Evaluating the Efficiency of Two Ecological Indices in Monitoring Forest Degradation in the Drylands of Sudan

Abstract

With increasing threats to forest resources, there is a growing demand for accurate, timely, and quantitative information on their status, trends, and sustainability. Satellite remote sensing provides an effective means of consistently monitoring large forest areas. Vegetation Indices (VIs) are commonly used to assess forest conditions, but their effectiveness remains a key question. This study aimed to assess and map forest degradation status and trends in Lagawa locality, West Kordofan State, Sudan using the soil adjusted and atmospheric resistant vegetation index (SARVI) to quantify the relationship between SARVI and the Normalized Difference Vegetation Index (NDVI) and compare the efficiency of both indices in detecting and monitoring changes in forest conditions. The study utilized four free cloud images (TM 1988, TM 1998, TM 2008, and OLI 2018), which were processed using Google Earth Engine (GEE) to derive the indices. The study found significant forest degradation over time, with 63% of the area categorized as moderately to severely degraded. A strong, positive relationship between SARVI and NDVI (R² = 0.9085, p < 0.001) was identified, indicating that both are effective in detecting forest changes. Both indices proved efficacy, cost-effectiveness, and applicable for monitoring forest changes across Sudan’s drylands. The study recommends applying similar methods in other dryland forests in other regions.

1. Introduction

According to the Food and Agriculture Organization (FAO), forests cover nearly one-third of the Earth’s land surface, but this area is shrinking, particularly in the tropics, despite efforts to halt deforestation and restore degraded lands. Additionally, the Global Forest Watch reports that in 2020, the world had 3.68 billion hectares of natural forest, extending over 28% of its land area, but it lost 23.9 million hectares of natural forest, equivalent to 14.7 gigatons of CO₂ emissions [1,2,3,4].

Forest ecosystems, especially in dryland regions, provide essential ecosystem services such as biodiversity conservation, carbon sequestration, and resources that support local communities [5,6]. In Sudan, forests play a vital role in the livelihoods of many, supplying goods like fuelwood, fodder, and building materials while maintaining critical ecological functions [6,7,8,9,10,11,12]. However, Sudan’s forest resources face immense pressure, primarily due to deforestation driven by agricultural expansion and increasing energy demands [6,7,11,13,14,15]. Forests are unevenly distributed across the country, with the majority found in the southern regions, while the north suffers from sparse and degraded vegetation [6,15,16]. Population growth in these areas has led to increased demand for forest products and services, exacerbating forest degradation [6,11,15,16,17].

The growing dependence on illegal timber harvesting by certain communities has further contributed to forest degradation. Many of these communities have limited awareness of the root causes of deforestation such as unsustainable land-use practices [15,18,19,20]. With a rising population and continued reliance on forest resources, conservation efforts are crucial to maintain ecological balance and protect local livelihoods [20,21,22]. Despite the importance of forests for Sudan’s ecological and economic stability, the country has experienced substantial forest cover loss in recent decades [6,23,24].

Ecologists and conservation programs have been working to develop reliable and timely metrics for monitoring forest ecosystems and degradation [23,25,26,27]. However, traditional forest monitoring methods often face significant challenges in developing countries like Sudan [28]. Forest inventories are often limited, covering only a small portion of the country’s forests [26,29,30,31]. This lack of consistent data on forest composition and condition makes it difficult to accurately estimate ecosystem services and plan effective conservation strategies [10,32,33]. Additionally, some forests are located in remote, inaccessible areas, making regular monitoring difficult [6,11,34,35]. The absence of a systematic monitoring program and appropriate tools for assessing forest health further complicates forest management efforts [26,30,35,36,37].

Up-to-date information on the distribution, composition, and abundance of tree species in Sudan’s forests is crucial for informed conservation planning [38,39]. Active, systematic monitoring using ecological indices can help address knowledge gaps. Such monitoring is essential for Sudan to meet its international commitments under the Convention on Biological Diversity (CBD) [5,6,40]. Remote sensing provides large-scale, up-to-date data on forest cover, composition, and changes over time, even in remote areas. Satellite imagery and aerial photography allow researchers to track deforestation, fragmentation, and other forms of degradation at a fraction of the cost and time of traditional field surveys [26,30,36,41,42,43]. GIS enables the spatial analysis of this data, offering insights into the distribution of forest resources, land-use impacts, and the relationship between environmental factors and forest health [26,36,44]. Together, these tools provide efficient, cost-effective monitoring, essential for improving conservation and forest management in regions like West Kordofan. This study aimed to (1) assess and map the status and trends of forest degradation in Lagawa locality, West Kordofan State, Sudan, using the Soil Adjusted Atmospheric Resistant Vegetation Index (SARVI); (2) quantify the relationship between SARVI and the Normalized Difference Vegetation Index (NDVI); and finally (3) compare the efficiency of both indices in detecting and monitoring changes in forest conditions. The findings are expected to provide valuable insights into enhancing sustainable forest management strategies.

2. Materials and Methods

2.1. Description of the Study Area

The study was conducted in Lagawa locality, located in West Kordofan State, Sudan. Lagawa lies within the savanna belt of the southern hemisphere tropics, characterized by a hot semi-arid climate. Temperatures range from an average maximum of 42 °C in summer to 25 °C in winter, with annual rainfall varying from 400 mm in the north to 800 mm in the south [45]. The locality is divided into three administrative units: Lagawa, Keilek, and El Sunut, and has a population of approximately 163,400. Its economy is largely based on traditional subsistence farming, rain-fed agriculture, and livestock rearing, with a mix of sedentary farming and transhumance systems [45]. The landscape is predominantly flat, composed of Nubian sandstones, granite, and green series rocks, with sandy, clay, and gardud soils [46]. The area supports two ecological zones, transitioning from thorny Acacia trees in the drier north to savannah woodland species such as Balanites aegyptiaca and Acacia seyal in the south [45,47]. Eight forests within the Lagawa Forest Circle were selected as study sites, located between latitudes 10°20′N and 11°20′N, and longitudes 28°00′E and 30°00′E (Figure 1).

2.2. Data Collection and Analysis

The study employed multiple methods for collecting, processing, and analyzing data. These methods included remote sensing, ground-truthing surveys, field observations, and interviews with forest experts working in the studied areas. Four satellite images were selected from Landsat 5 Thematic Mapper (TM) and Landsat 8 Operational Land Imager/Thermal Infrared Sensor (OLI/TIRS) for the years 1988, 1998, 2008, and 2018, as shown in

Table 1. Landsat 5 TM and OLI 8 data used for forest cover analysis.

Satellite/Sensor	Acquisition Date	Spectral Bands	Resolution (m)
Landsat 5 TM	28 January 1988	1, 2, 3, 4	30
Landsat 5 TM	26 January 1998	1, 2, 3, 4	30
Landsat 5 TM	15 January 2008	1, 2, 3, 4	30
OLI 8	01 January 2018	2, 3, 4, 5	30, 15

. The selection was based on the availability of free satellite data with less than 10% cloud cover. Images from 1991 and 2001 were used to assess changes in forest cover. Additionally, a structured survey was conducted to gather insights from forest experts on the threats and causes of forest degradation (Appendix A). These interviews aimed to evaluate the presence, severity, and underlying drivers of degradation affecting the forests within the study area.

First, the coordinates for each forest in the study area were specified, and the forest boundary was delineated and reviewed using the Google Earth Pro platform. Subsequently, images of the study area were downloaded from the free website (USGS, http://earthexplorer.usgs.gov and Google Earth Engine (GEE) (accessed on 15 November 2024)). These images were acquired during the dry season in January to create training data and perform classification. The dry period was ideal for obtaining cloud-free satellite images, which facilitated the differentiation of various land cover classes.

Second, image preprocessing, which included geometric and atmospheric corrections, was performed on all acquired images using geospatial analysis software (ArcGIS Pro version 3.4). The bands from the four image scenes were downloaded and saved as separate TIFF files. The individual bands were then combined sequentially: Bands 1 to 4 for TM 5 and Bands 2 to 5 for OLI 8, through the creation of a virtual raster. A false-color composite was generated for display purposes. Finally, a subset was created from the virtual raster and clipped to encompass the full extent of the study area, which was used to develop the training dataset for image classification. All images were projected using the WGS 84 Projection Coordinate System in the Universal Transverse Mercator (UTM) Zone 35N.

Third, unsupervised classification applied using the Natural Breaks (Jenks) method to categorize the images enabled effective distinction between different land-use classes in the study area while minimizing mixed pixels across classes. Additionally, SARVI thresholds (Table S1) were applied using the Natural Breaks (Jenks) method to classify vegetation cover effectively. This process identified four land types: Dense Forest (DF), Moderate Dense Forest (MDF), Low Dense Forest (LDF), and Semi-bare land/bare land (SBL/BL), with Semi-bare land present in some forests and bare land in others. Details on the characterization of these land types can be found in Table S2.

Fourth, different approaches for image processing were employed to derive the SARVI and NDVI indices for assessing forest degradation. These approaches included visual image interpretation and the biophysical indices method. Visual image interpretation was conducted before the field survey to analyze the study area’s features, identify targets, and plan the survey by examining variations in pattern, shape, texture, association, and size. The biophysical indices method utilized vegetation indices (VI) and ground truth data to improve the accuracy of forest degradation assessments, particularly in dryland regions. SARVI was selected for its effectiveness in evaluating vegetation degradation in semi-arid zones, outperforming other indices like NDVI. Meanwhile, NDVI was used to assess forest health conditions, equations and characteristics of two indices detailed in Table 2 and Table 3.

Fifth, the ground truth survey was conducted in the summer of 2018 to validate the land cover classes and accurately assess the classification results. A random sample of ground truth points was generated in ArcGIS Pro using a stratified random technique, with the following sample sizes for each forest: Alban Gadeid (90 points), Shingil Shariq (130 points), Alghara (160 points), Shag Addoban (180 points), Shingil Gharib (190 points), Ladi (190 points), Alkoua (200 points), and Mehaila (200 points). These samples were divided by class according to the area of each class to ensure accurate representation. The data was then exported to Google Earth Pro for visual interpretation. All sample points were entered into a Global Positioning System (GPS) for validation during the field survey.

Sixth, an accuracy assessment of the land cover maps was performed. This assessment involved calculating User’s Accuracy (UA), Producer’s Accuracy (PA), Overall Accuracy (OA), and Kappa coefficients (Tables S3–S6).

Seventh, the post-classification comparison method (PCC) was adopted to detect changes in forest cover within the studied areas. This method calculates the magnitude and percentage of change between different classes for selected forests over the study period. We chose these forests specifically because they are more severely degraded than others, allowing us to examine their past conditions and identify the causes that led to the current situation. An overview of all the research methodology steps used in this study is presented in Figure 2.

Finally, for data analysis, the study employed Google Earth Engine (GEE), a powerful cloud-based geospatial platform, to access, preprocess, and analyze multi-temporal satellite imagery [48,49]. GEE facilitated the calculation of vegetation indices, notably the soil adjusted and atmospheric resistant vegetation index (SARVI) and Normalized Difference Vegetation Index (NDVI), enabling efficient large-scale data processing. ArcGIS Pro was subsequently used to classify vegetation cover based on SARVI thresholds, as outlined in Table S1, applying the Natural Breaks (Jenks) classification method to effectively delineate vegetation classes. Microsoft Excel 2019 was utilized to organize and statistically analyze the satellite-derived data, as well as to generate visual representations. Additionally, Excel supported the analysis of structured survey data used to assess threat levels in the selected forest areas. Threats were quantified on a normalized scale ranging from 0.0 to 1.0, calculated as the ratio of the total observed threat score to the maximum possible score. A score of 0.0 indicated no observed threats, whereas a score of 1.0 represented the highest recorded threat level.

Table 2. Vegetation indices: definitions, formulas, and sources.

Vegetation Indexes	Abbreviation	Formula	Reference
Normalized Difference Vegetation Index	NDVI	${\displaystyle \frac{NIR-R}{NIR+R}}$	[50,51]
Soil adjusted and atmospheric resistant vegetation index	SARVI	${\displaystyle \frac{NIR-RB}{NIR+RB+L}}\ast \left(1+L\right)$ RB = R − γ (B − R), L = 0.5, γ = 1.0	[52,53]

Where NIR = near-infrared wavelength; R = red wavelength; B = blue wavelength; L = soil adjustment factor (typically 0.5); γ = aerosol coefficient (typically 1.0); RB = red wavelength × blue wavelength.

Table 2. Vegetation indices: definitions, formulas, and sources.

Vegetation Indexes	Abbreviation	Formula	Reference
Normalized Difference Vegetation Index	NDVI	${\displaystyle \frac{NIR-R}{NIR+R}}$	[50,51]
Soil adjusted and atmospheric resistant vegetation index	SARVI	${\displaystyle \frac{NIR-RB}{NIR+RB+L}}\ast \left(1+L\right)$ RB = R − γ (B − R), L = 0.5, γ = 1.0	[52,53]

Where NIR = near-infrared wavelength; R = red wavelength; B = blue wavelength; L = soil adjustment factor (typically 0.5); γ = aerosol coefficient (typically 1.0); RB = red wavelength × blue wavelength.

Table 3. Characteristics of the vegetation indices used in this research.

Indices	Characteristics
NDVI	⮚ As a simple transformation of spectral bands, ⮚ NDVI is easily computed without assumptions regarding land cover classes, soil type, or climatic condition. ⮚ Long-time series (more than 20 years) available.
SARVI	⮚ Combines ARVI with SAVI (the constant B is normally 1 but can be varied to correct for aerosol, e.g., 0.5 for Sahel dust) ⮚ Ability to discriminate different types of vegetation areas. ⮚ It can be used as a simple tool for estimating the vegetation maps with no need for ground truth mission (or high precision images). ⮚ Proved to have a better separability of vegetation classes. ⮚ Used to reduce aerosol atmosphere and soil noise.

Table 3. Characteristics of the vegetation indices used in this research.

Indices	Characteristics
NDVI	⮚ As a simple transformation of spectral bands, ⮚ NDVI is easily computed without assumptions regarding land cover classes, soil type, or climatic condition. ⮚ Long-time series (more than 20 years) available.
SARVI	⮚ Combines ARVI with SAVI (the constant B is normally 1 but can be varied to correct for aerosol, e.g., 0.5 for Sahel dust) ⮚ Ability to discriminate different types of vegetation areas. ⮚ It can be used as a simple tool for estimating the vegetation maps with no need for ground truth mission (or high precision images). ⮚ Proved to have a better separability of vegetation classes. ⮚ Used to reduce aerosol atmosphere and soil noise.

3. Results

3.1. Mapping the Distribution of Forest Cover in 2018

The 2018 OLI classified maps revealed alarming trends in forest cover across all sites. In the Alban Gadeid forest, dense forest constituted only 8.14% of the area, while low-dense forest dominated with 37.44%, and semi-bare land accounted for 25.57%. Combined, low-dense, and semi-bare land represented 63.01%, indicating significant degradation. Similarly, in Alghara forest, low dense forest was the most prevalent at 37.47%, with bare land adding 23.72%, cumulatively making up 61.19% of the area. In the Alkoua forest, low-dense and bare land together covered 52.88%, with low dense forest alone at 36.22%. Ladi forest showed a similar pattern, with low dense forest at 34.69% and bare land at 21.55%, cumulatively 56.24%. Mehaila forest had 35.52% low dense forest and 21.61% semi-bare land, totaling 57.13%. In the Shag Addoban forest, the situation was worse, with low dense forest at 42.95% and bare land at 17.06%, jointly comprising 60.01% of the area. In contrast, the Shingil Shariq forest exhibited a relatively stable condition: moderate dense forest dominated at 33.25%, and the combination of low dense and semi-bare land was 49.94%, balanced by 50.06% dense and moderate cover. Finally, the Shingil Gharib forest reflected poor conditions, with 36.22% low dense forest and 16.66% bare land, totaling 52.88%. Overall, most forests were heavily dominated by degraded covers (low dense and bare land), with cumulative percentages consistently exceeding 50%, indicating critical forest health decline across the study sites (Figure 3 and

Table 4. Distribution of forest cover area (ha) and percentage in the study area in 2018.

Forest name	Dense (ha)	%	M. Dense (ha)	%	Low Dense (ha)	%	Bareland (ha)	%	Semi-Bareland (ha)	%
Alban Gadeid	6.48	8.14	22.95	28.85	29.79	37.44	-	-	20.34	25.57
Alghara	36	10.42	98.1	28.39	129.51	37.47	81.99	23.72	-	-
Alkoua	151.56	15.26	316.44	31.86	359.73	36.22	165.42	16.66	-	-
Ladi	29.52	13.26	67.86	30.49	77.22	34.69	47.97	21.55	-	-
Mehaila	41.67	14.44	82.08	28.44	102.51	35.52	-	-	62.37	21.61
Shag Addoban	60.75	9.52	194.49	30.47	274.14	42.95	108.9	17.06	-	-
Shingil Shariq	12.06	16.81	23.85	33.25	22.05	30.74	-	-	13.77	19.2
Shingil Gharib	14.04	3.26	85.95	19.93	179.01	41.51	-	-	152.28	35.31

).

3.2. Spatiotemporal Trend of Forest Cover Degradation for Selected Forests

The spatiotemporal analysis of the Shingil Shariq Forest showed that dense cover initially decreased from 15.06% in 1988 to 9.41% in 1998, then increased to 16.81% by 2018. Moderate dense cover consistently rose from 27.35% to 33.25%. Low dense cover also increased from 28.23% to 30.74%, while semi-bare land decreased significantly from 29.39% to 19.2% over the 30 years (Figure 4 and

Table 5. Spatio-temporal trends of forest cover degradation from 1988 to 2018 in selected forests.

Shingil Gharib Forest
Class Name	1988		1998		2008		2018
	Area (ha)	%	Area (ha)	%	Area (ha)	%	Area (ha)	%
Dense	90.09	20.89	70.74	16.40	92.34	21.41	14.04	3.26
Moderate Dense	151.65	35.16	111.10	25.75	185.58	43.03	85.95	19.93
Low Dense	134.10	31.09	149.90	34.77	118.08	27.38	197	41.51
Semi-Bare Land	55.44	12.85	99.54	23.08	35.28	8.18	152.30	35.31
Ladi Forest
Dense	36.09	16.22	23.58	10.59	46.44	20.87	29.52	13.26
Moderate Dense	61.83	27.78	70.56	31.70	65.70	29.52	67.86	30.49
Low Dense	77.67	34.90	81.63	36.68	63.99	78.75	77.22	34.69
Bare Land	46.98	21.11	46.80	21.03	46.44	20.87	47.97	21.55
Shingil Shariq Forest
Dense	10.8	15.06	6.75	9.41	8.82	12.30	12.06	16.81
Moderate Dense	19.62	27.35	23.04	32.12	19.26	26.85	23.85	33.25
Low Dense	20.25	28.23	25.20	35.13	25.74	35.88	22.05	30.74
Semi-Bare Land	21.06	29.36	16.74	23.34	17.91	24.97	13.77	19.20

). This suggests a positive trend toward forest regeneration and improved vegetation density.

Analysis of the Ladi forest revealed notable fluctuations in forest cover over three decades. Dense forest cover declined from 16.22% in 1988 to 13.26% in 2018, reflecting an overall degradation trend. Moderate dense cover increased from 27.78% to 30.49%, while low dense cover showed a slight decrease from 34.9% to 34.69%. Bareland slightly increased from 21.11% to 21.55% during the same period (Figure 4, Table 5).

Conversely, the Shingil Gharib forest exhibited a sharp decline in dense cover, from 20.89% in 1988 to only 3.26% in 2018. Moderate dense cover similarly dropped from 35.16% to 19.93%. Meanwhile, low dense cover rose from 31.09% to 41.51%, and semi-bare land expanded dramatically from 12.85% to 35.31% (Figure 4, Table 5). These trends indicate significant forest degradation and an increase in sparsely vegetated or degraded land.

Overall, while Shingil Shariq showed signs of recovery, the Ladi forest experienced slight degradation, and Shingil Gharib faced substantial forest cover loss.

3.3. Trend of Forest Condition During Study Time

The SARVI analysis revealed notable changes in forest cover across the three forests between 1988 and 2018. In the Ladi forest, dense forest cover decreased by 5.62% from 1988 to 1998, increased by 10.27% by 2008, and then decreased by 7.60% by 2018. Moderate dense forest cover fluctuated, rising by 3.92% (1988–1998), dropping by 2.18% (1998–2008), and slightly increasing by 0.97% (2008–2018). Low forest cover showed minor increases and decreases, while bare land remained relatively stable with a slight increase of 0.69% between 2008 and 2018 (

Table 6. Trends in forest cover condition from 1988 to 2018 in selected forests.

Shingil Gharib Forest
Class Name	1988–1998		1998–2008		2008–2018
	Area (ha)	%	Area (ha)	%	Area (ha)	%
Dense Forest (DF)	−19.35	−4.49	+21.6	+5.01	−78.3	−18.16
Moderate Dense Forest (MDF)	−40.59	−9.41	−74.52	−17.28	−99.63	−23.10
Low Dense Forest (LDF)	+15.84	+3.67	−31.86	−7.39	−60.93	−14.13
Semi-Bare Land (SBL)	+44.1	+10.23	−64.26	−14.90	+117	+27.13
Ladi Forest
Dense Forest (DF)	−12.51	−5.65	+22.86	+10.27	−16.92	−7.60
Moderate Dense Forest (MDF)	+8.73	+3.92	−4.86	−2.18	+2.16	+0.97
Low Dense Forest (LDF)	+3.96	+1.78	−17.64	−7.93	+13.23	+5.94
Bare Land (BL)	−0.18	−0.08	−0.36	−0.16	+1.53	+0.69
Shingil Shariq Forest
Dense Forest (DF)	−4.05	−5.65	+2.07	+2.89	+3.24	+4.52
Moderate Dense Forest (MDF)	+3.42	+4.77	−3.78	−5.27	+4.59	+6.40
Low Dense Forest (LDF)	+4.95	+6.90	+0.54	+0.75	−3.69	−5.14
Semi-Bare Land (SBL)	−4.32	−6.02	+1.17	+1.63	−4.14	−5.77

).

In the Shingil Shariq forest, dense cover dropped by 5.65% from 1988 to 1998 but recovered with increases of 2.89% and 4.52% in the following decades. Moderate dense cover grew by 4.77% (1988–1998), declined by 5.27% (1998–2008), and then increased by 6.40% (2008–2018). Low dense cover expanded until 2008 but dropped by 5.14% thereafter, while semi-bare land generally decreased over the period (Table 6).

For the Shingil Gharib forest, dense cover declined sharply, especially between 2008 and 2018, losing 18.16%. Moderate dense cover steadily declined throughout, while low dense cover initially increased but later decreased. Semi-bare land fluctuated, with a substantial increase of 27.13% between 2008 and 2018, indicating significant forest degradation (Table 6).

3.4. The Relationship Between SARVI and NDVI

From a correlation test between the recent SARVI and NDVI, the study found strong correlation (p < 0.0001 and R² = 0.9085), as shown in Figure 5, which illustrates that this very good correlation between SARVI and NDVI can facilitate monitoring change in forest degradation in these forests with these indices values.

3.5. Comparison Between SARVI and NDVI in Different Aspects

Table 7. Comparison of SARVI and NDVI in different aspects.

No.	Aspect	SARVI	NDVI
1	Technical effort	More complex to compute due to additional corrections (blue band use); requires higher processing effort.	Simple calculation with fewer bands; computationally less demanding.
2	RS data required	Requires satellite sensors with a blue band and preferably medium to high spatial resolution; generally no ground truthing needed.	Works with all sensors providing red and near-infrared bands; often requires ground truthing for accuracy.
3	Sensors Compatibility	Limited to sensors that include the blue spectral band (e.g., Landsat, Sentinel-2).	Compatible with a wide range of satellite sensors lacking blue band (e.g., MODIS, AVHRR).
4	Accuracy	More accurate in areas with dense or sparse vegetation due to correction for soil background and atmospheric effects; better discrimination of vegetation types.	Performs well in dense, broadleaf vegetation but less accurate with sparse vegetation due to soil background influence and additive noise; limited ability to distinguish vegetation types.
5	Cost and Time	Cost-effective since many satellite images are freely available; however, more processing time is needed due to complexity.	Cost-effective and faster to process due to simpler computation.
6	Application and multidimensional	Effective for vegetation mapping without ground truth, aerosol and soil noise reduction, and monitoring dryland forest degradation with sparse vegetation.	Widely used for monitoring vegetation health, estimating crop yields, pasture performance, and rangeland carrying capacity; limited in correcting soil and atmospheric noise.

compares the SARVI and NDVI vegetation indices, outlining their respective advantages and limitations. SARVI provides improved accuracy in detecting vegetation across both dense and sparse areas by effectively correcting for soil background and atmospheric effects, making it especially valuable in dryland and heterogeneous landscapes without the need for extensive ground truthing. However, SARVI’s reliance on sensors with a blue spectral band and its more complex computational requirements limit its broader applicability and increase processing time. Additionally, while SARVI is designed to resist soil and atmospheric interference, it can still be affected by highly reflective soils and adverse atmospheric conditions such as heavy cloud cover or aerosols. Its sensitivity also varies with vegetation type and density, sometimes showing reduced responsiveness to changes in sparse vegetation compared to NDVI. Moreover, SARVI’s performance can differ across biomes and climatic zones, often necessitating region-specific calibration. In densely forested areas, it may experience saturation effects similar to NDVI, limiting its ability to detect subtle canopy variations. In contrast, NDVI is simpler and faster to compute, compatible with a wider range of satellite sensors, and widely used for general vegetation monitoring, crop yield estimation, and pasture assessment. However, NDVI’s sensitivity to soil reflectance, atmospheric interference, and saturation in dense vegetation reduces its precision, especially in sparse or mixed land cover environments. Recognizing these trade-offs is essential for selecting the most appropriate vegetation index for specific ecological and remote sensing applications.

3.6. Comparision and Temporal Trends of SARVI and NDVI in Selected Forest Sites (1988–2018)

Table 8. Comparison of SARVI and NDVI values for selected forests in 2018 and their changes from 1988 to 2018.

No.	Forest Name	2018		1988–2018
		SARVI	NDVI	SARVI	NDVI
1	Alban Gadeid	0.048	0.118	−0.136	0.074
2	Alghara	0.052	0.127	−0.130	0.093
3	Alkoua	0.057	0.132	−0.138	0.090
4	Ladi	0.042	0.125	−0.141	0.099
5	Mehaila	0.045	0.119	−0.148	0.079
6	Shag Addoban	0.052	0.143	−0.113	0.091
7	Shingil Shariq	0.047	0.139	−0.109	0.109
8	Shingil Gharib	0.062	0.128	−0.162	0.086

compares SARVI and NDVI values for eight forest sites between 1988 and 2018, revealing an interesting pattern: while NDVI values have generally increased, indicating improved vegetation greenness, SARVI values have consistently decreased, suggesting an overall decline in vegetation health and density when accounting for soil and atmospheric influences. This discrepancy can be explained by the inherent differences in how the two indices are computed and what they measure. As shown in the technical comparison table, NDVI is a simpler, widely used index that primarily relies on red and near-infrared bands, making it less sensitive to soil background and atmospheric effects. This means NDVI may overestimate vegetation conditions in areas with sparse or mixed vegetation, especially without extensive ground truthing. In contrast, SARVI incorporates the blue spectral band to correct for these effects, offering more accurate vegetation assessments in both dense and sparse vegetation types.

Over time, forest areas might exhibit increased surface greenness, due to seasonal changes, understory growth, or replanting efforts, which NDVI captures as positive trends. However, the decline in SARVI values suggests that these apparent improvements might be superficial, with underlying degradation such as canopy thinning, dryland stress, or soil exposure continuing unnoticed by NDVI alone. This highlights SARVI’s strength in detecting subtle ecological changes and underscores the importance of using advanced indices for long-term forest monitoring in dryland environments.

3.7. Forest Threats and Causes of Forest Degradation in Study Area

The study area is classified as low rainfall savanna, which is characterized by high biodiversity. In the past, the local people lived in harmony with their surrounding environment. Due to several factors, the stocking density of trees at the study area started to shrink year after year.

Table 9. Percentage occurrence of threats in each forest.

No.	Forest Name	Threat Metric	%
1	Ladi	0.40	40
2	Alkoua	0.34	34
3	Mehaila	0.31	31
4	Alban Gadeed	0.35	35
5	Alghora	0.33	33
6	Shingil Gharib	0.31	31
7	Shingil Shariq	0.38	38
8	Shag Aldoban	0.33	33

shows the percentage of threats in each forest and Figure 6 shows the main factors responsible for the deterioration of forest resources at the study area.

Overgrazing is ranked at the top of the factors responsible for the degradation of forest resources, representing (20%). In addition, it is considered as a main factor contributing to the deterioration of vegetation cover and forest cover particularly in the study area. Illegal tree felling represents (9.5%), which highly contributed to forest cover deterioration in the study area. Expansion of agricultural lands on the expense of the vegetation cover, represents (9%) of total factors. Due to climate change and variability, according to which there is significant variation in the rainfall pattern, which become sporadic, farmers attempted to expand their cultivable area for the sake of harvesting similar amount of crops equivalent to what they used to harvest from their lands in the past. Other factors responsible for the deterioration of vegetation and forest cover in the study area include frequent drought cycle (16%), fire (17%), floods (17%), and erosion (5%).

The impact of interrelated drivers on forest cover change is evident across the three study sites. In the Shingil Gharib Forest, dense forest cover declined by 18.16% from 2008 to 2018, while moderately dense and low-density forests decreased by 23.10% and 14.13%, respectively. During this period, semi-bare land increased by 27.13%, indicating significant landscape degradation. A similar trend is observed in the Ladi Forest, where dense forest cover temporarily increased by 10.27% from 1998 to 2008 but then declined by 7.60% from 2008 to 2018, accompanied by a modest rise in bare land (+0.69%). This decrease suggests that previous signs of regeneration were short-lived, likely due to anthropogenic pressures such as overgrazing, illegal logging, and agricultural expansion. In contrast, the Shingil Shariq Forest exhibited greater stability, with slight increases in dense and moderately dense forest classes, possibly attributed to improved protection measures or limited accessibility (as presented above Table 6). These findings underscore the complex nature of forest degradation, where human disturbances initiate processes exacerbated by climate stressors.

The conversion of dense and moderately dense forests into semi-bare or bare lands reflects deeper ecological degradation and socio-environmental vulnerability. Addressing these challenges requires integrated land management strategies that target human drivers, such as illegal logging and overgrazing, while incorporating climate-resilient agricultural practices, ecosystem restoration, and community-based conservation initiatives. A multi-sectoral approach is essential to halt degradation and ensure the long-term sustainability of forest ecosystems.

4. Discussion

The spatiotemporal analysis revealed significant trends of forest degradation across the study area, with varying degrees of severity among different forest sites. In 2018, most forests, including Alban Gadeid, Alghara, Alkoua, Ladi, Mehaila, and Shag Addoban, were predominantly characterized by low dense forest and bare or semi-bare lands, with degraded covers consistently exceeding 50% of the total area. For instance, Alban Gadeid exhibited 63.01% combined low dense and semi-bare land, while Alghara and Alkoua recorded 61.19% and 52.88%, respectively. These figures clearly reflect a high degree of forest degradation. Shingil Shariq, however, showed relatively better resilience, with dense and moderately dense forest covers collectively making up 50.06%, suggesting more stable forest conditions compared to other sites. Overall, the results are consistent with global observations, especially in tropical and semi-arid regions, where forest cover is rapidly shrinking due to anthropogenic activities such as deforestation and climatic factors like prolonged droughts [1,3].

Trend analysis from 1988 to 2018 further revealed contrasting dynamics among selected forests. Ladi forest experienced an overall decline in dense forest cover despite temporary increases, while Shingil Shariq exhibited gradual recovery in both dense and moderate dense forest categories. In stark contrast, the Shingil Gharib forest suffered severe degradation, marked by a sharp drop in dense cover from 20.89% to 3.26% and a dramatic increase in semi-bare land from 12.85% to 35.31%. Multiple socio-environmental pressures, including overgrazing, illegal tree cutting, agricultural expansion, recurring droughts, and frequent fires drive these patterns, which were identified as the major causes of forest deterioration. The findings emphasize the urgent need for targeted conservation measures, sustainable land-use planning, and active community participation to restore and safeguard the ecological integrity of these vulnerable forests.

Alongside traditional analysis methods, the application of the soil adjusted and atmospheric resistant vegetation index (SARVI) provided valuable insights into forest condition assessment. SARVI was found to be a more effective vegetation index than the commonly used NDVI, especially in arid and semi-arid environments such as those in Sudan [53]. SARVI’s advantage lies in its ability to reduce atmospheric and soil background noise by incorporating the blue band, thus improving vegetation monitoring accuracy, particularly under sparse vegetation conditions [54,55]. The SARVI-derived maps corresponded closely with ground-truth observations, demonstrating the current degraded state of the forests, with Ladi forest being classified under very severe degradation. Importantly, SARVI application in the study area was a novel approach, offering a new methodology for forest degradation assessment in dryland ecosystems where conventional NDVI may face limitations.

However, SARVI’s application is restricted to satellite sensors containing the blue band (e.g., TM, ETM, and OLI), while NDVI remains more versatile and compatible with a broader range of satellite imagery. Despite this limitation, SARVI’s enhanced capacity to discriminate vegetation types and detect forest degradation makes it particularly suitable for monitoring forest conditions in regions like Lagawa. The study also found a strong positive correlation between SARVI and NDVI (p < 0.0001, R² = 0.9085), suggesting that both indices can be complementary tools for future assessments of forest degradation. The SARVI-based maps highlighted that severe degradation was widespread, with significant expansion of low dense and semi-bare lands over the last thirty years. This deterioration is primarily attributed to the expansion of drought, intensified grazing pressure, recurrent fires, and unsustainable land management practices.

Moreover, the forests in the study area serve as vital rangelands, especially during the dry summer months when grass and shrubs are scarce. Herders often resort to lopping tree branches to feed their livestock, further exacerbating forest degradation. This critical interaction between local livelihoods and forest resources underscores the complexity of addressing forest degradation in such environments. Ultimately, the findings illustrate that the degradation trend is not only expanding but also accelerating, necessitating urgent, integrated strategies for forest restoration, climate adaptation, and sustainable resource management.

5. Conclusions

The spatiotemporal analysis of forest cover dynamics from 1988 to 2018 reveals a concerning trend of forest degradation in the study area. By 2018, degraded land types, specifically low dense forest and semi-bare land, constituted over 50% of the total area in most forests, including Alban Gadeid (63.01%), Alghara (61.19%), Alkoua (52.88%), Ladi (56.24%), Mehaila (57.13%), and Shag Addoban (60.01%). Shingil Gharib experienced the most severe degradation, with dense forest cover plummeting from 20.89% in 1988 to just 3.26% in 2018, alongside a 27.13% increase in semi-bare land. In contrast, Shingil Shariq demonstrated resilience, showing a combined increase of 8.76% in dense and moderately dense forest over the same period, underscoring the need for targeted conservation efforts.

The drivers of degradation included both anthropogenic and environmental factors: overgrazing (20%), illegal tree felling (9.5%), agricultural expansion (9%), recurrent droughts (16%), fire (17%), and floods (17%), as identified through structured expert interviews and threat assessments. These interconnected factors highlight the necessity for integrated land-use management strategies that balance ecological sustainability with community livelihoods.

The application of the soil adjusted and atmospheric resistant vegetation index (SARVI) significantly improved the detection of forest degradation compared to the conventional NDVI, particularly in dryland environments. SARVI was more effective at identifying degradation in sparsely vegetated areas by minimizing soil and atmospheric noise. This was supported by a strong correlation with NDVI values (R² = 0.9085, p < 0.0001), revealing declines in forest condition that NDVI alone could not detect.

Overall, the study confirms that forest degradation has increased substantially over the past three decades, jeopardizing biodiversity, essential ecosystem services, and local livelihoods. Immediate, targeted actions, such as sustainable grazing management, reforestation, and fire control, are crucial to halting further degradation. Integrating advanced remote sensing tools like SARVI into forest monitoring systems can enhance early detection and support informed, evidence-based land management. Future research should employ multi-index approaches (SARVI, NDVI, EVI) for more robust monitoring across ecological zones and incorporate socio-economic variables such as population pressure, grazing intensity, and land tenure to better understand the drivers of degradation. Additionally, studying climate variability and supporting ground-based assessments of biomass and biodiversity will be essential. Promoting community-led monitoring and restoration efforts can further enhance forest resilience and provide practical, sustainable policy solutions.