Next Article in Journal
Longitudinal Ceiling Temperature Profile in an Inclined Channel Induced by a Wall-Attached Fire
Previous Article in Journal
Volunteer Motivation in Firefighting Organisations: A Case of the Slovenian Firefighters Association
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Fire
Volume 8
Issue 6
10.3390/fire8060221
fire-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Hydroclimatic and Land Use Drivers of Wildfire Risk in the Colombian Caribbean

by
Yiniva Camargo Caicedo
1,2,*,
Sindy Bolaño-Diaz
1,2,
Geraldine M. Pomares-Meza
1,2,
Manuel Pérez-Pérez
1,2,
Tionhonkélé Drissa Soro
3,4,
Tomás R. Bolaño-Ortiz
5 and
Andrés M. Vélez-Pereira
6
1
Programa de Ingeniería Ambiental y Sanitaria, Facultad de Ingeniería, Universidad del Magdalena, Calle 29H3 No. 22-01, Santa Marta 470004, Colombia
2
Grupo de Investigación en Modelación de Sistemas Ambientales (GIMSA), Facultad de Ingeniería, Universidad del Magdalena, Calle 29H3 No. 22-01, Santa Marta 470004, Colombia
3
UFR Biosciences, Université Félix Houphouët-Boigny, Abidjan 22 BP 582, Côte d’Ivoire
4
Laboratoire des Milieux Naturels et Conservation de la Biodiversité, Université Félix Houphouët-Boigny, Abidjan 22 BP 582, Côte d’Ivoire
5
School of Natural Resources Engineering, Department of Agricultural Science, Universidad Católica del Maule, Curicó 3466706, Chile
6
Departamento de Ingeniería Mecánica, Facultad de Ingeniería, Universidad de Tarapacá, Arica 1000000, Chile
*
Author to whom correspondence should be addressed.
Fire 2025, 8(6), 221; https://doi.org/10.3390/fire8060221
Submission received: 26 April 2025 / Revised: 27 May 2025 / Accepted: 29 May 2025 / Published: 31 May 2025
(This article belongs to the Section Fire Science Models, Remote Sensing, and Data)
Browse Figures
Versions Notes

Abstract

Fire-driven land cover change has generated a paradox: while habitat fragmentation from agriculture, livestock, and urban expansion has reduced natural fire occurrences, human-induced ignitions have increased wildfire frequency and intensity. In northern Colombia’s Magdalena Department, most of the territory faces moderate to high wildfire risk, especially during recurrent dry seasons and periods of below-average precipitation. However, knowledge of wildfire spatiotemporal occurrence and its drivers remains scarce. This work addresses this gap by identifying fire-prone zones and analyzing the influence of climate and vegetation in the Magdalena Department. Fire-prone zones were identified using the Getis–Ord Gi* method over fire density and burned area data from 2001 to 2023; then, they were analyzed with seasonally aggregated hydroclimatic indices via logistic regression to quantify their influence on wildfires. Vegetation susceptibility was assessed using geostatistics, obtaining land cover types most affected by fire and their degree of fragmentation. Fire-prone zones in the Magdalena Department covered ~744.35 km2 (3.21%), with a weak but significant (τ = 0.20, p < 0.01) degree of coincidence between classification based on fire density, as pre-fire variable, and burned area, as a post-fire variable. Temporally, fire probability increased during the dry season, driven by short-lagged precursors such as Dry Spell Length and precipitation from the preceding wet season. Fire-prone zones were dominated by pastures (62.39%), grasslands and shrublands (19.61%) and forests (15.74%), and exhibited larger, more complex high-risk patches, despite similar spatial connectedness with non-fire-prone zones. These findings enhance wildfire vulnerability understanding, contributing to risk-based territorial planning.

1. Introduction

Wildfires have significant yet varying impacts across global ecosystems [1]. This spatiotemporal variability is particularly pronounced in tropical regions, where human activities have become the primary source of ignitions in forests, savannas, and agricultural landscapes, fundamentally altering natural fire regimes and ecosystem services [2,3,4]. These regime shifts are shaped by local precipitation and temperature response [5]. In South America, while overall wildfire activity has declined over the past two decades [6], extreme events driven by climatic variability remain oscillating [7]. For instance, in 2020, more than 50 million hectares were burned [8].
Tropical savannas represent the most fire-prone biome in the region, accounting for 43% to 55% of the total area affected over the last decade [8]. In this biome, fire regimes are characterized by frequent occurrences during dry seasons, creating spatial heterogeneity in vegetation structure [9,10]. However, increased grazing activities and urban and agricultural expansion have reduced fire frequency in areas with high anthropogenic pressure, particularly through deforestation practices, promoting woody species recruitment and landscape fragmentation [11,12,13,14]. Consequently, wildfire occurrence, intensity, and spread are altered, impacting communities, public health, and carbon emissions [15,16,17], while reflecting complex interactions between human activity, ecological systems, and vegetative productivity [18,19].
The distribution of wildfires is closely linked to climate change, particularly through variations in the El Niño–Southern Oscillation (ENSO) phenomenon, which increases temperatures and reduces precipitation in various regions [20,21]. This impact is particularly noticeable in boreal and temperate climate zones, where the risk of fires has increased [22,23,24]. Several studies have demonstrated that land cover is a key determinant of wildfire susceptibility, with a higher frequency observed in poorly managed areas dominated by grasslands and shrublands [25,26]. Population growth and associated economic activities are key factors in determining fire risk [27,28]. However, landscape fragmentation due to agricultural development can have contradictory effects: while modifying landscape dynamics, it can also reduce vegetative continuity, consequently reducing fire intensity and severity, as well as patch size [29].
In Colombia, fires have been studied from different perspectives. Bolaño-Diaz et al. [30] analyzed the distribution of active wildfires between 2000 and 2020. Significant differences between regions were found. The Orinoco region, followed by the Caribbean and Andean regions, identified high-fire density prone zones, which positively correlated with temperature. The results agree with Guevara [31], in the Orinoquia basin between Colombia and Venezuela, reported as more affected by fires, especially during the primary dry season (December-March) [32]. This distribution is linked to variations in the country’s hydrological cycle, with the western regions (Andes and Pacific) having a more intense cycle than the eastern regions (Caribbean, Orinoco and Amazon), due to the latter having less influence from ENSO, yet more direct influence from the Amazon basin, the Caribbean Sea and the Atlantic Ocean [33,34].
Fires in the Caribbean are driven by climatic, anthropogenic, and vegetation-related factors that shape their behavior at the regional scale. For example, the quasi-decadal oscillation of the Tropical North Atlantic (TNA) has been linked to fire events and variations in precipitation and temperature from 2003 to 2015 [35]. Recent evidence suggests an increased frequency of below-average annual rainfall in the Magdalena Department, with a pronounced negative trend ranging from −32 to −100 mm/decade, highlighting growing drought vulnerability, especially in the Sierra Nevada de Santa Marta (SNSM) strategic region and the center of the Department [36]. These drying patterns heighten wildfire risk, as demonstrated by the work carried out by Cabrera et al. [37] in the SNSM, where events were recorded in the southern and northeastern parts during the dry season, showing that factors such as the decrease in soil temperature reduce the occurrence of fire events, as well as the adoption of mitigation measures. Furthermore, the hazard analysis conducted by Gutiérrez et al. [38] showed that 84.61% of the area has a fire susceptibility between moderate and high, based on the theoretical evaluation of vegetation and climatic factors. However, no studies in the Department evaluate the spatial and temporal pattern of fires based on the number of fires and area burned.
The Magdalena Department’s climatic conditions make it particularly prone to fires, favoring their main effects. In this region, there is a geographic gap in the study of fires, their regime, and the identification of prone zones. No research has been conducted to analyze fire occurrence in this department, its relationship with hydroclimatic variables, and the main land covers. The present work aimed to (1) investigate the spatiotemporal relationship between pre-fire (thermal anomalies and fire density) and post-fire variables (burned area), (2) evaluate the influence of hydroclimatic variables on fire occurrence, and (3) identify fire-prone zones and the most susceptible land cover types in Magdalena Department.

2. Materials and Methods

2.1. Study Area

The study region is the Magdalena Department, located in Colombia’s northern region between latitudes 8°56′ and 11°36′ N and longitudes 73°32′ and 74°56′ W (Figure 1). This department has an approximate area of 23,188 km2, bordered by the Caribbean Sea to the north [39].
The Magdalena Department is located within the natural regions of the SNSM and the Bajo Magdalena Valley. The extension and geographical disposition give different biomes, such as tropical forests, wetlands, dry forests, rainforests, and coastal areas [40], playing a fundamental role in conserving biodiversity and regulating the global climate. Colombia has two main precipitation regimes: a monomodal regime present in regions such as Orinoco, Amazonia, and the Caribbean, and a bimodal regime found in the Andean region and the mountainous zones [41]. The Magdalena Department shows a bimodal pattern characterized by two dry and two rainy seasons in the eastern, central, and western zones and a monomodal pattern—one dry and one rainy season—in the northern and southern parts [39] (Figure S1.1), mainly influenced by the geomorphology of the SNSM and the interaction with climatic systems such as ENSO and the North Atlantic Oscillation (NAO) [36,42,43].
The study area’s extension comprises various biomes and strategic ecosystems, including semi-humid, dry tropical, and humid tropical forests, mangrove ecosystems, and coral reefs. Other biomes are associated with savannahs, mainly grasslands that cover a large part of the area. Water bodies are mostly wetlands, the delta of the Magdalena River, and protected areas such as the Ciénaga Grande de Santa Marta [44,45]. The SNSM contains rivers that originate in mountain peaks and converge with ecosystems such as paramo in the highlands, as well as vegetation including sub-Andean forests, dry tropical forests, grasslands, shrublands, and pastures in the middle and lowlands, all of which have been greatly affected by human activity [46,47].

2.2. Database

Wildfire data in the study area were obtained from 2001 to 2023 from the Moderate Resolution Imaging Spectroradiometer (MODIS) sensor, accessed through NASA’s Fire Information for Resource Management System (FIRMS) (https://firms.modaps.eosdis.nasa.gov/download/, accessed on 1 August 2024). This sensor allows for hot spots and burned areas, with many records. This provides an approximation of the occurrence of fires in areas with recurrent activity through the MCD14DL (Thermal Anomalies) and MCD64A1.061 (Burned Area) products [48]. Thermal anomaly data were obtained at 1 km spatial resolution with daily temporal coverage, while burned area data had 500 m spatial resolution and monthly temporal resolution. Thermal anomalies data were aggregated monthly by summing daily detections to match the temporal resolution of the burned area product, comparing the possible locations of monthly fire outbreaks, and corroborating the contributions of each product [30,49]. To avoid confusion with water reflections, thermal anomalies with confidence that were more significant than 80% were established as fire detection and selected to ensure the reliability of the information according to the FP_confidence ranges included in the product [50].
The meteorological information was obtained from the pluviometry land stations network operated by the Institute of Hydrology, Meteorology and Environmental Studies (also known by its acronym in Spanish, IDEAM) through http://dhime.ideam.gov.co/webgis/home/ (accessed on 8 October 2024). The variables selected for calculating hydroclimatic indices included total precipitation and dry days. Station filtering was conducted based on the criteria established by Pomares et al. [36]. Although various stations in the Department met these criteria, only those within zones later identified as fire-prone, determined through fire density analysis in Section 2.3, were included. To assign fire density weights to each station, a GIS layer of Voronoi polygons was created, delineating the influence area of each station relative to specific fire-prone zones [51,52,53]. Outlier treatment and handling of missing data involved applying the homogenization test proposed by Alexandersson [54] and adapted in recent hydroclimatic trend studies [55,56,57,58].
The information used for the analysis of fire-predominant land covers was obtained from the web service “Colombia en mapas” of the Geographic Institute Agustín Codazzi (also known by its acronym in Spanish, IGAC) (https://www.colombiaenmapas.gov.co, accessed on 6 February 2025).

2.3. Analysis of Fire Variables

2.3.1. Calculation of Fire Density and Burned Area

The fire events are categorized by the number of fires and fires per area (pre-fire) and burned area (post-fire)—reported in function of a 1000-hectare (100 km2) grid [59]. Fire density was calculated by averaging the total number of fires counted in each 1000-hectare (Ha) grid per year over the entire study period. Additionally, fire density was evaluated seasonally throughout the study period and grouped according to the Department’s typical wildfire seasonality: December through March (DJFM) representing the primary dry season with decreased precipitation and increased temperatures; April through July (AMJJ) marking the transition between dry and rainy seasons; and August through November (ASON) comprising the rainy season [35,37,60,61].

2.3.2. Spatial Autocorrelation and Identification of Fire-Prone Zones

Localized Getis–Ord Gi* index is a measure of spatial valuable autocorrelation for identifying zones with statistically significant high or low values. This index has been widely validated in fire research as an effective method for determining whether fires are clustered or randomly dispersed across landscape units and delimitating areas of intense fire activity. Its application to MODIS fire data has consistently supported the classification of fire-prone zones, demonstrating strong suitability for analyzing spatial patterns concerning underlying environmental factors [62,63,64,65].
The Gi* statistic measures the association between a given cell and its neighbors within a specified distance [66]. Here, the neighborhood was defined by the diagonal length of each 10 × 10 km cell, considering both cardinal and diagonal neighbors [67]. Each Gi* statistic represents a z-score, allowing the significance assessment at 1% (Extremely hot/cold), 5% (Hot/Cold), and 10% (Moderately hot/cold) significance levels to classify fire density cells [68]. Getis–Ord Gi* index was applied to both fire density and burned area, and the spatial coincidence of the resulting fire-prone zones was assessed using Pearson’s Chi-squared (χ2) and Kendall’s rank correlation (τ) tests [69]. Further explanation regarding the mathematical formulation of Gi* can be found in Section 2.2 of Supplementary Materials 1.

2.4. Fire Occurrence According to Hydroclimatic Variables

Hydroclimatic indices are crucial for assessing fire risk, as wet conditions promote biomass accumulation, while prolonged dry spells heighten the risk of combustion [7]. Precipitation intensity (INT) was calculated as the average rainfall intensity, derived from the total seasonal precipitation divided by the number of rainy days. Likewise, the Dry Spell Length (DSL) was determined by the ratio of total seasonal dry days (defined as days where rainfall >0.1 mm) and dry periods (consecutive dry days) [33].
A binary logistic regression model was employed to analyze the relationship between seasonal fire density and these indices. The quantile method was used for discretizing fire density, following Zhang et al. [70]’s approach for continuous variables with a similar negative asymmetric distribution. The optimal threshold for fire density was identified by maximizing the explained variance (Q) across potential thresholds, categorizing fire density into two groups: above or below the threshold [71].
The independence of predictor variables was assessed using the Variance Inflation Factor (VIF), with values above five indicating potential multicollinearity [72]. Therefore, the long-term seasonal effects of predictors were evaluated, focusing on the influence of droughts and extreme precipitation as precursors to fire risk [73].
The model was fitted for each season using discretized fire density as the dependent variable and predictor variables at lags 0, 1, and 2. Lag 0 indicates seasonal alignment between the dependent and the predictor variable, while lags 1 and 2 refer to one and two seasons prior, respectively. For each model, predicted probabilities were computed across the range of the predictor and plotted along with 95% confidence intervals derived from the standard errors. Their performance was evaluated from the calculation of the classification metrics described in Section S2.3 of Supplementary Materials S1, based on the work by Hosmer et al. [74].

2.5. Fire Occurrence According to Predominant Covers

A map algebra analysis was conducted to recalculate fire density and burned area per 1000 Ha for each land cover type across the Department. A thematic overlay using a land cover map at a 1:25,000 scale was applied to analyze the distribution of fire occurrence relative to land cover types.
To compare theoretical and actual fire risk, land covers were reclassified based on land cover-based fire risk zoning criteria from IDEAM [75]. The predominant land cover types in the Department—agricultural areas, forests, humid zones, grasslands, shrublands, pastures, permanent and temporary crops, sparsely vegetated areas, and water bodies—were assigned risk levels following the classification by Gutiérrez et al. [38], ranging from very low (1) to very high (5). These risk classes were grouped into three categories based on fuel type, fuel duration, and fuel load, forming an average composite land cover risk classification comprising very low, moderate, and high-risk classes.
A fragmentation-based assessment identified high-risk cover structures in fire-prone zones. Fragmentation is a landscape-level process involving various quantifiable properties—such as area, shape, and aggregation—measured via fragmentation indices or landscape metrics to outline mosaic presence and complexity [76,77,78]. Area metrics assess patch size, shape metrics evaluate patch complexity, and aggregation metrics measure the spatial connectedness of patches [79]. The analysis compared fire-prone zones to adjacent non-fire-prone control zones, using equal contiguous grids and controlling for other influencing factors, such as climate. These control zones matched in surface extent and types of cover presence but differed by lacking significant fire activity over the studied period, as identified by the Getis–Ord Gi* classification.
The R (v4.4.1) package landscapemetrics (v2.2.1) [80] was used to calculate eight landscape metrics (Table 1). Due to computational limitations, the vector land cover layer was converted to a raster format with a resolution of ~37 m using QGIS (v3.40.1). Connected dot plots were then generated to compare the fire-prone and non-fire-prone zones, grouping metrics by units to preserve their original ranges. The formula metrics are found in Section S2.4 of Supplementary Materials S1.

3. Results

3.1. Fire Variable Assessment

The spatiotemporal distribution of thermal anomalies and burned areas provided key insights for the study region. Figure 2 shows the distribution of these anomalies and burned areas during the study period. A total of 6500 thermal anomalies were detected across the Magdalena Department (Figure 2a), resulting in 163,285.15 Ha of burned area, grouped into 150 distinct events, excluding overlaps, within a total area of 2,318,800 Ha (Figure 2b). The high number of anomalies relative to burned areas is due to repeated detections of individual fires over time, reflecting the sensor’s temporal resolution and the multi-day or multi-week duration of many fires.
The monthly relationship between anomalies and burned areas is also consistent, with February and March having the highest detection values for both products, exceeding 1800 Ha burned and 600 fires in 2003 and 2004. Other relevant years within the analysis are mainly found within the first decade, when significant burned areas and relatively high numbers of thermal anomalies were detected, compared to the second decade, when the burning trend decreased.
The results of the distribution of thermal anomalies in the study area indicate that typical fire seasonality influences the temporal distribution of active fires in the Magdalena Department, Colombia. Figure 3 shows an increase in fire density during DJFM, especially in February and March, where, respectively, 21.38% and 70.82% of the total number of fires observed during the study period were recorded (Figure S2.1); these months are followed by April, with a contribution of 2.74% and July with 1.93% of the total number of fires. September and October, which belong to ASON, did not record any fires during the entire study period. Regarding the spatial distribution, the highest fire density values were concentrated in the central part of the department, ranging from 0.8 to 1.5 fires/1000 ha/year.
To further understand the spatial distribution of fire occurrences, fire density values were analyzed using a spatial autocorrelation approach to identify fire-prone zones in the area studied. Figure 4 displays the spatial distribution of fire density according to each weighted cell’s Getis–Ord Gi* significance value. Extremely hot grids (p < 0.01) represented 8.90% (n = 206) of the total, while composed hot and moderately hot grids accounted for 6.27% (n = 145). Statistically significant zones with low fire density values (cold spots) mainly were in northern Magdalena, while statistically substantial hot fire density cells were in the center, comprising an area of approximately 2256.27 km2.
In the Extremely hot zones, 95 grids (46.11%) presented an annual average fire density ≥0.435 fires/Ha/year, which accounted for 10–30 total fire counts during 2001–2023, whereas the hot and moderately hot zones exhibited 10–20 total fire counts (17.28%) and 10–17 (17.19%) total fire counts during the same period. Insignificant cells with more than 10 fire counts did not exceed 2.36% of the total area classified in this category, while null fire density mainly weighed cold and moderately cold cells.
Burned area per 1000 Ha was classified through Getis Ord-Gi* as well, where extremely hot cells represented 7.91% (n = 181) of the total number of cells, while composed hot and moderately hot accounted for 3.90% (n = 90) (Figure S2.2). Pearson’s Chi-squared test was conducted to evaluate the likelihood that any observed difference between the classification of fires according to fire density and burned area per 1000 Ha arose by chance. The test yielded a statistic of 4.93 and a p-value of 0.085, suggesting weak significance at the 10% level is insufficient to conclude a significant relationship between burned area and fire hot cell counts.
However, Kendall’s τ suggests a statistically significant direct spatial relationship between fires and burned area cells (τ = 0.20, p < 0.01). Despite this significance, the value indicates a weak agreement between the two spatial distributions of cells classified as hot zones for both variables. This is evident in Figure 5, where coincident extremely hot grids covered approximately 744.35 km2, accounting for only 3.21% of the total surface (Figure 5. Nonetheless, it is meaningful that the zone exhibiting the most statistically significant high-fire occurrence (p < 0.01) is where most cells coincide, as shown in Figure 5b.

3.2. Hydroclimatic Influence on Fire Occurrence

The binary logistic regression results in Figure 6 presented valuable insights into the relationship between seasonal fire density and hydroclimatic indices. The model using DSL as a predictor revealed a highly significant association (likelihood ratio test: p < 0.001). However, the predictive specificity of 72.4% suggests that, although true negatives are correctly identified, the model failed to predict a substantial proportion of true positives (sensitivity < 50%). In contrast, the model using INT as a predictor showed only a marginally significant relationship (p = 0.0485), with overall reasonable specificity and accuracy. Nonetheless, predicted probabilities remained below 50%, indicating the model strongly favors true negatives and, due to the absence of true positives, sensitivity could not be calculated, and the F1-score was null.
The results of the VIF for multicollinearity diagnosis exhibited that both DSL and INT were not only within the range of positive independence between variables but also presented extremely low values: 1.731 and 1.352, respectively. Moreover, this factor allowed testing the binary fire density response against total seasonal precipitation, which yielded a VIF value of 2.140. The corresponding model yielded a highly significant likelihood ratio test (p < 0.001) and a predictive sensitivity of 49.1%, the highest among the tested variables.
Logistic regression models assessing seasonal fire probability (DJFM, AMJJ, ASON) with lagged hydroclimatic indices revealed distinct patterns in predictor strength. Detailed information about performance metrics can be found in Table S2.1. During the DJFM fire season, synchronous DSL (Total Precipitation) was negatively (positively) associated with fire probability, acquiring a predictive sensitivity of 90.3% with p < 0.001 (100.0% with p < 0.001). This suggests increased precipitation during the primary dry season enhances fire probability, reflected in an upward-trending log-likelihood function.
In AMJJ, DSL at lag-2 (i.e., DSL from ASON) reached significance (likelihood ratio test: p < 0.01), though predictive sensitivity was null and specificity reached 99.32%, this was consistent with the steeper probability slope but broader confidence intervals (Figure 6g), which would suggest the presence of false positives and, consequently, reduced predictive power. In contrast, INT and precipitation effects across lags were marginal and less trend-like defined, though also accompanied by relatively narrow confidence intervals, signaling overall better prediction for non-fire occurrences during this season.
For ASON, model performance weakened across all predictors and lags. INT and precipitation curves were mainly flat, with low predictive sensitivity outcomes and minimal changes in log-likelihood, indicating limited predictive influence during the primary rainy season. Finally, DSL showed significance at lag-0 and lag-1 (likelihood ratio test: p < 0.001); however, this result was primarily driven by the predominance of true negatives, which is expected during this period, characterized by short dry spells followed by intense rainfall. As a result, the sensitivity of this model remained negligible, limiting its practical predictive value for rare fire occurrences during the wet season.

3.3. Fire Occurrence According to the Predominant Covers

Figure 7a displays the spatial distribution of land covers in the Magdalena Department, where pastures dominate, covering nearly 50% of the total area. Consequently, most fires recorded during the study period occurred in this land cover (67.96%), followed by grasslands and shrublands (15.62%) and forests (1.77%). However, the total number of fires alone does not fully capture fire occurrence, as the uneven land cover distribution influences it. To address this, fire activity was further analyzed in terms of both fire density and burned area per 1000 Ha.
Figure 7b presents the temporal distribution of the annual burned area per 1000 Ha per land cover and the years in which positive anomalies were observed. While pastures accounted for the highest total burned area detected (4117.58 Ha), transient crops (0.392 fires/cover, 16.94 Ha/1000 Ha/cover), humid zones (0.388 fires/cover, 33.89 Ha/1000 Ha/cover), and water bodies (0.04 fires/cover, 4.48 Ha/1000 Ha/cover) exhibited the most yearly extreme fluctuations, showing highly abnormal values for both fire occurrence and burned area per 1000 Ha during the same year (Table S2.2).
Figure 7c provides a more detailed evaluation of fire-prone land covers through fire density analysis. Yearly contributions indicate that pastures, grasslands, shrublands, and transient crops consistently had the highest fire densities, averaging 0.168, 0.129, and 0.119 fires/1000 Ha/cover, while all other land covers had values below 0.1 fires/1000 Ha/cover.
The fragmentation-based assessment revealed distinctive patterns between fire-prone and non-fire-prone zones (Figure 8). Although grasslands and shrublands, forests, and agricultural areas were theoretically classified as high-risk covers, humid zones, grasslands, shrublands, and pastures were proportionally the most affected covers by fire across the Magdalena Department in terms of burned area per 1000 Ha. Nevertheless, the fire-prone region was mainly composed of pastures (62.39%), grasslands and shrublands (19.61%), and forests (15.74%).
The fire-prone landscape demonstrated a considerably higher aggregation of high-risk cover classes, as evidenced by the higher AI (86.35% compared to 69.36%). This aggregated pattern is further substantiated by the notably larger AREA_MN in fire-prone zones (23.36 Ha) compared to non-fire-prone zones (6.56 Ha). The spatial distribution of high-risk cover patches showed that they constitute a significant portion (36.07%) of the total landscape in fire-prone zones while representing only a minor component (7.66%) in non-fire-prone landscapes.
Nonetheless, both the CONTIG_MN and COHESION results showed minimal variation between the two landscape types, suggesting comparable levels of spatial connectedness in both zones. This disparity in patch distribution and size indicates that fire-prone zones harbor more extensive, connected areas of vegetation or land cover types susceptible to fire.
Despite these aggregation and area coverage differences, the complexity of shapes exhibited only subtle variations between the two landscape types. The mean fractal dimension index indicates marginally simpler patch shapes in non-fire-prone zones, though these differences are not substantial.

4. Discussion

The spatiotemporal wildfire analysis in the Magdalena Department identified key fire-prone zones, mainly in central and northeastern areas. These regions showed high fire density and extensive burned areas, particularly during the primary dry season (DJFM) and during the El Niño events of 2002–2004 (moderate) and 2014–2017 (very strong) [81]. Similarly, the study by Ramos Rodriguez et al. [82] in the canton of Loja, Ecuador, found a pronounced fire period from August to November, with a strong correlation between reported fires and hotspots identified by the FIRMS product. Both results showed a significant correlation between the hotspots and fire foci, with the latter occurring in lower proportions because of fire detectability limitations. However, in both cases, the spatial and temporal distributions are consistent.
Moreover, Zubieta et al. [83] evaluated the agreement between the hotspots derived from three satellite products, including MODIS, and the burned area classified by the Normalized Burned Area index. The results showed that the sensor hotspots fell within the burned areas delineated by the index, with errors of less than 1% associated with a higher confidence level in the hotspots. However, these products significantly omitted burned areas under 25 Ha due to the coarse spatial and temporal resolution, highlighting a limitation in detecting small, minor, short-duration burns. In the tropical Andes, the MODIS burned area product detected only 14.90% of the fire scars identified by the Fire Atlas from Landsat imagery in paramo ecosystems in northern South America, with detection difficulty during periods of low fire activity [84]. In the uplands of the SNSM, fire density grids were classified from Insignificant to Extremely cold, which aligned with comparable studies such as the one referenced, may provide evidence of fire activity being underestimated in topographically complex regions, underscoring the importance of using higher-resolution imagery and complementary fire data sources for effective monitoring in such ecosystems.
The identified fire-prone zones are neither topographically nor climatically complex. Even so, the logistic regression results indicate that increased precipitation and decreased DSL during the primary dry season significantly enhance fire probability. These relationships—evidenced by upward- and downward-trending log-likelihood functions, respectively—suggest that anomalous wet conditions tend to synchronize with heightened fire activity; meanwhile, fire probability diminishes during periods of anomalous dryness. This behavior can be attributed to the lack of post-fire fuel accumulation, as extended drought conditions trigger a cycle that exacerbates fuel scarcity [85] and compromises vegetation recovery [86]. These results align with previous studies that have analyzed similar dryness indices about fire outbreaks in regions like the Magdalena Department, which are characterized by a consistent wet season typically occurring approximately four to five months before fire events, leading to a significant buildup of fire fuel [87].
For instance, Alizadeh et al. [88] noted that elevated Vapor Pressure Deficit (VPD) and Fuel Moisture (FM) levels correlate with increased susceptibility of US western climate ecosystems to ignition and rapid fire spread. Notably, their findings revealed that the peak of VPD anomalies occurs about one week after a fire event begins, indicating a synchronization between fire activity and moisture availability.
Moreover, the seasonal fire density analysis with hydroclimatic indices revealed that short-lagged precursors, particularly DSL, significantly influence fire probability. The strongest model fits were observed during the DJFM and AMJJ seasons, where fire probability increased alongside DSL during the wet season—Lag-1 and Lag-2, respectively—while higher precipitation totals led to decreased fire activity.
Qu et al. [89] categorized hydroclimatic indices as top-down fire precursors and analyzed them alongside variables such as maximum temperature to evaluate their predictive power. The results exhibited that the time lags correlated with top-down precursors are generally shorter, approximately three to five months, compared to bottom-up precursors, associated with fuel characteristics exceeding eight months. These outcomes are consistent with those presented above and can be attributed to the greater seasonal variability in fire weather conditions, compared to sub-annual variability [90].
In regions with a pronounced and prolonged wet season, where fuel availability is a non-limiting factor, top-down precursors gain greater influence over vegetation abundance, particularly in zones where fine fuels, which dry rapidly under hot conditions, predominate [89,91]. These results support the development of real-time monitoring software based on fire weather-driven indices in the Magdalena Department. Di Giuseppe et al. [92] used Fire Weather Index (FWI), a meteorologically based fire risk metric derived from both HRES-IFS (ECMWF) and observational data, to forecast fire events in Chile, Portugal, and California. Their analysis concluded that extreme FWI values predicted 7–10 days before large wildfires, driven by prolonged dry conditions. However, limitations arise when anthropogenic fires are predominant, as meteorological-based forecast systems design fails to capture ignition sources.
Human activities, conditioning land use, and driving fire activity in the Magdalena Department are activities that include livestock, agriculture, and forest use, which have increased recently, generating conflicts that lead to wildfires in the central zone [47,93]. According to Sánchez-Cuervo et al. [94], since 2001 and throughout the following decade, agriculture and herbaceous vegetation types dominated this region’s landscape, at the expense of forest cover gain. Moreover, approximately 705 km2 underwent cover transformation in recent years, with 81% of these changes involving the conversion of sparsely vegetated areas and forest covers to agricultural and livestock areas. In contrast, forest-to-pasture transitions represented less than 10% of the changes [95]. These transformation processes were predominantly concentrated in the center, which coincided with identified fire-prone zones.
In frontier agricultural regions, fire drives and results from long-term anthropogenic cover dominance, often favoring fire-prone regrowth [96]. Landscape metrics showed that high fire-risk zones presented larger mean and core areas and slightly higher cohesion. To the northeast of the Magdalena Department, along the foothills of the SNSM, very dry tropical forests dominate, featuring xerophytic ecosystems with sparse shrub layers, whereas in the center, there are extensive patches of natural and anthropic tropical savannahs, covered by fine grass fuels. While the SNSM is susceptible to wildfires due to its vegetation, fire incidence is lower in the highlands, where paramo ecosystems, intense rainfall, and indigenous stewardship offer protection mechanisms [97]. The degree of fragmentation of these high-risk covers suggests that in this context, the continuity of fire-prone vegetation, not fragmentation per se, enhances fire susceptibility.
Furthermore, Maillard et al. [98] linked fire risk to fragmentation and edge effects in forest covers. In the fire-prone zones, gallery forest type predominates, with naturally elongated and irregular patches that, combined with fire use [99,100], agriculture [101], and livestock expansion [102], contribute to greater shape complexity, as reflected in the metric results. Evidence from the Magdalena Department forest loss [103] reflects an increasing fragmentation trend, where isolated, post-fire patches transition into degraded soils in response to inadequate forest management [104].
Nonetheless, a broader understanding of fragmentation dynamics is limited by insufficient multitemporal temperature data and a ground-based wildfire database to validate satellite-detected occurrences. Future research would benefit from integrated approaches that link remote sensing, emissions inventories, and health impact assessments to evaluate the combined effects of fire, fragmentation, and air quality degradation [105].

5. Conclusions

Significant fire distribution patterns were identified in the central and northeastern regions of the Magdalena Department, with marked seasonality during the dry season (DJFM) when February and March alone accounted for over 92% of total fires. The central region constitutes a critical fire hotspot with high-density clusters of thermal anomalies (p < 0.01) representing 8.90% of the study area. This spatiotemporal pattern demonstrates the predictable nature of fire occurrence in the region, which can contribute to prevention and reduction efforts.
The hydroclimatic analysis revealed a counterintuitive relationship where increased precipitation during the primary dry season enhanced fire probability (Sensitivity = 90.3%, p < 0.001). At the same time, the DSL index showed a significant negative association with fire occurrence. This paradoxical relationship suggests that dry season rainfall promotes vegetation growth that subsequently becomes fuel during prolonged dry conditions, highlighting the complex ecological interactions where fuel accumulation appears to be a more significant driver of fire risk than immediate moisture conditions.
Land cover analysis demonstrated that while pastures dominated the total burned area (4117.58 Ha) and fire occurrence (67.96%), they exhibited consistently higher vulnerability with elevated fire densities. In fire-prone zones, high-risk vegetation classes displayed significantly greater aggregation (AI = 86.35% vs. 69.36%) and larger mean patch areas (23.36 Ha vs. 6.56 Ha), indicating that xerophytic ecosystems and tropical savannas are disproportionately susceptible to wildfires. These spatial configuration patterns contribute substantially to fire risk, emphasizing that landscape structure is as crucial as vegetation type when assessing and managing wildfire vulnerability.

Supplementary Materials

The following supporting information can be downloaded at https://www.mdpi.com/article/10.3390/fire8060221/s1. Figure S1.1: Monthly climograms according to precipitation regimes in the Magdalena Department. (a) Simplified Caldas-Lang classification based on precipitation regimes in the Magdalena Department; (b) Monthly precipitation distribution in bimodal zones; (c) Monthly precipitation distribution in mixed (bimodal/monomodal) zone; (d) Monthly precipitation distribution in monomodal zone. Figure S2.1: Detailed information on the results of the monthly distribution of fire density (number of fires per 1000 Ha) between 2001 and 2023 in the Magdalena Department. Mentioned in Section 3.1. Figure S2.2: Classification of the Magdalena Department according to burned area. (a) Average burned area per 1000 Ha map. (b) Classification results by Getis–Ord Gi* p-value based on gridded burned area values. Table S2.1: Results of the binary logistic regression analysis for fire density ≥ optimal threshold (Q) in fire-prone zones of the Magdalena Department, with evaluation metrics derived from the classification of predicted probabilities from the fitted model using P Y = 1     X ≥ 0.5 as the classification criterion. * 0.1 < p < 0.05, ** 0.05 < p < 0.01, *** p < 0.01. Table S2.2: Year-wise average pre- and post-fire occurrence per land cover in the Magdalena Department. Pre-fire is described as average fire density, and post-fire as average burned area per 1000 Ha.

Author Contributions

Conceptualization, S.B.-D., T.D.S., A.M.V.-P. and Y.C.C.; methodology, S.B.-D., T.D.S., A.M.V.-P. and Y.C.C.; software, G.M.P.-M. and M.P.-P.; validation, T.D.S., A.M.V.-P. and Y.C.C.; formal analysis, A.M.V.-P. and Y.C.C.; investigation, S.B.-D., Y.C.C., G.M.P.-M. and M.P.-P.; data curation, S.B.-D., T.D.S., A.M.V.-P. and Y.C.C.; writing—original draft preparation, G.M.P.-M. and M.P.-P.; writing—review and editing, A.M.V.-P., Y.C.C. and T.R.B.-O.; visualization, G.M.P.-M. and M.P.-P.; supervision, Y.C.C.; funding acquisition, Y.C.C. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by Universidad del Magdalena through Grant FONCIENCIAS 2024.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The datasets employed for the analysis in this study are publicly accessible. Archived active fire and burned area data are available through the NASA FIRMS website at https://firms.modaps.eosdis.nasa.gov/download/, accessed on 1 August 2024. As for precipitation data, they can be accessed through the DHIME platform at http://dhime.ideam.gov.co/webgis/home/, accessed on 8 October 2024, while land cover map of the Magdalena Department can be retrieved from the “Colombia en mapas” platform at https://www.colombiaenmapas.gov.co, accessed on 6 Febraury 2025.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Bowman, D.M.; Balch, J.K.; Artaxo, P.; Bond, W.J.; Carlson, J.M.; Cochrane, M.A.; d’Antonio, C.M.; DeFries, R.S.; Doyle, J.C.; Harrison, S.P. Fire in the Earth System. Science 2009, 324, 481–484. [Google Scholar] [CrossRef] [PubMed]
  2. Sánchez, S.; Grilli, M.; Karlin, M.; Fachinetti, R.; Ravelo, A. Determinación de Regímenes de Incendios y Sequías Usando Información Satelital y Meteorológica Para Córdoba, Argentina. Agriscientia 2022, 39, 1–10. [Google Scholar] [CrossRef]
  3. Bonfanti, F.A.; Sánchez, M.E. Los Incendios Forestales En El Este Del Chaco Durante El Período 2015–2020 y Su Implicancia Ambiental. Una Visión Desde La Geografía. In Proceedings of the XL Encuentro de Geohistoria Regional; Consejo Nacional de Investigaciones Científicas y Técnicas: Buenos Aires, Argentina, 2021. [Google Scholar]
  4. Moritz, M.A.; Batllori, E.; Bradstock, R.A.; Gill, A.M.; Handmer, J.; Hessburg, P.F.; Leonard, J.; McCaffrey, S.; Odion, D.C.; Schoennagel, T. Learning to Coexist with Wildfire. Nature 2014, 515, 58–66. [Google Scholar] [CrossRef] [PubMed]
  5. Yue, W.; Ren, C.; Liang, Y.; Liang, J.; Lin, X.; Yin, A.; Wei, Z. Assessment of Wildfire Susceptibility and Wildfire Threats to Ecological Environment and Urban Development Based on GIS and Multi-Source Data: A Case Study of Guilin, China. Remote Sens. 2023, 15, 2659. [Google Scholar] [CrossRef]
  6. Andela, N.; Morton, D.C.; Giglio, L.; Chen, Y.; van der Werf, G.R.; Kasibhatla, P.S.; DeFries, R.S.; Collatz, G.J.; Hantson, S.; Kloster, S.; et al. A Human-Driven Decline in Global Burned Area. Science 2017, 356, 1356–1362. [Google Scholar] [CrossRef]
  7. Swain, D.L.; Prein, A.F.; Abatzoglou, J.T.; Albano, C.M.; Brunner, M.; Diffenbaugh, N.S.; Singh, D.; Skinner, C.B.; Touma, D. Hydroclimate Volatility on a Warming Earth. Nat. Rev. Earth Environ. 2025, 6, 35–50. [Google Scholar] [CrossRef]
  8. Global Wildfire Information System Annual Area Burnt by Wildfires 2025. Available online: https://gwis.jrc.ec.europa.eu/apps/gwis.statistics/seasonaltrend (accessed on 21 May 2025).
  9. Yang, X.; Jin, X.; Zhou, Y. Wildfire Risk Assessment and Zoning by Integrating Maxent and GIS in Hunan Province, China. Forests 2021, 12, 1299. [Google Scholar] [CrossRef]
  10. Xie, L.; Zhang, R.; Zhan, J.; Li, S.; Shama, A.; Zhan, R.; Wang, T.; Lv, J.; Bao, X.; Wu, R. Wildfire Risk Assessment in Liangshan Prefecture, China Based on an Integration Machine Learning Algorithm. Remote Sens. 2022, 14, 4592. [Google Scholar] [CrossRef]
  11. United Nations Environment Programme-UNEP. Waves of Extreme Wildfires. In Frontiers 2022: Noise, Blazes and Mis-Matches—Emerging Issues of Environmental Concern; UNEP: Nairobi, Kenya, 2022; pp. 24–37. ISBN 978-92-807-3917-6. [Google Scholar]
  12. Pérez-Verdín, G.; Márquez-Linares, M.A.; Cortés-Ortíz, A.; Salmerón-Macías, M. Análisis Espacio-Temporal de La Ocurrencia de Incendios Forestales En Durango, México. Madera y Bosques 2013, 19, 37–58. [Google Scholar] [CrossRef]
  13. Ávila-Florez, D.Y.; Pompa-García, M.; Vargas-Pérez, E. Análisis Espacial de La Ocurrencia de Incendios Forestales En El Estado de Durango. Rev. Chapingo. Ser. Cienc. For. y del Ambiente 2010, 16, 253–260. [Google Scholar] [CrossRef]
  14. Nagle, E.J. Análisis de La Sucesión Vegetal En Áreas Quemadas Por Incendios Forestales Mediante Teledetección. Bachelor’s Thesis, Universidad de Manizales, Manizales, Colombia, 2022. Available online: https://ridum.umanizales.edu.co/xmlui/handle/20.500.12746/6629 (accessed on 8 April 2025).
  15. Earl, N.; Simmonds, I. Spatial and Temporal Variability and Trends in 2001–2016 Global Fire Activity. J. Geophys. Res. Atmos 2018, 123, 2524–2536. [Google Scholar] [CrossRef]
  16. Sayedi, S.S.; Abbott, B.W.; Vannière, B.; Leys, B.; Colombaroli, D.; Romera, G.G.; Słowiński, M.; Aleman, J.C.; Blarquez, O.; Feurdean, A. Assessing Changes in Global Fire Regimes. Fire Ecol. 2024, 20, 1–22. [Google Scholar] [CrossRef]
  17. Meng, Y.; Hao, Z.; Feng, S.; Zhang, X.; Hao, F. Increase in Compound Dry-Warm and Wet-Warm Events under Global Warming in CMIP6 Models. Glob. Planet. Change 2022, 210, 103773. [Google Scholar] [CrossRef]
  18. Pausas, J.G.; Keeley, J.E. Wildfires and Global Change. Front. Ecol. Environ. 2021, 19, 387–395. [Google Scholar] [CrossRef]
  19. Krawchuk, M.A.; Moritz, M.A.; Parisien, M.-A.; Van Dorn, J.; Hayhoe, K. Global Pyrogeography: The Current and Future Distribution of Wildfire. PLoS ONE 2009, 4, e5102. [Google Scholar] [CrossRef]
  20. Abatzoglou, J.T.; Williams, A.P.; Barbero, R. Global Emergence of Anthropogenic Climate Change in Fire Weather Indices. Geophys. Res. Lett. 2019, 46, 326–336. [Google Scholar] [CrossRef]
  21. Richardson, D.; Black, A.S.; Irving, D.; Matear, R.J.; Monselesan, D.P.; Risbey, J.S.; Squire, D.T.; Tozer, C.R. Global Increase in Wildfire Potential from Compound Fire Weather and Drought. NPJ Clim. Atmos. Sci. 2022, 5, 23. [Google Scholar] [CrossRef]
  22. Senande-Rivera, M.; Insua-Costa, D.; Miguez-Macho, G. Spatial and Temporal Expansion of Global Wildland Fire Activity in Response to Climate Change. Nat. Commun. 2022, 13, 1208. [Google Scholar] [CrossRef]
  23. Jones, M.W.; Smith, A.; Betts, R.; Canadell, J.G.; Prentice, I.C.; Le Quéré, C. Climate Change Increases the Risk of Wildfires. Available online: https://ueaeprints.uea.ac.uk/id/eprint/77983 (accessed on 30 March 2025).
  24. Jones, M.W.; Abatzoglou, J.T.; Veraverbeke, S.; Andela, N.; Lasslop, G.; Forkel, M.; Smith, A.J.; Burton, C.; Betts, R.A.; van der Werf, G.R. Global and Regional Trends and Drivers of Fire under Climate Change. Rev. Geophys. 2022, 60, e2020RG000726. [Google Scholar] [CrossRef]
  25. Pricope, N.G.; Binford, M.W. A Spatio-Temporal Analysis of Fire Recurrence and Extent for Semi-Arid Savanna Ecosystems in Southern Africa Using Moderate-Resolution Satellite Imagery. J. Environ. Manag. 2012, 100, 72–85. [Google Scholar] [CrossRef]
  26. Argañaraz, J.P. Dinámica Espacial del Fuego en las Sierras de Córdoba. Ph.D. Thesis, Universidad Nacional de Córdoba, Córdoba, Argentina, 2016. Available online: https://www.researchgate.net/publication/301635295_Dinamica_espacial_del_fuego_en_las_Sierras_de_Cordoba (accessed on 30 March 2025).
  27. Meira, A.C.; Nunes, A.; Sousa, A.; Lourenço, L. Mapping the Causes of Forest Fires in Portugal by Clustering Analysis. Geosciences 2020, 10, 53. [Google Scholar] [CrossRef]
  28. de Oliveira-Junior, J.F.; Filho, W.L.F.C.; Alves, L.E.R.; Lyra, G.B.; de Gois, G.; da Silva, C.A.; Dos Santos, P.J.; Sobral, B.S. Fire Foci Dynamics and Their Relationship with Socioenvironmental Factors and Meteorological Systems in the State of Alagoas, Northeast Brazil. Environ. Monit. Assess. 2020, 192, 654. [Google Scholar] [CrossRef] [PubMed]
  29. Laris, P.; Jo, A.; Wechsler, S.P. Effects of Landscape Pattern and Vegetation Type on the Fire Regime of a Mesic Savanna in Mali. J. Environ. Manag. 2018, 227, 134–145. [Google Scholar] [CrossRef] [PubMed]
  30. Bolaño-Díaz, S.; Camargo-Caicedo, Y.; Soro, T.D.; N’Dri, A.B.; Bolaño-Ortiz, T.R. Spatio-Temporal Characterization of Fire Using MODIS Data (2000–2020) in Colombia. Fire 2022, 5, 134. [Google Scholar] [CrossRef]
  31. Guevara, B.L. Análisis Espacio-Temporal de Incendios Del Norte y Centro de Suramérica: 2009–2019. Agric. Habitat 2021, 4, 13–23. [Google Scholar] [CrossRef]
  32. Comisión Nacional Asesora para la Investigación en Gestión del Riesgo de Desastres (CNAIGRD). In Investigaciones en Gestión del Riesgo de Desastres Para Colombia: Contribuciones Locales, Regionales y Nacionales 2023; Universidad Nacional de Colombia: Bogotá, Colombia, 2023; ISBN 978-958-55-0927-6.
  33. Mesa, O.; Urrea, V.; Ochoa, A. Trends of Hydroclimatic Intensity in Colombia. Climate 2021, 9, 120. [Google Scholar] [CrossRef]
  34. Montoya, A.F.H.; Sánchez, Ó.J.M. Climate Change and Space-Time Variability of Precipitation in Colombia. Rev. EIA/Engl. Version 2015, 12, 129–147. Available online: https://revistas.eia.edu.co/index.php/Reveiaenglish/article/view/913 (accessed on 24 April 2025).
  35. Hoyos, N.; Correa-Metrio, A.; Sisa, A.; Ramos-Fabiel, M.A.; Espinosa, J.M.; Restrepo, J.C.; Escobar, J. The Environmental Envelope of Fires in the Colombian Caribbean. Appl. Geogr. 2017, 84, 42–54. [Google Scholar] [CrossRef]
  36. Pomares-Meza, G.M.; Camargo-Caicedo, Y.; Vélez-Pereira, A.M. Long-Term Spatiotemporal Analysis of Precipitation Trends with Implications of ENSO-Driven Variability in the Department of Magdalena, Colombia. Water 2024, 16, 3372. [Google Scholar] [CrossRef]
  37. Cabrera, A.; Ferro, C.; Casallas, A.; López-Barrera, E.A. Wildfire Scenarios for Assessing Risk of Cover Loss in a Megadiverse Zone within the Colombian Caribbean. Sustainability 2024, 16, 3410. [Google Scholar] [CrossRef]
  38. Gutiérrez, L.C.; Gutiérrez, Y.; Noriega, O.d.J.; Rangel, R.B.; Fonseca Gamba, C.; Ortega, E.; García, H. Caracterización, Diagnóstico y Análisis de Vulnerabilidades y Amenazas en el Departamento del Magdalena: Incendios forestales. 2017. Available online: https://www.corpamag.gov.co/archivos/riesgosAmbientales/2017_Doc07_Incendios.pdf (accessed on 14 March 2025).
  39. Instituto de Hidrología, Meteorología y Estudios Ambientales-IDEAM. Atlas Climatológico de Colombia; Imprenta Nacional de Colombia: Bogotá, Colombia, 2017; ISBN 978-958-8067-95-7. [Google Scholar]
  40. Instituto Geográfico Agustín Codazzi-IGAC. Estudio General de Suelos y Zonificación de tierras: Departamento del Magdalena, escala 1:100.000; IGAC, Ed.; Imprenta Nacional de Colombia: Bogotá, Colombia, 2009; ISBN 978-958-8323-30-5. [Google Scholar]
  41. Urrea, V. Variabilidad Espacial y Temporal Del Ciclo Anual de Lluvia en Colombia. Master’s Thesis, Universidad Nacional de Colombia, Medellín, Colombia, 2017. Available online: https://repositorio.unal.edu.co/handle/unal/59836 (accessed on 9 April 2025).
  42. Revueltas, J.E.; Zabaleta, A.; Mercado, T.; Aguirre, S. Cambios En El Clima Local y Su Efecto En La Regulación Hídrica En Microcuencas Del Departamento Del Magdalena, Norte de Colombia. Inf. Tecnológ. 2020, 31, 193–206. [Google Scholar] [CrossRef]
  43. Cerón, W.L.; Andreoli, R.V.; Kayano, M.T.; Canchala, T.; Ocampo-Marulanda, C.; Avila-Diaz, A.; Antunes, J. Trend Pattern of Heavy and Intense Rainfall Events in Colombia from 1981–2018: A Trend-EOF Approach. Atmosphere 2022, 13, 156. [Google Scholar] [CrossRef]
  44. Rangel-Ch, O.J. Ecosistemas del Caribe Colombiano. In Colombia Diversidad Biótica XII: La Región Caribe de Colombia; Rangel, O.J., Ed.; CÓDICE Ltda: Bogotá, Colombia, 2012; pp. 963–1009. ISBN 978-958-761-215-8. [Google Scholar]
  45. Corporación Autonóma Regional del Magdalena-CORPAMAG. Determinantes Ambientales del Departamento del Magdalena, Para la Orientación de los Modelos de Ocupación Territorial. 2018. Available online: https://www.corpamag.gov.co/archivos/mapas/DeterminantesAmbientalesDptoMagdalena.pdf (accessed on 9 April 2025).
  46. Beltrán Polo, Y.T. Estudio y Localización de Lagos de Alta Montaña de la Sierra Nevada de Santa Marta en el Departamento del Magdalena. Bachelor’s Thesis, Universidad Cooperativa de Colombia, Santa Marta, Colombia, 2020. Available online: https://repository.ucc.edu.co/entities/publication/5d4a80e0-3473-4a98-a32a-dc479eada3ef (accessed on 20 May 2025).
  47. Instituto Geográfico Agustín Codazzi-IGAC. Atlas de Suelos: Departamento Del Magdalena; IGAC, Ed.; Imprenta Nacional de Colombia: Bogotá, Colombia, 2024; ISBN 978-958-5494-78-7. [Google Scholar]
  48. Chuvieco, E.; Cifuentes, Y.; Hantson, S.; López, A.A.; Ramo, R.; Torres, J. Comparación Entre Focos de Calor MODIS y Perímetros de Área Quemada En Incendios Mediterráneos. Teledetección 2012, 37, 9–22. [Google Scholar]
  49. Guillén, M.G.; Figuereido, M.S. Análisis de Incendios En Honduras a Partir de Imágenes Del Sensor MODIS C6 (2013–2018). Portal de la Cienc. 2022, 1, 32–40. [Google Scholar] [CrossRef]
  50. Giglio, L.; Schroeder, W.; Hall, J.; Justice, C. MODIS Active Fire and Burned Area Products—Active Fires. Available online: https://modis-fire.umd.edu/af.html (accessed on 19 October 2023).
  51. Ibrahim, M.A. Risk of Spontaneous and Anthropogenic Fires in Waste Management Chain and Hazards of Secondary Fires. Resour. Conserv. Recyc. 2020, 159, 104852. [Google Scholar] [CrossRef]
  52. Portela, M.M.; Espinosa, L.A.; Zelenakova, M. Long-Term Rainfall Trends and Their Variability in Mainland Portugal in the Last 106 Years. Climate 2020, 8, 146. [Google Scholar] [CrossRef]
  53. Lobodzinskyi, O.; Vasylenko, Y.; Koshkina, O.; Nabyvanets, Y. Assessing the Impact of Climate Change on Discharge in the Horyn River Basin by Analyzing Precipitation and Temperature Data. Meteorol. Hydrol. Water Manag. 2023, 11, 93–106. [Google Scholar] [CrossRef]
  54. Alexandersson, H. A Homogeneity Test Applied to Precipitation Data. J. Climatol. 1986, 6, 661–675. [Google Scholar] [CrossRef]
  55. Meseguer-Ruiz, O.; Ponce-Philimon, P.I.; Guijarro, J.A.; Sarricolea, P. Spatial Distribution and Trends of Different Precipitation Variability Indices Based on Daily Data in Northern Chile between 1966 and 2015. Int. J. Climatol. 2019, 39, 4595–4610. [Google Scholar] [CrossRef]
  56. Meseguer-Ruiz, O.; Ponce-Philimon, P.I.; Quispe-Jofré, A.S.; Guijarro, J.A.; Sarricolea, P. Spatial Behaviour of Daily Observed Extreme Temperatures in Northern Chile (1966–2015): Data Quality, Warming Trends, and Its Orographic and Latitudinal Effects. Stoch. Environ. Res. Risk Assess. 2018, 32, 3503–3523. [Google Scholar] [CrossRef]
  57. Caloiero, T.; Filice, E.; Coscarelli, R.; Pellicone, G. A Homogeneous Dataset for Rainfall Trend Analysis in the Calabria Region (Southern Italy). Water 2020, 12, 2541. [Google Scholar] [CrossRef]
  58. Javanshiri, Z.; Pakdaman, M.; Falamarzi, Y. Homogenization and Trend Detection of Temperature in Iran for the Period 1960–2018. Meteorol Atmos. Phys. 2021, 133, 1233–1250. [Google Scholar] [CrossRef]
  59. Soro, T.D.; Koné, M.; N’Dri, A.B.; N’Datchoh, E.T. Identified Main Fire Hotspots and Seasons in Côte d’Ivoire (West Africa) Using MODIS Fire Data. S. Afr. J. Sci. 2021, 117, 1–13. [Google Scholar] [CrossRef]
  60. Palacio, M.E.Q.; Villa, H.M.S. Manejo Del Fuego Como Alternativa Frente a Los Incendios Forestales. El Caso Del Parque Entrenubes. Ambiente y Desarro. 2019, 23, 2. [Google Scholar]
  61. Alvarado, S.T.; Hantson, S. Dinámica espacio-temporal de incendios forestales en la región de Los Llanos en Colombia y Venezuela. In Proceedings of the Congreso Internacional de Ciencias Básicas e Ingeniería, Villavicencio, Colombia, 1 February 2023; p. 5. [Google Scholar]
  62. Xie, W.-F.; Li, J.-K.; Peng, K.; Zhang, K.; Ullah, Z. The Application of Local Moran’s I and Getis–Ord Gi* to Identify Spatial Patterns and Critical Source Areas of Agricultural Nonpoint Source Pollution. J. Environ. Eng. 2024, 150, 04024011. [Google Scholar] [CrossRef]
  63. Kumari, S.; Mamgain, S.; Roy, A.; Prince, H.C. Characterising Spatial Clusters of Forest Fire Activity in the Western Himalayan Region of India: Implications for Conservation and Management. Int. J. Wildland Fire 2025, 34, 2. [Google Scholar] [CrossRef]
  64. Lanorte, A.; Nolè, G.; Cillis, G. Application of Getis-Ord Correlation Index (Gi) for Burned Area Detection Improvement in Mediterranean Ecosystems (Southern Italy and Sardinia) Using Sentinel-2 Data. Remote Sens. 2024, 16, 2943. [Google Scholar] [CrossRef]
  65. Mupfiga, U.N.; Mutanga, O.; Dube, T.; Kowe, P. Spatial Clustering of Vegetation Fire Intensity Using MODIS Satellite Data. Atmosphere 2022, 13, 1972. [Google Scholar] [CrossRef]
  66. Siabato, W.; Guzmán-Manrique, J.; Siabato, W.; Guzmán-Manrique, J. La autocorrelación espacial y el desarrollo de la geografía cuantitativa. Cuad. de Geogr. Rev. Colomb. de Geogr. 2019, 28, 1–22. [Google Scholar] [CrossRef]
  67. Wang, Z.; and Lam, N.S.N. Extending Getis–Ord Statistics to Account for Local Space–Time Autocorrelation in Spatial Panel Data. Prof. Geogr. 2020, 72, 411–420. [Google Scholar] [CrossRef]
  68. Getis, A.; Ord, J.K. The Analysis of Spatial Association by Use of Distance Statistics. Geogr. Anal. 1992, 24, 189–206. [Google Scholar] [CrossRef]
  69. Teegavarapu, R.S.V. Chapter 1—Methods for Analysis of Trends and Changes in Hydroclimatological Time-Series. In Trends and Changes in Hydroclimatic Variables; Teegavarapu, R., Ed.; Elsevier: Amsterdam, The Netherlands, 2019; pp. 1–89. ISBN 978-0-12-810985-4. [Google Scholar]
  70. Zhang, F.; Zhang, B.; Luo, J.; Liu, H.; Deng, Q.; Wang, L.; Zuo, Z. Forest Fire Driving Factors and Fire Risk Zoning Based on an Optimal Parameter Logistic Regression Model: A Case Study of the Liangshan Yi Autonomous Prefecture, China. Fire 2023, 6, 336. [Google Scholar] [CrossRef]
  71. Vélez-Pereira, A.M.; De Linares, C.; Canela, M.-A.; Belmonte, J. Logistic Regression Models for Predicting Daily Airborne Alternaria and Cladosporium Concentration Levels in Catalonia (NE Spain). Int. J. Biometeorol. 2019, 63, 1541–1553. [Google Scholar] [CrossRef] [PubMed]
  72. Myers, R.H. Classical and Modern Regression with Applications; PWS-KENT: Belmont, CA, USA, 1990; ISBN 978-0-534-92178-1. [Google Scholar]
  73. Giorgi, F.; Im, E.-S.; Coppola, E.; Diffenbaugh, N.S.; Gao, X.J.; Mariotti, L.; Shi, Y. Higher Hydroclimatic Intensity with Global Warming. J. Clim. 2011, 24, 20. [Google Scholar] [CrossRef]
  74. Hosmer, D.W.; Lemeshow, S.; Sturdivant, R.X. Applied Logistic Regression, 3rd ed.; John Wiley & Sons: Hoboken, NJ, USA, 2013; ISBN 978-0-470-58247-3. [Google Scholar]
  75. Instituto de Hidrología, Meteorología y Estudios Ambientales-IDEAM. Protocolo Para la Realización de Mapas de Zonificación de Riesgos a Incendios de la Cobertura Vegetal en Colombia—Escala 1:100.000; IDEAM: Bogotá, Colombia, 2011; ISBN 978-958-8067-41-4.
  76. Crist, M.R. Rethinking the Focus on Forest Fires in Federal Wildland Fire Management: Landscape Patterns and Trends of Non-Forest and Forest Burned Area. J. Environ. Manag. 2023, 327, 116718. [Google Scholar] [CrossRef]
  77. Montibeller, B.; Kmoch, A.; Virro, H.; Mander, Ü.; Uuemaa, E. Increasing Fragmentation of Forest Cover in Brazil’s Legal Amazon from 2001 to 2017. Sci. Rep. 2020, 10, 5803. [Google Scholar] [CrossRef]
  78. Hagmann, R.K.; Hessburg, P.F.; Salter, R.B.; Merschel, A.G.; Reilly, M.J. Contemporary Wildfires Further Degrade Resistance and Resilience of Fire-Excluded Forests. For. Ecol. Manag. 2022, 506, 119975. [Google Scholar] [CrossRef]
  79. McGarigal, K.S.; Cushman, S.; Neel, M.; Ene, E. FRAGSTATS: Spatial Pattern Analysis Program for Categorical Maps 2015. Available online: https://www.researchgate.net/publication/259011515_FRAGSTATS_Spatial_pattern_analysis_program_for_categorical_maps (accessed on 30 March 2025).
  80. Hesselbarth, M.H.K.; Sciaini, M.; Nowosad, J.; Hanss, S. Landscapemetrics: Landscape Metrics for Categorical Map Patterns 2025, 2.2.1. Available online: https://CRAN.R-project.org/package=landscapemetrics (accessed on 30 March 2025).
  81. NOAA Cold & Warm Episodes by Season. Available online: https://origin.cpc.ncep.noaa.gov/products/analysis_monitoring/ensostuff/ONI_v5.php (accessed on 24 April 2025).
  82. Ramos Rodríguez, M.; Poma Cabrera, E.; Reyes Chancay, J. Estadísticas de Incendios Forestales En El Cantón Loja, Provincia Loja, Ecuador, En El Periodo 2011–2020. Bosques Latid. Cero 2024, 14, 64–76. [Google Scholar] [CrossRef]
  83. Zubieta, R.; Ccanchi, Y.; Liza, R. Performance of Heat Spots Obtained from Satellite Datasets to Represent Burned Areas in Andean Ecosystems of Cusco, Peru. Remote Sens. Appl. Soc. Environ. 2023, 32, 101020. [Google Scholar] [CrossRef]
  84. Obando-Cabrera, L.; Díaz-Timoté, J.J.; Bastarrika, A.; Celis, N.; Hantson, S. The Paramo Fire Atlas: Quantifying Burned Area and Trends across the Tropical Andes. Environ. Res. Lett. 2025, 20, 054019. [Google Scholar] [CrossRef]
  85. Sharma, S.; Carlson, J.D.; Krueger, E.S.; Engle, D.M.; Twidwell, D.; Fuhlendorf, S.D.; Patrignani, A.; Feng, L.; Ochsner, T.E. Soil Moisture as an Indicator of Growing-Season Herbaceous Fuel Moisture and Curing Rate in Grasslands. Int. J. Wildland Fire 2020, 30, 57–69. [Google Scholar] [CrossRef]
  86. Bousquet, E.; Mialon, A.; Rodriguez-Fernandez, N.; Mermoz, S.; Kerr, Y. Monitoring Post-Fire Recovery of Various Vegetation Biomes Using Multi-Wavelength Satellite Remote Sensing. Biogeosciences 2022, 19, 3317–3336. [Google Scholar] [CrossRef]
  87. Ermitão, T.; Gouveia, C.M.; Bastos, A.; Russo, A.C. Interactions between Hot and Dry Fuel Conditions and Vegetation Dynamics in the 2017 Fire Season in Portugal. Environ. Res. Lett. 2022, 17, 095009. [Google Scholar] [CrossRef]
  88. Alizadeh, M.R.; Adamowski, J.; Entekhabi, D. Land and Atmosphere Precursors to Fuel Loading, Wildfire Ignition and Post-Fire Recovery. Geophys. Res. Lett. 2024, 51, e2023GL105324. [Google Scholar] [CrossRef]
  89. Qu, Y.; Miralles, D.G.; Veraverbeke, S.; Vereecken, H.; Montzka, C. Wildfire Precursors Show Complementary Predictability in Different Timescales. Nat. Commun. 2023, 14, 6829. [Google Scholar] [CrossRef]
  90. Resco de Dios, V.; Cunill Camprubí, À.; Pérez-Zanón, N.; Peña, J.C.; Martínez del Castillo, E.; Rodrigues, M.; Yao, Y.; Yebra, M.; Vega-García, C.; Boer, M.M. Convergence in Critical Fuel Moisture and Fire Weather Thresholds Associated with Fire Activity in the Pyroregions of Mediterranean Europe. Sci. Total Environ. 2022, 806, 151462. [Google Scholar] [CrossRef]
  91. Gincheva, A.; Pausas, J.G.; Torres-Vázquez, M.Á.; Bedia, J.; Vicente-Serrano, S.M.; Abatzoglou, J.T.; Sánchez-Espigares, J.A.; Chuvieco, E.; Jerez, S.; Provenzale, A.; et al. The Interannual Variability of Global Burned Area Is Mostly Explained by Climatic Drivers. Earth’s Future 2024, 12, e2023EF004334. [Google Scholar] [CrossRef]
  92. Di Giuseppe, F.; Vitolo, C.; Krzeminski, B.; Barnard, C.; Maciel, P.; San-Miguel, J. Fire Weather Index: The Skill Provided by the European Centre for Medium-Range Weather Forecasts Ensemble Prediction System. Nat. Hazards Earth Syst. Sci. 2020, 20, 2365–2378. [Google Scholar] [CrossRef]
  93. Departamento Administrativo Nacional de Estadistica-DANE. Encuesta Nacional Agropecuaria (ENA). Available online: https://www.dane.gov.co/index.php/estadisticas-por-tema/agropecuario/encuesta-nacional-agropecuaria-ena (accessed on 21 May 2025).
  94. Sánchez-Cuervo, A.M.; Aide, T.M.; Clark, M.L.; Etter, A. Land Cover Change in Colombia: Surprising Forest Recovery Trends between 2001 and 2010. PLoS ONE 2012, 7, e43943. [Google Scholar] [CrossRef]
  95. Instituto de Hidrología, Meteorología y Estudios Ambientales-IDEAM. Sistema de Monitoreo de Las Coberturas de La Tierra—SIMCOT. Available online: https://experience.arcgis.com/experience/6f82270e8c2b4872979478f43e1494d5/page/Mapa-de-Cambios (accessed on 24 April 2025).
  96. San Martín, R. Fires, Land Use, and Forest Loss in the South American Chaco: Understanding the Links between Fires, Climate, Ecosystems, and Human Activity through Remote Sensing. Ph.D. Thesis, Université Paris-Saclay, Orsay, France, 2024. Available online: https://theses.hal.science/tel-04885407 (accessed on 11 April 2025).
  97. Duran-Izquierdo, M.; Olivero-Verbel, J. Vulnerability Assessment of Sierra Nevada de Santa Marta, Colombia: World’s Most Irreplaceable Nature Reserve. Glo. Ecol. Conser. 2021, 28, e01592. [Google Scholar] [CrossRef]
  98. Maillard, O.; Vides-Almonacid, R.; Flores-Valencia, M.; Coronado, R.; Vogt, P.; Vicente-Serrano, S.M.; Azurduy, H.; Anívarro, R.; Cuellar, R.L. Relationship of Forest Cover Fragmentation and Drought with the Occurrence of Forest Fires in the Department of Santa Cruz, Bolivia. Forests 2020, 11, 910. [Google Scholar] [CrossRef]
  99. Mojica, J.I.; González-Afanador, E.; Donato-Rondón, J.; Guillot-Monroy, G.H. Chapter 3—Magdalena. In Rivers of South America; Graça, M.A.S., Callisto, M., de Mello, F.T., Rodríguez-Olarte, D., Eds.; Elsevier: Amsterdam, The Netherlands, 2025; pp. 57–91. ISBN 978-0-12-823429-7. [Google Scholar]
  100. Guáqueta-Solórzano, V.-E.; Postigo, J.C. Indigenous Perceptions and Adaptive Responses to the Impacts of Climate Variability in the Sierra Nevada de Santa Marta, Colombia. Front. Clim. 2022, 4, 910294. [Google Scholar] [CrossRef]
  101. Ministerio de Ambiente y Desarrollo Sostenible. Plan Integral de Gestión Del Cambio Climático Territorial Del Departamento de Magdalena; Ministerio de Ambiente y Desarrollo Sostenible: Bogotá, Colombia, 2015; p. 144.
  102. Garrido, A.P.; Bernal, F.T.; Fontanilla, J.D.; Caicedo, Y.C.; Vélez-Pereira, A.M. Assessment of Livestock Greenhouse Gases in Colombia between 1995 and 2015. Heliyon 2022, 8, e12262. [Google Scholar] [CrossRef] [PubMed]
  103. University of Maryland; World Resources Institute Magdalena, Colombia Deforestation Rates & Statistics. Available online: https://www.globalforestwatch.org/dashboards/country/COL/19?category=forest-change (accessed on 24 April 2025).
  104. Braun, A.C.; Faßnacht, F.; Valencia, D.; Sepulveda, M. Consequences of Land-Use Change and the Wildfire Disaster of 2017 for the Central Chilean Biodiversity Hotspot. Reg. Environ. Change 2021, 21, 37. [Google Scholar] [CrossRef]
  105. Wang, Y.; Tian, Y.; Shi, Y. Estimation of Biomass Burning Emissions in South and Southeast Asia Based on FY-4A Satellite Observations. Atmosphere 2025, 16, 582. [Google Scholar] [CrossRef]
Fire 08 00221 g001
Figure 1. Study area which the right panel shows the Magdalena Department in Colombia and South America, while the left panel shows the area with its different land covers.
Figure 1. Study area which the right panel shows the Magdalena Department in Colombia and South America, while the left panel shows the area with its different land covers.
Fire 08 00221 g001
Fire 08 00221 g002
Figure 2. Distribution of thermal anomalies and burned area, Magdalena Department (2001–2023). (a) Distribution of thermal anomalies (2001–2023). (b) Distribution of burned area (2001–2023).
Figure 2. Distribution of thermal anomalies and burned area, Magdalena Department (2001–2023). (a) Distribution of thermal anomalies (2001–2023). (b) Distribution of burned area (2001–2023).
Fire 08 00221 g002
Fire 08 00221 g003
Figure 3. Time-averaged spatial distribution of average fire density of the Magdalena Department (2001–2023). (a) Average fire density map. (b) Seasonal contribution to average fire density, obtained from intra-annual fire occurrence in Colombia.
Figure 3. Time-averaged spatial distribution of average fire density of the Magdalena Department (2001–2023). (a) Average fire density map. (b) Seasonal contribution to average fire density, obtained from intra-annual fire occurrence in Colombia.
Fire 08 00221 g003
Fire 08 00221 g004
Figure 4. Classification results by gridded Getis Ord-Gi* p-value and value of fire density.
Figure 4. Classification results by gridded Getis Ord-Gi* p-value and value of fire density.
Fire 08 00221 g004
Fire 08 00221 g005
Figure 5. Overlay analysis for identification of fire-prone zones. (a) Spatial distribution of fire-prone coincident areas between Getis Ord-Gi* classification of fire density and burned area per 1000 Ha (b) Relationship between fire density and burned area per 1000 Ha classification. Flow width represents the proportion of grid cells transitioning between categories of significance.
Figure 5. Overlay analysis for identification of fire-prone zones. (a) Spatial distribution of fire-prone coincident areas between Getis Ord-Gi* classification of fire density and burned area per 1000 Ha (b) Relationship between fire density and burned area per 1000 Ha classification. Flow width represents the proportion of grid cells transitioning between categories of significance.
Fire 08 00221 g005
Fire 08 00221 g006
Figure 6. Comparative results of logistic regression performed between seasonal fire density binary response and hydroclimatic variables. (a–c) Long-term seasonal relationships; (d–f) DJFM synchronized and precursor relationships; (g–i) AMJJ synchronized and precursor relationships; and (j–l) ASON synchronized and precursor relationships.
Figure 6. Comparative results of logistic regression performed between seasonal fire density binary response and hydroclimatic variables. (a–c) Long-term seasonal relationships; (d–f) DJFM synchronized and precursor relationships; (g–i) AMJJ synchronized and precursor relationships; and (j–l) ASON synchronized and precursor relationships.
Fire 08 00221 g006
Fire 08 00221 g007
Figure 7. Distribution of fire variables across land covers in the Magdalena Department (2001–2023). (a) Land covers in the Magdalena Department, red polygons symbolize fire-prone zones. (b) Temporal distribution of annual burned area per 1000 Ha per land cover, compared to fire density. (c) Year-wise contribution of each land cover to fire density in the Magdalena Department.
Figure 7. Distribution of fire variables across land covers in the Magdalena Department (2001–2023). (a) Land covers in the Magdalena Department, red polygons symbolize fire-prone zones. (b) Temporal distribution of annual burned area per 1000 Ha per land cover, compared to fire density. (c) Year-wise contribution of each land cover to fire density in the Magdalena Department.
Fire 08 00221 g007
Fire 08 00221 g008
Figure 8. Fragmentation-based assessment results for fire-prone zones in the Magdalena Department. (a) Fire risk-based reclassification of land covers, red and blue polygons symbolize fire-prone and non-fire-prone zones, respectively. (b) Comparison of landscape metrics for the fire-prone and non-fire-prone zones.
Figure 8. Fragmentation-based assessment results for fire-prone zones in the Magdalena Department. (a) Fire risk-based reclassification of land covers, red and blue polygons symbolize fire-prone and non-fire-prone zones, respectively. (b) Comparison of landscape metrics for the fire-prone and non-fire-prone zones.
Fire 08 00221 g008
Table 1. Landscape metrics selected for fragmentation-based assessment of high-risk covers in both fire-prone and non-fire-prone zones.
Table 1. Landscape metrics selected for fragmentation-based assessment of high-risk covers in both fire-prone and non-fire-prone zones.
TypeMetricAbbr.Definition
AggregationAggregation index (%)AIConnectedness of a class based on the ratio of actual like adjacencies to the maximum possible, with higher values indicating lower spatial distribution.
Cohesion index (%)COHESIONConnectedness of a class, proportional to the area and shape metrics, indicating whether the patches belonging to the same class are aggregated or isolated.
Mean contiguity indexCONTIG_MNConnectedness of a class, based on quantifying the patches’ links. Ranges from 0 to 1, with higher values indicating more continuous and connected patches.
ShapeMean shape indexSHAPE_MNAverage class shape complexity is based on the perimeter ratio of the area’s square root.
Mean fractal dimension FRAC_MNAverage class shape complexity based on how the perimeter scales with area, summarizing each class as the mean fractal dimension of its patches. Ranges from 1 to 2, with higher values indicating a highly irregular, plane-filling shape.
AreaMean area (Ha)AREA_MNAverage class area of all areas belonging to the same class.
Mean core area (Ha)CORE_MNAverage core area of a class, where the core consists of all patches without neighboring patches of a different class.
Percentage of landscape (%)PLANDPercentage of landscapes conforming to the same class.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Camargo Caicedo, Y.; Bolaño-Diaz, S.; Pomares-Meza, G.M.; Pérez-Pérez, M.; Soro, T.D.; Bolaño-Ortiz, T.R.; Vélez-Pereira, A.M. Hydroclimatic and Land Use Drivers of Wildfire Risk in the Colombian Caribbean. Fire 2025, 8, 221. https://doi.org/10.3390/fire8060221

AMA Style

Camargo Caicedo Y, Bolaño-Diaz S, Pomares-Meza GM, Pérez-Pérez M, Soro TD, Bolaño-Ortiz TR, Vélez-Pereira AM. Hydroclimatic and Land Use Drivers of Wildfire Risk in the Colombian Caribbean. Fire. 2025; 8(6):221. https://doi.org/10.3390/fire8060221

Chicago/Turabian Style

Camargo Caicedo, Yiniva, Sindy Bolaño-Diaz, Geraldine M. Pomares-Meza, Manuel Pérez-Pérez, Tionhonkélé Drissa Soro, Tomás R. Bolaño-Ortiz, and Andrés M. Vélez-Pereira. 2025. "Hydroclimatic and Land Use Drivers of Wildfire Risk in the Colombian Caribbean" Fire 8, no. 6: 221. https://doi.org/10.3390/fire8060221

APA Style

Camargo Caicedo, Y., Bolaño-Diaz, S., Pomares-Meza, G. M., Pérez-Pérez, M., Soro, T. D., Bolaño-Ortiz, T. R., & Vélez-Pereira, A. M. (2025). Hydroclimatic and Land Use Drivers of Wildfire Risk in the Colombian Caribbean. Fire, 8(6), 221. https://doi.org/10.3390/fire8060221

Article Metrics

Back to TopTop