In the Tamil Nadu State, Vellore District is located between 78° 20′to 78° 50′east longitude and 12°15′and 13°15′north latitudes. Tamil Nadu is divided into 32 districts, one of which is Vellore. Vellore District covers 6,077 sq. km geographically. The Vellore district has 27% forest cover, around 162,286 ha, most located in the taluks of Gudiyatham, Tirupattur, and Vellore. (Vellore Reserve Forest (RF) 2023).

Located in the eastern area of Tirupattur Taluk, the Javadhu Hills are the highest mountain in the Vellore district. The Javadhu Hills have peaks reaching up to 1,280.16 m, and their elevation is 762 m above mean sea level. The 975.36 m tall Yelagiri Hills are placed at the midpoint of the Tirupathur Taluk. The district is home to the Palar and Ponnai, two significant rivers. The capability for irrigation-led agriculture is restricted because both rivers are seasonal and rain-fed. Red loamy soil is in the other sections, and black soil is in the tank and river bottoms (District Administration V. and State Planning Commission Tamil Nadu in association with University V, 2017).

The grasslands and mountain vegetation of the Vellore district are blended with the forests. In addition, the wetland/dry land cover and forest classifications for 2001 and 2006 were determined. There were notable differences in dry farmland, hilly areas with vegetation, and agricultural areas throughout 10 years. Between 2001 and 2006, there was an approximate 6% and 23% decrease in forest or bushland and open area cover types, respectively, while agricultural and built-up areas and water areas had an approximate 19%, 4%, and 7% increase, accordingly (Gandhi et al., 2015).

In 2010, Vellore had 36.1 thousand hectares of natural forest, outspreading over 6.0% of its land area. In 2023, it lost 4 ha of natural forest; as of 2000, 8.6% of Vellore’s land area was >30% tree cover, natural forest 52.0 kha, plantations 499 ha, and other land cover 554 thousand hectares.

In 2020, Vellore had >30% tree cover by land as follows: forest 52% (179 thousand hectares), grassland 20% (66.3 thousand hectares), settlement 16% (54.2 thousand hectares), wetland <0.1% (155 ha), cropland 12% (40.1 thousand hectares), Other <0.1%. While tree cover density is high, in 2020, Vellore had 470 ha of land above 10% tree cover, extending over 77.7% of its land area. As of 2001, 27% of Vellore’s total tree cover was primary forest. Primary Forest 14.2 thousand hectares, Other Tree Cover 38.3 thousand hectares, non-forest 554 kha. Furthermore, from 2002 to 2023, Vellore lost 80 ha of humid primary forest, making up 24% of its total tree cover loss. In the same period, the total area of humid primary forest in Vellore decreased by 0.56%. From 2013 to 2023, 97% of tree cover loss in Vellore occurred within natural forests (Global Forest Watch, Vellore Reserve Forest (RF) 2023) (Figure 1).

This present research paper relies on independent metrics, such as the fuel danger index (FDI), activity danger index (ADI), weather danger index (WDI), and topographic danger index (TDI), as independent value factors of forest fires due to metrics’ immediate impact on the occurrence of fires (Sivrikaya and Küçük, 2022) (Figure 2).

As a metric for measuring fire severity, the Normalized Burn Ratio (NBR) is a well-known index often utilized in research to emphasize scorched regions in huge fire areas. The Normalized Burn Ratio (NBR) is the proportion of the near-infrared (NIR) to short wavelength infrared (SWIR) bands. In turn, burned places reflect a more critical quantity of radiation in the wavelength’s visible and short wavelength infrared (SWIR) regions and obtain more radiation in the NIR; green vegetation consumes red light in the visible band (Zhao et al., 2021; Pramanick et al., 2023). NBR images were created for pre- and post-fire accidents, so maps of NBR-pre (NBR before the fire) and NBR-post (NBR following the fire) were generated (Alcaras et al., 2022).

Land-sat data were utilized in this study for NBR-pre- and post-fire and merging the use of both NIR (B5) and SWIR (B7) wavelengths by using a Land-sat nine data dated February/2023 with 30 m resolution from USGS Global Visualization Viewer website https://glovis.usgs.gov/app. Employ the NIR and SWIR parameters to simulate NBR using the bands coupled with each data for both pre-and post-fire images, Equation 5 (Khoirunisa and Mucsi, 2020; Zubkova et al., 2021): N B R = N I R − S W I R 2 N I R + S W I R 2 (5)

Where:

NIR: near-infrared.

SWIR: Short wavelength infrared.

For Landsat 9: NBR = B 5 − B 7 B 5 + B 7

NBR was conducted for the Landsat nine satellite data taken for the pre-fire (14/Feb/2023) and post-fire (19/Oct/2023) time and measured using Equation 5.

The Normalized Burn Ratio (NBR) is a typically used method that presents the entire burning area for the prior-fire and post-fire period. However, the most often employed satellite based on data parameters to simulate burning intensity is the Delta Normalized Burn Ratio (dNBR) (Uttaruk et al., 2022) (Figure 3).

Additionally, the dNBR is measured by deducting the pre-fire NBR from the post-fire NBR, Equation 6 for computing dNBR (Khoirunisa and Mucsi, 2020): d N B R = N B R p r e − f i r e − N B R p o s t − f i r e (6)

The dNBR ranged from −2 to +2 (Uttaruk et al., 2022). Values above 0 denote regions with more severe blazes, while rates less than 0 represent areas with fewer severe burns. One study suggested that the appropriate data rank between −0.5 and 1.3, where dNBR lower than −0.5 or more than 1.3 may occur but are usually not elaborated as scorched areas. Instead, they are likely due to misregistration, clouds, or other sources, not due to actual variations in land cover (Syaufina et al., 2022).

The dNBR is a valuable factor for determining fire severity in forest ecosystems, providing insights into the impacts of wildfires on greenery cover and natural ecosystem quality. One study noted that high dNBR metrics refer to severe burn severity, leading to enormous damage to flora and fauna within ecosystems. Severe wildfires with high dNBR rates can disrupt environmental processes like nutrients and the water cycle, affecting the ecosystem balance. A moderate positive bond between dNBR elements and prior-fire vegetation was displayed, which denoted fire severity impacted by the amount of fuel in nature. Also, as the topography greatly influences forest fire behavior, it impacts fire severity in different areas (Heidari and Arfania, 2022).

All the independent factors referred to above have the ability to ignite and cause wildfires. Therefore, this paper has taken the historical forest fire events and their size as the dependent parameters. Vellore Reserve Forest (RF) (2023) confirmed that all the preceding metrics elevated the fire risk in the Vellore district; naturally occurring forest covered 36.1 kha (or 6.0%) of Vellore’s total land region in 2010. A 4-ha natural forest was lost in 2023. Twelve hectares of tree cover were lost to fires in Vellore between 2001 and 2023, whereas 410 ha were lost to all other causes of loss. With 2 ha lost to blaze, or 5.0% of all tree cover loss for that year, 2019 had the most due to fires during this era. Whereas between the 16th of August 2021 and the 12th of August 2024, Vellore witnessed 293 VIIRS Alerts fire alerts (Global Forest Watch 2024) https://www.globalforestwatch.org/dashboards/global/; In brief, several factors, including fuel elements like vegetation density, kind and moisture content, climate conditions, topography factors, and human activities, influence the incidence and spread of wildfires. Likewise, field observations help validate remote sensing data and models, provide information on the effects of fires on vegetation degradation and environmental changes brought on by wildfires, and provide valuable data for calculating fire risk and simulating fire behavior.

Finally, the study assessed the Fuel Danger Index (FDI) based on density and moisture content. The greenery density was measured using the Normalized Difference Vegetation Index (NDVI); a more excellent score means the vegetation is more susceptible to fire in a highly intensive fire. In addition, since drier fuels have a higher ignite potential, the vegetation’s wetness content was measured using the Normalized Difference Moisture Index (NDMI). These two components, density and moisture, were combined to replicate the overall FDI. Because forest fires are more likely to start when human activity is close to a forest, the Human Activity Danger Index (ADI) was enhanced using the distance between highways and communities and the forest. Meteorological metrics involving temperature, relative humidity, and wind speed are considered by the Weather Danger Index (WDI), which directly affects how fires behave. Slope, aspect, and elevation are used in the Topographic Danger Index (TDI) to determine the impacts on fire severity and spread. In addition, this research work applied the Normalized Burn Ratio (NBR) and Differenced NBR (dNBR) to rate the fire’s severity based on pre- and post-fire satellite data. Combining many variables makes a comprehensive assessment of wildfire danger and behavior in the Vellore Forest region possible.

In Vellore woods, vegetation type and density play a crucial role as igniting agents in forest fires. The mountain vegetation and grasslands blend with the trees. This study analyzed Shrubland and temperate and tropical rainforests in the woods. Due to the extended dry season, dense grasses such as lemon grass and shrubs take the place of trees in the forest, making them an excellent ignition source. In this study area, the NDVI ranges from −0.188 to +0.947. The NDVI was classified into five classes using ArcGIS 10.8 to assess fire danger. The Natural Breaks Jenks Distribution technique was applied, which provides a more balanced distribution of values between classes, reducing contrast within groups and increasing contrast between the groups in Table 1.

Density may be dangerous if the NDVI value falls between 0.538 and 0.947. Higher values indicate thicker vegetation. Typically, standard NDVI values range from −1 to +1. Table 1 indicates that the case study’s actual NDVI values range from −0.188 to +0.947.

Based on vegetation type, density, and moisture, the Fuel Danger Index (FDI) was utilized in a study area to determine the fire risk. Other research used similar techniques to examine the relationship between FDI and measurements like NDVI and NDMI, which measure vegetation. The Forest Fire Danger Rating System (FDRS) in Australia is one index that this technique corresponds with research that has been employed (Ibrahim et al., 2024). The research in the Vellore district further distinguished itself from earlier studies by providing comprehensive data on vegetation density and fire vulnerability, emphasizing grassland’s important role in fire events.

Conversely, a different study examined the coniferous kind, which has a higher resinous content and is more likely to cause wildfires in the Mediterranean region. For instance, the study demonstrated that pine woods are prone to wildfires (Sivrikaya and Küçük, 2022).

In contrast, the conventional NDMI used a weighting system of −1 to +1; a lower weight indicates drier and more flammable vegetation. Fire risk was partitioned into five classes according to moisture content by the Raster Calculator, then Map Algebra from Spatial Analyst in ArcGIS 10.8, which gives a value of five to very dry and one to exceptionally wet. Table 1 shows that the actual NDMI values in the study area range from 0.4 to −0.42, which were used to classify the fire risk into five bands. Wetness of the vegetation is deemed hazardous when the NDMI value is less than 0.0097; a value of five denotes arid vegetation, and a value of one denotes extremely wet vegetation is considered hazardous if its moisture content is less than 0.0097.

Standard FDI values are contingent upon the studied region’s particular vegetation type, density, and moisture content. About Table 4, the research region’s actual FDI value rates range from −0.119 to 1.15. The high and extremely high-threat areas are described in Figure 5A. Vellore lost 80 ha of its humid primary forest between 2002 and 2023, accounting for 24% of the district’s overall tree cover loss. During this time, Vellore’s humid primary forest’s total area shrank by 0.56%. As mentioned earlier, Figure 4A illustrates how the high vegetation density in the east contributed to the rise in the fuel hazard index value.

The research also highlighted that the areas with high density are very high-risk areas. This paper emphasizes the importance of considering regional situations when estimating fire risk.

Human activities vary in the study area, where the roads and settlements are widely spread through the forests in the research area, facilitating the population to visit the forests for activities involving rangeland use, hunting, and woodcutting. Since these parameters are considered average values, it is essential to note that they also raise the incidence of fire, gunfire, and other mishaps. The present study classified the ADIr into five categories using the Euclidean Distance tool from ArcGIS 10.8. A multi-buffer analysis was performed to categorize the values into five risk classes utilizing the Natural Break (Jenks) distribution. As demonstrated in (Table 1), a value of 0.037–0.063 is regarded as very low risk, and a value of 0–0.0052 increases the danger level. As indicated in (Table 4), the simulated human danger index (ADI) rates are from 12,000 to 0 in Figure 4B. A multi-buffer analysis was produced to classify the values into five risk classes utilizing the Natural Break (Jenks) distribution. As illustrated in (Table 1), a value of 6,590–11,700 is regarded as very low risk, and a value of 0–1,290 expands the risk level.

Since ADI is one of the fire indexes used in this research paper, it rates fire risk based on human activities proximate to the roads and settlements. This approach is similar to related studies that used GIS-based analytical network process (ANP) simulations to calculate wildfire risk regions (Mangiameli et al., 2021).

Furthermore, the study in the Vellore district provides more detailed data on specific human activities related to roads and populated areas and their effects on fire danger. The present work depends on GIS modeling to evaluate forest fire risk regions compared to related studies that applied different approaches. It also underlines the importance of the vicinity of roads and settlements to the forests in fire risk computing and prevention, an essential factor in wildfire risk modeling.

Conversely, to related surveys that utilized the same method, one relevant research conducted by Gheshlaghi discussed the influence of closeness to roads and populated areas on fire events, which is a crucial factor in forest fire danger modeling and impacts the possibility of fire incidents and protection. In the study conducted in Noshahr Forests, North Iran, the author applied the GIS-based analytical network process (ANP) model to assist forest fire risk areas. Also, the author pointed out that ADI raises the forest fire risk due to human activities near the forests (Abedi Gheshlaghi, 2019). Due to ADI, researchers and agricultural department managers can identify and mitigate the fire damage caused by human presence and activities in forested areas.

The weather status in the area governs the vegetation type dominance there, as the area will be susceptible to fires in an area with a drier climate. Standard climatic rates are based on climate metrics such as temperature, relative humidity, and wind speed/direction.

Weather Danger Index (WDI) simulation was generated by ArcGIS 10.8, starting with Multidimensional tools and then using “IDW” from Interpolation in Spatial Analyst Tool, which classifies the temperature index into five classes by applying Natural Breaks (Jenks) from very low danger up to very high risk. On the other hand, “IDW” was applied to categorize wind speed and direction and put it into five classes relying on Natural Breaks (Jenks). The same division was used for relative humidity, producing five categories from very low to very high danger. The final output for WDI was created utilizing the Raster Calculator in ArcGIS 10.8. WDI was classified into five classes, from a very low danger fire index to a very high fire danger index using Natural Breaks (Jenks), as shown in (Table 1) (Rabiei et al., 2022).

(Table 4) explains absolute weights for previous elements in the study area. In turn, and shows WDI categories in five levels of risk based on the simulated weather danger index (WDI) values, which ranged from 338 to 348 (Figure 5A). 345–348 refers to very high risk, while 338–341 is very low. The fire season, which lasts from May to July, leads to hydric pressure in the woods, which causes a cumulative proportion of dry organic substances, increasing the fuel for fire ignition.

The present study used weather indexes like temperature, relative humidity, precipitation, wind speed, and direction to estimate fire risk and behavior. It provided more detailed information on the climatic metrics and their impacts on fire risk. This paper agrees with relevant studies that studied the effects of climate change on wildfire risk modeling and protection. This research work also meets with other studies that have assured that meteorological factors directly impact fire danger and behavior by influencing fuel moisture content and combustion capability.

Moreover, dryness enhances fire risk by drying out vegetation and creating a hospitable environment for fire inflammation. Vellore’s study output differs from other regions, such as the Mediterranean, with different weather and natural situations (Ibrahim et al., 2024). This comparison highlights the importance of regional variations in fire risk evaluation.

In the research area, the terrain is crucial to spreading fire since steep slopes facilitate the fire’s rapid expansion and eventual annihilation of all vegetation species. Fire travels upward fastest on slopes. In this study, the software was used for slope categorizing into five classes and applying Natural Breaks (Jenks) distribution due to their impacts on the fire outspread in the study region, starting from the very low risk with one value up to very high risk, with five values in (Table 1). The present study has classified the aspect into four categories by applying Natural Break (Jenks) distribution. So, the southern aspect was classified as high ignition risk, and the northern aspect had the lowest rate, so it was classified as very low risk (Table 1). An elevation map was simulated using the Natural Breaks (Jenks) distribution, which was sorted into five classes with rating values between 1,340 and 55, where a rate of five was given to the highest-risk group and one to the very low-risk class (Table 1).

In the topography of current work, the slope is the leading cause of fire spread; aspect and elevation are not as dangerous as the slope. There is an increased risk of a fire spreading because most of the study area’s forests are dispersed over sloping terrain. As with the WDI index, the topographic danger index TDI value, which ranges from one to 5, provides the basis for five risk categories (Table 4).

The methodology used in this research marked the significance of topographic metrics, such as slope, aspect, and elevation, in fire risk mapping Figure 5B. This paper is consistent with other studies that have found that topography is crucial in wildfire behavior. The study used an integration of topographic factors to evaluate fire danger; it also highlighted the value of topography in spreading fire, a central parameter in wildfire danger simulation. The Vellore study found very high topographic risk values between 934 and 1,690, while medium risk values between 56.5 and 1,690 (Table 4).

Compared to a study conducted in Menderes, Turkey, the results indicated that topography plays a significant role in determining the behavior of wildfires. Turkish forests with a higher slope tend to spread fires more quickly, especially those containing Black and Calabrian pine species, with higher canopy closure rates than 71%. When determining the fire risk, the south aspect—which faces portions of Turkish forests—is essential since regions facing south are more prone to fires. The comparison with related studies emphasizes the importance of considering topography when evaluating a fire’s risk and how good fire behavior forecasting requires more precise data (Çolak and Sunar, 2020).

To summarize, although this study established that slope is the primary topographic component influencing the spread of fires, previous research has stressed the significance of aspect and elevation in determining the risk of fires and the extent of burn damage.

The present study used remote sensing data and a GIS-based model to estimate fire severity. The research also noted the value of various land cover types in fire seriousness simulation, a fundamental parameter in wildfire risk calculation and prevention. Compared to related research, the study utilized dNBR to compute fire severity. A similar method was used in other studies that used a remote sensing approach to evaluate fire severity. Foremost, NBR pre- and post-fire and dNBR were measured for data analysis. Afterward, the results of the indexes were various under pixel values, and the regions that formed the fire danger were defined. Relying on the fire severity model regarding the dNBR index explained that not all areas are endangered by forest fires (Mamgain et al., 2022). According to the NBR simulation performed during the pre-fire period, populated areas, highways, water bodies, and forest areas all have reflection parameters.

NBR rates between −1 and +1, where the weight closer to value + 1 indicates the vegetation area (unburned area), and the value closer to −1 points out the scorched region. The factor was modeled by applying the raster calculator in ArcGIS 10.8 and using the Natural Breaks (Jenks) distribution. It was sorted into five groups: pre-fire and post-fire, as shown in (Table 2).

By calculating the fire severity estimation using dNBR, the study presents a complete understanding of the forest fire patterns in the Vellore district. Combining these different metrics allowed the complete modeling of fire risk and sensitivity in the study area (Figure 6). So, a standard rate for NBR ranges between −1 and +1, while the dNBR ranges from −2 to +2 (Genç et al., 2023). The dNBR usually is scaled by 103, which rates between +59.43 and +754.7, indicating high severity fire (Table 3). The current results demonstrate the pre-NBR values between −0.661 and +0.545; on the other hand, post-NBR values between −0.442 and +0.542 (Table 2). Meanwhile, dNBR ranges from −1 to +754.7 (Table 4).

This was approved with one study in Kastamonu, Turkey, that described that dNBR is a valuable element for identifying fire severity and produced a result that the dNBR index was used to compute the unburned region after a forest fire in 2020. The study site exposed enhanced regrowth, unburned, low severity, moderate severity, and high severity classes based on dNBR values, indicating variable levels of damage (Genç et al., 2023).

The paper meets the findings and approaches of a pertinent study on using dNBR to evaluate fire severity. Examining the impact of different land cover types on interpreting the fire severity indices derived from remote sensing makes the comparison crucial. A variety of topics were examined in this paper and related research, including the effects of fire on vegetation cover, the use of fire science, NDVI computing, the function of science in land management, and wildfire-exposing techniques in diverse environments (Mahfoud and Ali, 2017; Mahfoud, 2020). The researchers emphasized the importance of data quality in a fire study and the usefulness of data processing techniques, such as pre-treatment satellite data, for precise computations in NDVI simulation. Combining GIS-based modeling and remote sensing improves the study’s enforcement across various forest types and landscapes by thoroughly understanding fire behavior and danger parameters. The papers examined the advanced use of technologies such as remote sensing and GIS-based modeling to map burned regions, forecast fire danger, and study fire behavior.

The study’s authors stress a more cogent approach to fire management, prudent management of natural resources, and the promotion of effective fire danger mitigation techniques; they also emphasize the ecosystem impact of fires on vegetation strain, soil degradation, and changes in land cover and use (Sivrikaya and Küçük, 2022; Sandal Erzurumlu and Yıldız, 2024). GIS and RS technologies were used to determine fire risk categories, calculate fire hazard factors, and create fire severity maps to develop effective management strategies. Concentrating on fuel administration in areas with high vegetation species reduces the fire risk and spread. Also, acknowledging that human activity plays a crucial role in initiating fires is imperative. Simultaneously, to prevent wildfires, it is imperative to keep an eye on geographic features and weather patterns, especially during dry seasons. In addition to academic institutes, collaboration with forestry and environmental authorities is crucial for gathering field and wildfire data that will provide a thorough understanding of forest fire dynamics and optimal reduction strategies.