Assessment and Mapping of Forest Cover Change in Dryland, Sudan Using Remote Sensing

Abstract

Forest resources in the arid and semi-arid of Sudan are experiencing significant fluctuations in tree cover and ecological functionality. This study aims to bridge this gap by utilizing multi-temporal Landsat imagery and mapping forest cover change in the Nabag Forest Reserve (NFR) in South Kordofan State, Sudan. For this assessment, two cloud-free images (TM from 2011 and OLI from 2021) were downloaded and analyzed using ArcMap 10.7 and ERDAS 2014 software. Supervised classification techniques were applied, corroborated by GPS point verification and field surveys, to quantify changes in forest cover over the decade. The results revealed that dense forest cover increased from 9% in 2011 to 38.9% in 2021, while light forest cover decreased from 34.4% in 2011 to 30.9% in 2021. Additionally, the area occupied by agriculture and barren land declined from 37.2% and 19.4% in 2011 to 18.7% and 11.5% in 2021, respectively. Rapid shifts were observed in all LULC categories during the study period. The primary causes of deforestation and forest degradation were tree felling, unsustainable grazing practices, and construction activities. These findings are crucial for guiding future forest rehabilitation and creating targeted management plans for the local communities reliant on these forests.

1. Introduction

Forest resources in Sudan’s drylands are subject to variations in tree cover and ecological functionality. Reliable and up-to-date information on these resources is essential for addressing their socio-economic and environmental roles within the national environmental policy framework [1, 2, 3, 4]. Accurate measurements of the current extent and rate of change in forest areas are imperative for devising effective management strategies to sustain the diverse ecosystem services they offer [5, 6, 7, 8, 9].

Forests cover over 11 billion acres in arid regions globally and are integral to local ecosystems and traditional food supply chains [10, 11, 12, 13, 14, 15, 16]. However, countries, such as Sudan, near deserts are experiencing a decline in tree cover and associated ecosystem services. Indiscriminate clearing for urbanization and economic pursuits threatens the study area’s protected forests. Escalating land prices have accelerated deforestation, transforming previously forested reserves into areas for housing construction, unauthorized residential developments, and various agricultural activities. Illegal logging and firewood harvesting persist as concerns [5, 17].

For sustainable resource management, precise and current land use and cover data are indispensable. Understanding the causes and impacts of land cover changes is crucial for identifying their adverse effects on biological diversity and human development [18, 19, 20, 21, 22]. Remote sensing technologies offer valuable tools for monitoring these changes [23, 24, 25, 26]. Forest composition and distribution alterations substantially influence various biological, biochemical, and ecological processes [10, 11, 12]. Data from remote sensing are often used for natural resource analysis, including tracking shifts in land use, such as forest degradation [25, 26, 27, 28].

Temporal comparisons of satellite imagery facilitate easily identifying landscape changes [29]. Landsat data, frequently employed to monitor land cover changes at regional and global scales, helps identify and map landscape features with a high level of detail [30]. Up-to-date resource inventories are vital for effective land use planning and sustainable management [31]. By integrating remote sensing, GIS, and landscape metrics, researchers can achieve more spatially consistent outcomes, better pinpointing the social and biophysical factors behind landscape fragmentation [10, 32, 33, 34].

This study aims to assess and map temporal changes in the forest cover of Nabag Forest Reserve using multi-temporal satellite imagery, ground-truth data, and GIS integration. Information from these sources was synthesized and analyzed using matrix analysis methods to provide a comprehensive view of forest cover dynamics.

2. Material and methods

2.1 Study area

The Nabag Forest Reserve (NFR) is situated in the northern region of South Kordofan State in Sudan. Spanning 4174.2 hectares, its geographical coordinates range from 12° 30′ 0″ N to 12° 36′ 0″ N in latitude and 29° 36′ 0″ E to 29° 58′ 0″ E in longitude (Figure 1). The reserve lies within a low-rainfall woodland savannah zone, experiencing an annual precipitation of 350–900 mm, primarily between May and September. The area’s temperature varies between 30 and 35°C.

Figure 1.

Maps for the location of South Kordofan state in the southern part of Sudan (A) and the location of Nabag natural Forest reserve (study area) in South Kordofan state (B), as well as the overview of Nabag natural Forest reserve (C).

The soil composition in the NNFR predominantly consists of clay plains interspersed with sandy clay, locally referred to as “Gardud.” The vegetative cover could be more sparse and more degraded. It features a scattering of acacia trees, with Acacia seyal, Acacia mellifera, and Acacia Senegal being the dominant species. Other tree varieties, such as Balanitis aegyptiaca, are also present but less prevalent.

This setting provides a challenging backdrop for the conservation and management of the forest reserve, highlighting the importance of rigorous, science-based approaches to sustain its ecological functionality.

2.2 Detecting land use/land cover, changes

We utilized remotely sensed data from Landsat 5 and Landsat 8 satellites to assess changes in land use and land cover. These images, featuring a spatial resolution of 30 meters, were cloud-free and sourced from the United States Geological Survey’s GloVis website. Specifically, we downloaded images from path 175 and row 51 for 2011 and 2021 (Table 1). Landsat 5 and 8 were selected by their geographical coverage and temporal availability, ensuring that the data would be relevant and current for our study area.

Satellite name	Sensors	Resolution	Acquisition date	Path/Row	Band used
Landsat5	TM	30 m	2011/01/11	175/051	1, 2, 3, 4
Landsat8	OLI/TRIS	30 m, 15 m	2021/01/06	175/051	2, 3, 4, 5

Table 1.

Landsat 5 TM and OLI that used in LULC determination of the study area.

Our methodology employed an integrated approach that combined remote sensing data with field information for comprehensive data collection and analysis. This integrated process is illustrated in Figure 2 of the paper. By synthesizing satellite imagery and on-the-ground data, we aimed to generate a robust, multidimensional perspective on land use and land cover changes within the Nabag Natural Forest Reserve.

Figure 2.

Flowchart for data collection and analysis.

The analysis of satellite images was performed using ERDAS Imagine 2014 and QGIS (Version 3.22.1) software, supplemented by Microsoft Excel 2019 for statistical calculations. Various stages of image processing were executed, including calibration, geometrical, atmospheric corrections, layer stacking, and composite banding. These steps converted the individual bands from each year’s dataset into single-layer files. After the preprocessing, sub-scenes from the larger images were extracted (clipped) to focus on the area of interest. Supervised classification was then carried out using the maximum likelihood classifier. This method was employed to classify land use and land cover (LU/LC) in the acquired Landsat images from 2011 and 2021. To evaluate the accuracy of the classification, we relied on ground truth data, Google Earth imagery, and preexisting knowledge of the study area. Quantitative metrics, including users’ and producers’ accuracies, overall accuracy, and Kappa coefficients, were calculated to assess the reliability of the classification. To evaluate the accuracy of the classification, we relied on ground truth data, Google Earth imagery, and pre-existing knowledge of the study area. Quantitative metrics, including users’ and producers’ accuracies, overall accuracy, and Kappa coefficients, were calculated to assess the reliability of the classification. The results demonstrated that each classified image’s overall accuracy and Kappa coefficients exceeded 85%. This high level of accuracy underscores the classification process’s reliability, confirming the approach’s utility in capturing the dynamics of land use and land cover changes within the Nabag Natural Forest Reserve (Tables 2 and 3).

Class	Bareland	Agriculture	Light forest	Dense forest	Total	UA %
Bare land	47	7	1	3	58	81
Agriculture	4	119	3	3	129	92.3
Light forest	1	1	26	1	29	89.7
Dense forest	0	0	5	29	34	85.3
Total	52	127	35	36	250
PA %; OA = 88.4%; Cappa = 82%	90.4	93.7	74.3	80.6

Table 2.

Accuracy assessment of classified map of 2021.

Class	Bare land	Agriculture	Light forest	Dense forest	Total	UA %
Bare land	85	13	2	0	100	81
Agriculture	5	89	6	0	100	83
Light forest	0	0	29	6	35	82.9
Dense forest	0	0	0	15	15	100
Total	90	102	37	21	250
PA %; OA = 80%; Cappa = 81%	94.4	87.3	78.4	71.4

Table 3.

Accuracy assessment of classified map of 2011.

3. Results and discussion

3.1 Land use/land cover results in 2021

The supervised image classification identified four distinct land cover classes: Dense forest, light forest, agriculture, and bare land. These categories are visually represented in Figure 3. According to the OLI 2021 classification results, dense forests accounted for 38.9% of the total area, equivalent to 1623.78 hectares. Light forest occupied 30.9%, translating to 1289.97 hectares. Agricultural land constituted 18.7% of the area, or 780.75 hectares, while bare land comprised 11.5%, totaling 479.97 hectares. These figures and their corresponding percentages are further detailed in Table 4.

Figure 3.

Distribution of LU/LC classes detection in NFR in 2021.

Class Name	2021
	Area (ha)	%
Dense forest	1623.78	38.9
Light forest	1289.97	30.9
Agriculture	780.75	18.7
Bare land	479.97	11.5
Total	4174.11	100

Table 4.

Distribution of LULC in the study area (ha) in 2021.

This classification provides a comprehensive up-to-date understanding of land use and land cover within the Nabag Forest Reserve. The results offer valuable insights into the current state of the forest and the broader landscape, essential for informed decision-making in conservation and land management strategies.

3.2 Assessment of land use/land cover change

Globally, changes in land use and land cover (LULC) are among the most significant and enduring modifications to the Earth’s surface [35, 36]. Monitoring these changes is essential for generating baseline thematic maps and facilitating ongoing assessments. In the current study, as illustrated in Figure 4 and Table 5, substantial LULC changes were observed during the study periods of 2011 and 2021, attributable to both human and environmental factors.

Figure 4.

Maps of land cover change in NFR between 2011 and 2021.

Class Name	2011		2021
	Area (ha)	%	Area (ha)	%
Dense forest	377.82	9	1623.78	38.9
Light forest	1435.86	34.4	1289.97	30.9
Agriculture	1550.79	37.4	780.75	18.7
Bare land	809.64	19.4	479.97	11.5
Total	4174.11	100	4174.11	100

Table 5.

LU/LC classes and their area in percentage for 2011 and 2021.

Specifically, the results reveal a significant increase in dense forest cover, growing from 9% in 2011 to 38.9% in 2021—an impressive net change of 29.9% and an annual growth rate of 2.99%. Studies on forest rehabilitation indicate that tree planting typically accelerates the recovery of forest vegetation [37]. Hemida et al. [38] found that the increase in forest cover during a certain period can be attributed to the Taungya agroforestry program initiated by the FNC in early 2005 and continuing to the present day. Salih [39] also reported that the Taungya program positively contributed to rehabilitating 3024 ha of NFR between 2005 and 2013. Similarly, a study from Nigeria highlighted the advantages of the Taungya system in promoting forest conservation and regeneration [40]. Conversely, light forest cover decreased from 34.4% in 2011 to 30.9% in 2021, with a net change of 3.5% and an annual decline rate of 0.35%. Agricultural land also reduced from 37.2% in 2011 to 18.7% in 2021, marking a net change of 18.5% and an annual decrease of 1.85%. Lastly, bare land areas contracted from 19.4% in 2011 to 11.5% in 2021, experiencing a net change of 7.9% and an annual reduction rate of 0.79%. It is important to note that other factors, including environmental, political, and agricultural influences, might have contributed to the observed changes in the reserve during this period. These factors significantly impact land use and land cover transformations [1, 41]. Historical reviews of the study region reveal that since 1982, there have been recurrent dry years. Additionally, frequent civil wars have resulted in significant losses of vegetation cover [42].

A study by [1, 2, 38, 43] corroborates these findings, suggesting deforestation, flooding, soil erosion, and unplanned urban and agricultural expansion could lead to such LULC changes. The study also emphasizes that LULC modifications, influenced by varying environmental, political, demographic, and socio-economic conditions, are dynamic and directly affect communities living near forests.

In light of our findings, it is evident that anthropogenic activities are the driving forces behind these significant changes and disruptions, impacting the natural vegetation cycles in the study area. This research provides valuable insights into the human and environmental dimensions of LULC changes, underscoring the urgency for informed and adaptive land management strategies.

Land use and land cover (LULC) change is not merely a straightforward or linear conversion between different land types but involves a complex interplay of human activities, environmental conditions, and policy frameworks [43, 44]. This multifaceted nature of LULC makes its study particularly compelling as it can offer insights into broader sociocultural and environmental trends and processes.

Utilizing temporal remote sensing data enables the analysis of dynamic processes, emphasizing not only static snapshots of land cover at specific times but also the sequences and rates of their changes [2, 45]. In the case of the Nabag Forest Reserve, the dynamic nature of land transitions is evident spatially and temporally, especially in recent years. Use of transition matrices (as shown in Table 4 and the maps in Figure 5) underscores the complex and dynamic landscape of the forest’s changes. Transition matrices provide valuable analytical tools for understanding how these complex systems evolve, particularly in semiarid regions, where the conservation of natural resources is a significant concern [46].

Figure 5.

LU/LC change trajectory matrix 2011–2021.

Transition matrices allow for a multidimensional view of change over time by showing what changes have occurred and the extent and direction of these changes. In essence, they can provide insights into what drives the change in each class and help predict future trends based on past and current activities [47].

By employing matrix analysis tools, you have captured the systematic transitions in land use and land cover over the study period of 2011–2021. These matrices, visualized in Figure 5 and Table 6, offer a nuanced understanding of the factors affecting Nabag forest, thereby providing a basis for more informed and targeted conservation and land management strategies.

			Year	2021
	Class name	Dense forest	Light forest	Agriculture	Bare land	Total (ha)
Year	Dense forest	179.37	668.43	570.87	205.02	1623.69
2011	Light forest	80.19	392.85	510.66	306.27	1289.97
	Agriculture	91.98	246.6	283.95	157.95	780.48
	Bare land	26.28	127.98	185.31	140.4	479.97
	Total (ha)	377.82	1435.86	1550.79	809.64	4174.11

Table 6.

Land use and land cover trajectory matrix 2011–2021.

Note: Highlighted values illustrate areas and percentages of the unchanged classes.

Understanding the multilayered factors driving LULC changes is essential for developing robust and sustainable management plans. It enables stakeholders, from local communities to policymakers, to make evidence-based decisions considering the complex realities of land transformation and its implications.

The intricacies of land use and land cover changes in the Nabag Natural Forest Reserve from 2011 to 2021 are highly revealing. The transition matrices offer a nuanced view, showing that the area under dense forest has seen a significant net gain of 1326.04 hectares over the decade, largely attributed to the Forest National Corporation’s agroforestry activities. However, this positive trend is counterbalanced by a loss of 118.26 hectares of dense forest, mostly converted to agriculture and bare land. This indicates that issues, such as rainfed agriculture still significantly threaten forest conservation. Furthermore, the area under light forest, categorized as fragile, has seen degradation, largely through its conversion to dense forest. This could signal environmental stress due to natural and anthropogenic factors. The decline in agricultural land could be seen as beneficial for natural ecosystems. However, it also raises concerns about declining soil fertility and the socio-economic implications for local communities. A similar decline in bare land, partially converted to dense forest, appears to be a positive trend but necessitates further inquiry into the driving factors for long-term planning. The data also flags a concerning lack of participation from local communities in forest management, emphasizing the need for more community-based, sustainable approaches. Ultimately, the transition matrix is an invaluable ledger for policymakers, offering a detailed account of losses and gains in land cover categories. Given the dynamic nature of these changes, adaptive evidence-based management strategies are crucial for the long-term conservation and sustainable development of the Nabag Forest Reserve (Table 2).

Sudan forests grapple with deforestation, primarily driven by agricultural growth and infrastructure projects. Despite initiatives promoting reforestation and sustainable land use, there remains a pressing need for a unified national strategy. Such a strategy should not only focus on forest conservation but also intertwine with community development, emphasizing the forest’s role in climate change mitigation and biodiversity preservation (Table 3) [48, 49, 50, 51].

4. Conclusion and recommendations

A multifaceted approach involving satellite imagery, geospatial technology, and ground inventories was utilized to detect, assess, monitor, and map land use and land cover (LULC) changes in the Nabag Natural Forest Reserve in South Kordofan State, Sudan. This study elucidated spatial and temporal patterns of LULC variations, offering insights for effective forest management and future rehabilitation plans. The research found substantial LULC alterations between 2011 and 2021, with differing trends and percentages. Specifically, the analysis revealed that dense forest cover increased from 9% in 2011 to 38.9% in 2021, while light forest cover declined from 34.4% to 30.9% in the same period. Agricultural and bare land areas also decreased, dropping from 37.2 and 19.4% in 2011 to 18.7 and 11.5% in 2021, respectively. These shifts were primarily driven by anthropogenic activities such as illegal tree felling, woodcutting, overgrazing, and infrastructural development. These factors contribute to both deforestation and alterations in forest cover. Such changes are anticipated to affect the forest’s ecosystem services significantly. If the current degradation rate persists, the long-term sustainability of the forest could be jeopardized. Therefore, the Forestry National Corporation (FNC) should focus on preserving the remaining areas of Nabag forest. The active participation of local communities in rehabilitation efforts while respecting their interests is crucial for the success of any restoration program. Ignoring this could lead to the failure of rehabilitation initiatives. The outcomes of this study can inform the development of comprehensive forest management and rehabilitation plans that are conscious of community needs.

Acknowledgments

This chapter has been supported by the BorderEye project (TKP2021-NVA-13), which has been implemented with the support provided by the Ministry of Culture and Innovation of Hungary from the National Research, Development and Innovation Fund, financed under the TKP2021-NVA funding scheme. We also wish to express our gratitude to our colleagues for their contributions, both direct and indirect, to the data collection and analysis. The authors affirm that there are no conflicts of interest that could have affected the publication of this paper.