality is satisfied if the constraints hold. Follow the low-entropy hypothesis, we postulate a modeling mechanism—a weak equality (of modeling) that holds only on the subspace of the theory space of physical models defined by some modeling constraints. By applying the modeling mechanism, we obtain a specific model of modified gravity under specific modeling conditions. We derive a novel equation of modeling from the mechanism, that describes how different gravitational models emerge. The solution of the modeling equation naturally turns out to be the model of R 2-gravity (with additional terms) if ordinary matter is negligible. We also found that this mechanism leads to two models: large-field inflation and wave-like dark matter (DM). Interestingly, the wave-like DM model is supported by the most recent observations of Einstein rings.

Path shadowing Monte Carlo

We introduce a Path Shadowing Monte Carlo method, which provides the prediction of future paths, given any generative model. At any given date, it averages future quantities over generated price paths whose past history matches, or ‘shadows’, the actual (observed) history. We test our approach using paths generated from a maximum entropy model of financial prices, based on a recently proposed multi-scale analogue of the standard skewness and kurtosis called ‘Scattering Spectra’. This model promotes the diversity of generated paths while reproducing main statistical properties of financial prices, including stylized facts such as volatility roughness. Our method yields state-of-the-art predictions for future realized volatility and allows one to determine conditional option smiles for the S&P500 that outperform both the most recent low-parametric models and the option market itself. The code is available at https://github.com/RudyMorel/shadowing (This work is supported by the PRAIRIE 3IA Institute of the French ANR-19-P3IA-0001 program and the ENS-CFM models and data science chair.).

Visibility-informed mapping of potential firefighter lookout locations using maximum entropy modelling

Visibility-informed mapping of potential firefighter lookout locations using maximum entropy modelling

Predicting Polygonum capitatum distribution in China across climate scenarios using MaxEnt modeling

Climate change affects the geographical distribution of species. Predicting the future potential areas suitable for certain species is of great significance for understanding their distribution characteristics and exerting their value. Based on the data of 276 effective distribution points of Polygonum capitatum and 20 ecological factors, the maximum entropy (MaxEnt) model and the ArcGIS software were employed to predict the areas suitable for P. capitatum growth, and the main environmental factors affecting the geographical distribution of this species were explored. Under the current climatic conditions, the areas highly suitable for P. capitatum are mainly distributed in southwestern China, with a small number of sites in coastal areas and most sites in Guizhou Province. Under different climate scenarios, the suitable areas were reduced to varying degrees. The dominant environmental variables affecting the distribution of P. capitatum were precipitation in the driest month, annual precipitation, and elevation, with a cumulative contribution rate of 84.1%. Against the background of a changing climate, the areas suitable for P. capitatum in China will be widely distributed in the southwestern region, with Guizhou Province and Yunnan Province as the main distribution areas; some sites will also be distributed throughout the southwest of Tibet Autonomous Region, the south of Sichuan Province, the north of Guangxi Autonomous Region, and the coastal area of Fujian Province. Optimal conditions for P. capitatum include a dry month precipitation range of 13.4 to 207.3 mm, elevations from 460.3 to 7214.3 m, and annual precipitation between 810 and 1575 mm. Given these insights, we recommend enhanced conservation efforts in current prime habitats and exploring potential cultivation in newly identified suitable regions to ensure the species’ preservation and sustainable use.

Problems with combining modelling and social science approaches to understand artisanal fisheries bycatch

AbstractAimArtisanal fisheries account for 40% of the world's fisheries catch, yet its environmental impacts remain poorly understood. This is especially the case in developing countries. In this study, we sought to integrate Local Fisher's Knowledge with distribution modelling to estimate the annual bycatch of Titicaca Grebe (Rollandia microptera), an endangered endemic bird from Lake Titicaca whose main anthropogenic threat is bycatch.LocationLake Titicaca, Peru and Bolivia.MethodsWe conducted transect and point counts of fishing nets in March–September 2022 and conducted interviews with fishers across the Lake Titicaca region. Using bathymetry, distance from shore, distance from a settlement, distance from the protected area, presence/absence of aquaculture, distance from aquaculture, and wetland cover, we constructed a distribution model of fisheries using maximum entropy modelling. We conducted interviews with fishers asking about the frequency of grebe bycatch and conducted short‐term monitoring at various sites while conducting transect points for dead grebes.ResultsWe estimate 3270 km2 of the surface area of Lake Titicaca is used for fishing, which amounts to 39.40% of the lake's surface area. The area under the curve (AUC) of the distribution model was 0.89 and the True Skill Statistic was 0.67, which suggests maximum entropy modelling can model fisheries occurrence. The results of our interviews suggested a biologically implausible large number of grebes caught as bycatch annually. The cultural context of the interviews, including potential influences of non‐response and social‐desirability bias, being with fishers who often view the Titicaca Grebe as a nuisance species, might have caused over‐reporting of bycatch and hence led to these implausible figures.Main ConclusionsIt is possible to map fisheries using distribution models as one might with species. However, obtaining accurate measures of fisheries bycatch through interviews is more difficult, due to cultural factors which affect the accuracy in fisher's responses. While we hope that this method provides a low‐cost alternative to monitoring, it is not a suitable replacement for it.

Predicting the potential global distribution of Ixodes pacificus under climate change.

In order to predict the global potential distribution range of Ixodes pacificus (I. pacificus) under different climate scenario models in the future, analyze the major climate factors affecting its distribution, and provide references for the transformation of passive vector surveillance into active vector surveillance, the maximum entropy model (MaxEnt) was used in this study to estimate the global potential distribution range of I. pacificus under historical climate scenarios and different future climate scenarios. The global distribution data of I. pacificus were screened by ENMtools and ArcGIS 10.8 software, and a total of 563 distribution data of I. pacificus were obtained. Maxent 3.4.1 and R 4.0.3 were used to screen climate variables according to the contribution rate of environmental variables, knife cutting method and correlation analysis of variables. R 4.0.3 was used to calculate model regulation frequency doubling and feature combination to adjust MaxEnt parameters. The model results showed that the training omission rate was in good agreement with the theoretical omission rate, and the area under ROC curve (AUC) value of the model was 0.978. Among the included environmental variables, the Tmin2 (minimum temperature in February) and Prec1 (precipitation in January) contributed the most to the model, providing more effective information for the distribution of I. pacificus. MaxEnt model revealed that the distribution range of I. pacificus was dynamically changing. The main potential suitable areas are distributed in North America, South America, Europe, Oceania and Asia. Under the future climate scenario model, the potential suitable areas show a downward trend, but the countries and regions ieeeeeeenvolved in the suitable areas do not change much. Therefore, the invasion risk of the potential suitable area of I. pacificus should be paid attention to.

An integrated urban flooding risk analysis framework leveraging machine learning models: A case study of Xi'an, China

As global climate change intensifies and urbanization progresses, the frequency and extent of urban flooding have been increasing. This trend poses a significant threat to the safety and property of urban residents and severely impedes the advancement of sustainable development goals (SDGs). This study proposed an integrated urban flooding risk analysis framework to provide an essential basis for developing effective adaptation and mitigation measures. The kernel density estimation was adopted to analyze the distribution characteristics of urban flooding hotspots. The maximum entropy (MaxEnt) model was adopted for urban flooding risk probability assessment based on multi-source data. This study conducted an in-depth analysis from three perspectives including spatial analysis, risk planning and model comparison. The MaxEnt model and multiple exploratory spatial data analysis methods were combined to reveal spatial distribution patterns at grid cell scale. The MaxEnt and Zonation models were coupled to generate an urban flooding risk map. Further, six traditional machine learning models were compared with the MaxEnt model. The results indicate that rapid urbanization and intense human activities have significantly increased the number of buildings, leading to loss of previously existing blue and green spaces. These urban characteristics increase the vulnerability of cities to extreme rainfall. Compared to traditional machine learning models, the MaxEnt model has significant advantages. In the field of urban flooding risk, the MaxEnt model demonstrates significant potential for practical application. In summary, the framework empowers urban managers to comprehensively evaluate flooding risk in planning, supporting urban design and risk prevention.

The Potential Habitat Response of Cyclobalanopsis gilva to Climate Change.

Cyclobalanopsis gilva, a valuable timber species in China, holds significant importance for understanding the constraints imposed by climate change on the dynamic geographic distribution of tree species. This study utilized the MaxEnt maximum entropy model to reconstruct the migratory dynamics of C. gilva geographical distribution since the Last Glacial Maximum. The objective was to comprehend the restrictive mechanisms of environmental factors on its potential geographical distribution, aiming to provide insights for mid-to-long-term afforestation planning of C. gilva. The optimized MaxEnt model exhibited a significantly high predictive accuracy, with an average AUC value of 0.949 ± 0.004 for the modern suitable habitat model of C. gilva. The total suitable habitat area for C. gilva in contemporary times was 143.05 × 104 km2, with a highly suitable habitat area of 3.14 × 104 km2. The contemporary suitable habitat was primarily located in the southeastern regions of China, while the highly suitable habitat was concentrated in eastern Fujian and central-eastern Taiwan. Bioclimatic variables such as mean diurnal range (Bio2), min temperature of coldest month (Bio6), precipitation of driest quarter (Bio17), and precipitation of driest month (Bio14) predominantly influenced the modern geographic distribution pattern of C. gilva, with temperature factors playing a leading role. With global climate warming, there is a risk of fragmentation or even loss of suitable habitat for C. gilva by 2050 and 2090. Therefore, the findings of this study can significantly contribute to initiating a habitat conservation campaign for this species.

Desert oasis vegetation information extraction by PLANET and unmanned aerial vehicle image fusion

Sparse vegetation is a key factor in maintaining the health and sustainability of oasis ecosystems under extreme drought conditions. Combining the advantages of both satellites and drones, using image processing and machine learning technology, it can efficiently and accurately extract information on the vegetation of desert oases, providing a scientific basis for ecological monitoring and management. Therefore, in this study, the Dariyabui natural oasis in the hinterland of the Taklamakan Desert, China, was used as the study area to fuse PLANET and UAV images to screen 19 feature factors from different data sources. The machine learning binary classification model was evaluated based on the overall accuracy (OA, %), kappa, balanced accuracy (BA, %), F1_score (F1, %), and Area Under Curve (AUC) values. The results were validated using aerial imaging data from different months. The results show that (1) the primary feature factors for extracting sparse ground vegetation information in a desert oasis include the soil-adjusted vegetation index (SAVI); modified soil-adjusted vegetation index (MSAVI); optimized soil-adjusted vegetation index (OSAVI); and the mean, entropy, contrast, homogeneity, and digital surface models (DSM). (2) The evaluation indices of the fused images in the vegetation information extraction were better than those of the original satellite images; the difference in the extraction effect with the original UAV images was not significant. (3) The random forest (RF) algorithm achieved the best classification accuracy in extracting sparse vegetation information (Kappa = 0.87, OA = 94.12 %, BA = 93.53 %, F1 = 92.13 %, and AUC = 0.967).

Habitat for Coilia nasus in southern Zhejiang Province, China, based on a maximum entropy model

As an important fishery resource and endangered species, studying the habitat of Coilia nasus (C. nasus) is highly significant. This study used fishery survey data from southern Zhejiang coastal waters from 2016 to 2020, employing a maximum entropy model (MaxEnt) to map the habitat distribution of C. nasus. Model performance was evaluated using two metrics: the area under the curve (AUC) of the receiver operating characteristic curve for the training and test sets and true skill statistics (TSS). This study aimed to predict the habitat distribution of C. nasus and explore how environmental variables influence habitat suitability. The results indicated that the models for each season had strong predictive performance, with AUC values above 0.8 and TSS values exceeding 0.6, indicating that they could accurately predict the presence of C. nasus. In the study area, C. nasus was primarily found in brackish or marine waters near bays and coastal islands. Among all environmental factors, salinity (S) and bottom temperature (BOT) had the highest correlations with habitat distribution, although these correlations varied across seasons. The findings of this study provide empirical evidence and a reference for the conservation and management of C. nasus and for the designation of its protected areas.

Estimating velocity distribution and flood discharge at river bridges using entropy theory – insights from computational fluid dynamics flow fields

Abstract. Estimating the flow velocity and discharge in rivers is of particular interest for monitoring, modeling, and research purposes. Instruments for measuring water level and surface velocity are generally mounted on bridge decks, and this poses a challenge because the bridge structure, with piers and abutments, can perturb the flow field. The current research aims to investigate the applicability of entropy theory to estimate the velocity distribution and the discharge in the vicinity of river bridges. For this purpose, a computational fluid dynamics (CFD) model is used to obtain three-dimensional flow fields along a stretch of the Paglia River (central Italy), where a historical multi-arch bridge strongly affects flood flows. The input data for the entropy model include the cross-sectional bathymetry and the surface velocity provided by the numerical simulations. A total of 12 samples, including three different flow conditions at four cross-sections, one upstream and three downstream of the bridge, are considered. It is found that the entropy model can be reliably applied upstream of the bridge, also when forced with a single (i.e., the maximum) value of the surface velocity, with errors on total discharge below 13 % in the considered case. By contrast, downstream of the bridge, the wakes generated by the bridge piers strongly affect the velocity distribution, both in the spanwise and in the vertical directions and for very long distances. Here, notwithstanding the complex and multimodal spanwise distribution of flow velocity, the entropy model estimates the discharge with error lower than 8 % if forced with the river-wide distribution of the surface velocity. The present study has important implications for the optimal positioning of sensors and suggests the potential of using CFD modeling and entropy theory jointly to foster greater knowledge of river systems.

Effects of environment and human activities on rice planting suitability based on MaxEnt model.

Rice is one of the major food crops, and the study of suitable planting areas for rice plays an important role in improving rice yield and optimizing the production layout. This study used Maximum Entropy (MaxEnt) model to simulate and predict the distribution of suitable rice planting areas in China from 1981 to 2020 by combining the climate, soil, and human activities, analyzed the spatial and temporal changes of suitable rice planting areas in China, and determined the main factors affecting rice planting suitability. The results indicated that the main factors influencing the distribution of suitable planting areas for rice in China were gross domestic product (GDP), population density (Pop), and annual sunshine duration (Sun), with human activities playing a dominant role. The high suitable planting areas of rice were mainly distributed in Hubei, Hunan, Jiangxi, Anhui, Guangdong, southeastern Sichuan and western Guizhou. The total suitable planting areas for rice were 346.00 × 104 km2, 345.66 × 104 km2, 347.01 × 104 km2, and 355.57 × 104 km2 from 1981 to 1990, 1991 to 2000, 2001 to 2010 and 2011 to 2020, respectively. With the passage of time, the area of unsuitable areas for rice gradually decreased, and the area of medium suitable areas increased, with large changes in the area of high- and low-suitable areas. Moreover, due to the transfer of a large number of rural laborers to the cities in recent years, the tension between people and land caused by the population explosion has led to the increasing impact of Pop on rice suitable areas and the relatively weakening of the impact of GDP on rice production interventions. The results can be used to provide scientific evidence for the management of rice cultivation and food production safety, with a view to reducing the impacts of climate change on agricultural production in the context of global climate change.

Invasive alien plants in the Qinghai-Tibetan Plateau (China): Current state and future predictions

Invasive alien plants have received increasing attention due to their worldwide environmental impacts, ecological concerns, and significant economic consequences. The Qinghai-Tibetan Plateau is an important global biodiversity hotspot hosting rich biodiversity and serving as a crucial component of the southwest ecological security barrier, with global conservation value. However, in recent years, invasive plants have been detrimental to the conservation of biodiversity and native ecosystems on the Qinghai-Tibetan Plateau. Here, 197 invasive plant species were recognized from the Qinghai-Tibetan Plateau, encompassing 133 genera across 47 families. We used the maximum entropy model (MaxEnt) method to predict the potential distribution of invasive alien plants on the Qinghai-Tibetan Plateau. Furthermore, potential geographic distribution regions suitable for 43 invasive alien plant species were found to be primarily concentrated in central and eastern Qinghai, southern and eastern Tibet, southwest Xinjiang, and the Qinghai-Tibetan Plateau part of Sichuan, Yunnan, and Gansu in China. Notably, the current potential distribution area for all 43 invasive alien plant species accounts for 29.92 % of the total area, with an increase of 32.72–59.39 % projected by 2050. Moreover, risk screening results confirm the high risk of invasion of most alien plants on the Qinghai-Tibetan Plateau. This study provides a foundation for assessing the risks associated with introducing and managing invasive alien plants in the region. It also offers scientific support for conserving the Qinghai-Tibetan Plateau’s unique biodiversity, ecological originality, and sustainable development during economic growth.

Efficient First-Order Algorithms for Large-Scale, Non-Smooth Maximum Entropy Models with Application to Wildfire Science.

Maximum entropy (MaxEnt) models are a class of statistical models that use the maximum entropy principle to estimate probability distributions from data. Due to the size of modern data sets, MaxEnt models need efficient optimization algorithms to scale well for big data applications. State-of-the-art algorithms for MaxEnt models, however, were not originally designed to handle big data sets; these algorithms either rely on technical devices that may yield unreliable numerical results, scale poorly, or require smoothness assumptions that many practical MaxEnt models lack. In this paper, we present novel optimization algorithms that overcome the shortcomings of state-of-the-art algorithms for training large-scale, non-smooth MaxEnt models. Our proposed first-order algorithms leverage the Kullback-Leibler divergence to train large-scale and non-smooth MaxEnt models efficiently. For MaxEnt models with discrete probability distribution of n elements built from samples, each containing m features, the stepsize parameter estimation and iterations in our algorithms scale on the order of O(mn) operations and can be trivially parallelized. Moreover, the strong ℓ1 convexity of the Kullback-Leibler divergence allows for larger stepsize parameters, thereby speeding up the convergence rate of our algorithms. To illustrate the efficiency of our novel algorithms, we consider the problem of estimating probabilities of fire occurrences as a function of ecological features in the Western US MTBS-Interagency wildfire data set. Our numerical results show that our algorithms outperform the state of the art by one order of magnitude and yield results that agree with physical models of wildfire occurrence and previous statistical analyses of wildfire drivers.

Modeling the richness and spatial distribution of the wild relatives of Iranian pears (Pyrus L.) for conservation management

The preservation of the genetic resources of crop wild relatives (CWRs) is crucial for food production systems and is considered a vital measure for global agricultural health and food security. The identification of potential areas where CWRs can thrive is one of the first steps towards their conservation. In this study, we used a maximum entropy model (MaxEnt) to determine the habitat suitability of seven wild relatives of pears (Pyrus L.) for the first time. We aimed to identify high-priority areas for conservation and determine the hotspots for rich biodiversity in Iran. The study showed excellent predictive performance for all species studied (AUC value ≥ 90). The soil depth, solar radiation, minimum temperature of the coldest month (Bio6), and precipitation of the wettest quarter (Bio16) were the main environmental factors that influenced the habitat suitability of all seven species, according to permutation importance. The projected maps revealed that P. elaeagnifolia had the largest suitable habitat area, while P. glabra had the lowest. The results also showed that less than 5% of the suitable habitats for these seven species were in protected areas. This research highlights the need for national preservation policies and the development of cultivation and rehabilitation strategies for these threatened species.

The influence of biophysical characteristics on elephant space use in an African savanna

Understanding the influence of biophysical variables on spatial distribution and seasonal locations of elephants is valuable inter alia in understanding their movement patterns, and the carrying capacity to optimise the conservation of their habitats. Furthermore, understanding elephant distribution in time and space is beneficial for predicting locations for tourist viewing and determining potential areas of human–wildlife conflicts. In this study, we examined the environmental drivers of seasonal variations in elephant ranges and movement using telemetry data. Using the elephant tracking data and biophysical characteristics, the Adaptive-Local Convex Hull (a-LoCoH) and the Maximum Entropy Modelling (MaxEnt) algorithm were adopted to determine their potential locations and distribution in a protected area. The jackknife approach was used to assess the environmental factors influencing seasonal elephant movements based on collar data from ten individual elephants tracked between December 2015 and April 2018. The performance of the elephant distribution models for the two seasons was assessed using the area under the curve from the receiver operating characteristic based on a 30% test holdout data. Results showed that the core a-LoCoH home of elephants was larger in the dry than wet season (Wilcoxon signed-rank test; W = 3, Z = 2.4973, p < 0.012). However, no noticeable differences were observed (Wilcoxon signed-rank test; W = 10, Z = 1.7838, p < 0.07) between the dry and wet season for the total (90%) a-LoCoH ranges. The MaxEnt modelling results showed that distance to the artificial water holes was the most influential variable in predicting the potential elephant spatial distribution in both wet and dry seasons, while the normalised difference vegetation index, a proxy for forage availability, was more influential in the dry season in predicting elephant distribution. The study concludes that seasonal elephant space use is mostly influenced by water availability in both seasons, while forage availability and variability influence elephant distribution more strongly in the dry than wet season. The study provides valuable insight into the biophysical characteristics of elephant home ranges. This is useful for the sustainable management of elephants and their habitats.

Landslide susceptibility prediction based on landform predisposing indexes − An example from the Beiluo River Basin

The Beiluo River basin, which flows through the central part of the Loess Plateau, has experienced intense soil erosion and significant geomorphic change, which has provided favorable conditions for the occurrence of a large number of landslides. Landform indexes, which can express geomorphologic development state and internal rules, can transfer the development process information of surface morphology into the evaluation of landslide susceptibility, and help get more accurate landslide susceptibility prediction results. Taking the Beiluo River Basin as an example, a landslide susceptibility prediction model based on landform index is proposed by comparing the importance of landform index. In order to improve the accuracy of LSP, 10 kinds of general predictors indexes, 5 kinds of landform predisposing indexes and 1821 landslide points were compiled, and the geographic information system of Beiluo River Basin was constructed. Through the correlation test and CF model, the environmental indexes were evaluated to obtain the sensitive index results, and the combination of different environmental predictors indexes were classified according to the sensitive index results. Based on the combined classification results, the Max Entropy (MaxEnt) model was used to evaluate the Landslide susceptibility prediction (LSP), while the calculated results were evaluated and compared using the receiver operating characteristic (ROC) curve and landslide density. The results show that the vertical erosion factor, elevation, rainfall, horizontal erosion factor, slope angle and NDVI play a key role in controlling the spatial distribution of landslides in the study area. At the same time, the accuracy of landslide susceptibility is compared by AUC value. According to the calculation results, the Group5 (AUC = 0.803) with reasonable terrain index performs better in the training and test stages, and the relative accuracy is improved by 6.22 % compared with the non-introduction of terrain index and the omission rate difference is the best (omission rate difference = 0.0005), indicating that the introduction of landform index can effectively improve the landslide susceptibility prediction. The distribution of different sensitive areas was observed. The high sensitive areas and very high sensitive areas are mainly distributed in the southern Luochuan loess tableland and the northern Wuqi loess hilly area. The research results provide a scientific basis for landslide susceptibility prediction with rational introduction of landform indexes and regional infrastructure construction.

Multirate Progressive Entropy Model for Learned Image Compression

Multirate Progressive Entropy Model for Learned Image Compression

Determination of forest fire risk with respect to Marchalina hellenica potential distribution to protect pine honey production sites in Turkey.

Turkey is the leading producer of pine honey worldwide, accounting for 90% of global production, largely due to the presence of Marchalina hellenica populations. However, in recent years, devastating forest fires have caused substantial damage to Pinus brutia forests and M. hellenica populations, leading to a dramatic decline in pine honey production areas. The urgency for early intervention procedures against forest fires and relocation of M. hellenica populations to other P. brutia forests has become apparent. A comprehensive assessment of 25 criteria was conducted to investigate the thresholds and behaviors of each criterion, which play a vital role in the distribution of M. hellenica, using the maximum entropy model (MaxEnt). To evaluate the impact of forest fires on the distribution of M. hellenica, the potential locations of pine honey production areas were determined for 2022. Furthermore, the susceptibility of forest fires was modeled for all pine honey production months. The findings revealed that forest fires have destroyed 384.9 km2 (12.8% of the total pine honey production area), predominantly in August and September, with the most severe damage in Marmaris (156 km2) and significant impacts in Ula, Köyceğiz, and Milas. The analysis facilitates the estimation of new areas suitable for M. hellenica and pine honey production while providing valuable insights into strategies for mitigating forest fires and formulating proactive protection protocols.

The importance of geography in forecasting future fire patterns under climate change

An increasing amount of California's landscape has burned in wildfires in recent decades, in conjunction with increasing temperatures and vapor pressure deficit due to climate change. As the wildland-urban interface expands, more people are exposed to and harmed by these extensive wildfires, which are also eroding the resilience of terrestrial ecosystems. With future wildfire activity expected to increase, there is an urgent demand for solutions that sustain healthy ecosystems and wildfire-resilient human communities. Those who manage disaster response, landscapes, and biodiversity rely on mapped projections of how fire activity may respond to climate change and other human factors. California wildfire is complex, however, and climate-fire relationships vary across the state. Given known geographical variability in drivers of fire activity, we asked whether the geographical extent of fire models used to create these projections may alter the interpretation of predictions. We compared models of fire occurrence spanning the entire state of California to models developed for individual ecoregions and then projected end-of-century future fire patterns under climate change scenarios. We trained a Maximum Entropy model with fire records and hydroclimatological variables from recent decades (1981 to 2010) as well as topographic and human infrastructure predictors. Results showed substantial variation in predictors of fire probability and mapped future projections of fire depending upon geographical extents of model boundaries. Only the ecoregion models, accounting for the unique patterns of vegetation, climate, and human infrastructure, projected an increase in fire in most forested regions of the state, congruent with predictions from other studies.