Dinder Natural Park (DNP) is one of the most important natural reserves in Sudan and one of the oldest in Africa, having been established in 1935. Dinder has been included in the UNESCO list of biosphere reserves since 1987 ( Figure 1 ). It covers an area of 10,197 km² ( MPDNP, 2004 ).

The Dinder and Rahad are the two rivers that flow through the DNP region in the northwest direction, and the banks of the two rivers consist of a group of lakes or back swamps (i.e., Mayas) ( Abdel Hameed et al., 1997 ). The soil in (DNP) consists of two types: vertisols and entisols ( Holsworth, 1968 ; Dasmann, 1972 ). The climate of the park is characterized by two seasons: the hot and humid rainy season (May-November) and the cool and dry season (December-April). The northeastern part of the park has the least rainfall (400 - 800 mm), which gradually increases with distance towards the southeast of the park (800 - 1000 mm) ( MPDNP, 2004 ). The vegetation in the park consists of three ecosystems: the A. seyal-Balanites aegyptiaca ecosystem, the Mayas ecosystem, and the riverine ecosystem ( Abdel Hameed et al., 1996 ). The Mayas wetlands are the most unique feature and one of its three major ecosystems. “Mayas” is a local name for floodplain wetlands that are found on both sides of the Dinder and Rahad rivers. More than 40 Mayas are part of the river Dinder and Rahad ecosystems inside the DNP ( Hassaballah et al., 2016 ).

Remote sensing (RS) is defined as “the science and art of obtaining information about an object, area, or phenomenon through the analysis of data acquired by a device that is not in contact with the object, area, or phenomenon under investigation” ( Lillesand & Kiefer, 1979 ).

The present study employed the image processing technique that was used to detect burned scars. Mapping is based on analysis of the remote sensing data time series of multispectral satellite imagery from the National Aeronautics and Space Administration (NASA) archive by Moderate Resolution Imaging Spectroradiometer (MODIS) data ( Atta, 2007 ; Elmardi, 2009 ; García et al., 2014 ). Finally, a fire frequency map was prepared by using fire maps from each year from 2010 to 2020.

The image processing technique that was used to classify fire scars in each individual MODIS scene consists of the following general steps:

1) The MODIS satellite images 8-day surface reflectance product of 250m resolution during the fire season was obtained freely from the web. The data were downloaded from https://search.earthdata.nasa.gov/ .

2) Subsetting the study area of the DNP.

3) Burned areas were detected for the years 2010-2020 based on MODIS surface reflectance data. Each of the resulting windowed images was then used to calculate a Normalized Difference Vegetation Index (NDVI) that is sensitive to the presence and abundance of green vegetation. The NDVI was formulated as follows:

(NIR band − RED band)/(NIR band + RED band)

NDVI was calculated to provide a basis for detecting burned areas, since burning removes vegetation, giving rise to a black layer of ash, and burned material also reduces NDVI. This makes it possible to detect burned vegetation, as pixels can be classified into ranges of burnt scar areas that take the lowest NDVI values, with NDVI values ranging from 1 to −1. The selection of bands is crucial for good visual interpretation, since burn areas present low reflectance in red and near-infrared. Burned areas were detected from the NDVI time series of Al Dinder by subtracting each earlier image from each later one, and then burning pixels were determined from the resulting difference image as those whose NDVI had reduced substantially between dates. The changes (burned area) between each of the two successive NDVI images were detected and saved in a new image.

4) The above detection of burned areas was achieved using threshold burned areas from each difference image. A colour density slide was created by using the NDVI difference values observed for the pixel at the minimum and maximum limits of the darkness range of the burned area. The region of burned area with NDVI difference values similar to those of burned areas was interactively removed (masked out) to create a new noise-free image, which helped produce a high-quality fire scar map.

5) The resulting burned area images were then converted from raster to vector to facilitate the preparation of burned area maps in ArcMap.

6) Each satellite MODIS scene and the burned areas shapefile were rectified to UTM.

Zone 36 Projection to be compatible with existing spatial data so that it can be clipped to the geographic boundaries of the federal properties.

7) Finally, all the burned areas detected between successive images were summed together to obtain one image containing all the burned areas of the years. These yearly images were then added together to produce a new image showing the number of times that each pixel within al Dinder was burned during the years 2010 to 2020 of the study.

Fire mapping accuracy was evaluated between a 2020 fire map image from MODIS satellite imagery at 250 m-pixel resolution and November 2020 Landsat satellite imagery at 30 m-pixel resolution with path 171 and row 51, showing a mapping accuracy of 70%.

A total of 9 samples were randomly obtained near roads and accessibility locations (3 samples from each fire frequency) ( Figure 2 ), evenly distributed across three categories of fire frequency: low (0 - 1 burn), moderate (2 - 5 burns), and high (6 - 11 burns), utilizing ArcGIS 10.3 software. Due to the country’s circumstances, we were unable to conduct a second planned fieldwork to cover two types, vertisols and entisols data, in DNP, and increase soil samples.

A comprehensive review of the study area for relevant literature and historical cartographic records to classify soils within the park was conducted. For accurate geospatial sampling, a Global Positioning System (GPS) receiver was used, and a Toshiba Satellite C855 laptop and a printer with documentation of Microsoft Office applications were used.

A soil paste was made by carefully adding distilled water to 250 g of air-dried soil (2 mm) until the paste satisfied the criteria for saturation.

The pH of saturated soil paste was determined by a glass electrode pH meter. This was done after calibrating the pH meter to pH 4.01 and 7.00 using buffer solutions. The pH was measured by immersing the electrode into the upper part of the soil paste, and the pH value was recorded ( Mclean, 1982 ).

Total nitrogen in soil was determined by the Kjeldahl digestion and distillation procedure following the method developed by ( Bremner and Mulvaney,1982 ). In brief, an exact amount of 1 g (finely crushed) soil was weighed into a Kjeldahl digestion flask, and a tablet of selenium and 5 ml of concentrated H₂SO₄ were added to the soil. The flask was placed on a Kjeldahl digestion apparatus, heated initially gently and later vigorously for at least 3 hours. The flask was removed after a clear mixture was obtained and then allowed to cool down. To dissolve the digested material, about 5 mL of distilled water was added to the digested material and transferred into a 100 mL distillation tube. Then, about 20 ml of 40% NaOH was added to the solution, and then distilled for the collection of ammonia. The digested material was distilled for 4 minutes, and the distillate was collected in a flask containing 20 ml of 2% boric acid (H₃BO₃) prepared with universal indicator (methyl red and bromocresol green) to produce approximately 75 ml of the distillate. In the end, the color changes from pink to blue. The concentration of ammonia was determined by back titration against 0.01 N HCl from a microburette. At the endpoint, when the solution changed from weak blue to pink, the volume of 0.02 N HCl used was recorded, and % N was calculated. Blank distillation and titration were also carried out to account for traces of nitrogen in the reagents and the water used.

Calculations:

% Total N in soil sample = (sample titration − Blank) × Normality × 14 × dilution⁄sample weight.

Soil organic carbon was determined by the modified Walkley-Black method ( Walkley & Black, 1934 ), as described by Nelson and Sommers (1982) . The procedure involves wet combustion of the organic matter with a mixture of potassium dichromate and concentrated sulfuric acid. After the reaction, the excess dichromate was titrated against ferrous ammonium sulfate. Approximately 1.0 g of air-dried soil was weighed into a clean and dry 250 ml Erlenmeyer flask. A reference sample and a blank were included. A 10 ml 0.1667 M potassium dichromate (K₂Cr₂O₇) solution was accurately dispensed into the flask using the custom laboratory dispenser. The flask was swirled gently so that the sample was made wet. Then, using an automatic pipette, 20 ml of concentrated sulfuric acid (H₂SO₄) was dispensed rapidly into the soil suspension and swirled vigorously for one minute, and allowed to stand on a porcelain sheet for about 30 minutes, after which 100 ml of distilled water was added and mixed well. Ten ml of orthophosphoric acid and 1 ml of diphenylamine indicator were added and titrated by adding 1.0 M ferrous ammonium sulfate from a burette until the solution turned dark green at the endpoint from an initial purple color. About 0.5 ml of 0.1667 M K₂Cr₂O₇ was added to restore excess K₂Cr₂O₇, and the titration was completed by adding FeSO₄ dropwise to attain a stable endpoint. The volume of FeSO₄ solution used was recorded, and % C calculated.

% C was calculated as follows:

$N\times 0.39\times \text{mcf}\times \left(V_1-V_2\right)/S$

where:

N = normality of ferrous sulphate solution.

V₁ = mL of ferrous ammonium sulphate solution used for blank.

V₂ = mL of ferrous ammonium sulfate solution used for the sample.

S = weight of air-dry sample in grams.

Mcf = moisture-correcting factor (100 × % moisture)/100.

0.39 = 3 × 0.001 × 100% × 1.3 (3 = equivalent weight of carbon).

1.3 = a compensation factor for the incomplete combustion of organic carbon.

The fire frequency map showed that the riverine areas and Maya areas are the least burned areas due to the humidity of these places ( Figure 3 ). This reduced fire activity is attributed to higher humidity levels near water sources. On the other hand, these conditions promote enhanced vegetation growth, which increases the potential for wildfires over time. Conversely, areas along the park’s borders show low fire frequency, likely due to overexploitation by local inhabitants, leading to insufficient biomass to sustain annual fires.

The most vulnerable areas are in the middle of the park because they contain large quantities of weeds. These maps could be used to determine priorities for areas to open fire lines in DNP.

The fire season normally starts in October and extends until December. During these months, the areas that burn inside the park differ from year to year ( Figure 4 ). This could be correlated with the annual precipitation, which triggers the availability of vegetation. This result is in agreement with Zhang et al . (2016) , who stated that the precipitation intensity and intermittency play an important role in the dynamics of woody vegetation cover and deep soil moisture. In arid and semiarid regions, as the annual precipitation increases, the rate of woody vegetation cover increases as a power-law function.

The statistical results showed that the maximum area burned in the park in 2011 was 12.8%, and the minimum area burned in 2016 was 4.1%. The results showed management activities were absent ( Figure 5 ). The wildfire in the DNP burned without any barrier, and it stopped only when it encountered a natural obstacle such as a drainage system ( Figure 6 ). A fire usually starts in the northeastern part of the park immediately after the rainy season because it is closer to the villages that surround the park, and some residents of the villages surrounding the park start fires for hunting and collecting honey ( Elmardi, 2009 ). Also, the north side is drier than the southern area because of the higher amount of rainfall.

The results ( Figure 7 ) showed that the different fire frequencies did not affect nitrogen accumulation in the soil, although the results also show that soil nitrogen decreased slightly with an increase in fire frequency. This result is acceptable because fires with high temperatures affect nitrogen fixation processes ( DeBano, 1990 ). The results ( Figure 8 ) & ( Figure 9 ) showed that the levels of organic carbon and soil pH decreased slightly with an increase in fire frequency. This finding aligns with a study on the effects of fire on soil properties in Fwangnin, Bokkos District, Nigeria ( Wuyep et al., 2024 ). These results are in agreement with Hosseini et al. (2017) , who stated that severe fires can generally remove organic carbon, and Mayor et al. (2016) , who mentioned that the trend of increasing fire recurrence in Southern Europe may result in losses or alterations of soil organic matter and early warning indicators for shifts in soil fertility in response to fire recurrence. Also, ash accumulation after fires often raises soil pH, especially in neutral soils, due to alkaline carbonates in the ash ( Zitta et al., 2022 ; Kemmitt et al., 2006 ). Changes in organic matter (OM) affect soil properties like electrical conductivity, pH, water retention, and carbon levels ( Fonseca et al., 2017 ).