Advancing Soil Erosion Assessment: Application of Remote Sensing and Geospatial Techniques in Bulango Ulu Reservoir Basin

. Soil erosion is an important concern due to the steepness of the terrain and the significant elevation differential between the upstream and downstream regions of the basin. Revised Universal Soil Loss Equation (RUSLE) was integrated with Remote Sensing (RS) and Geographic Information System (GIS) in the current work establish the annual soil erosion map in the Bulango Ulu Reservoir basin. The RUSLE model incorporated soil erosion zonation features such as rainfall erosivity, soil erodibility, topographic, vegetation cover, and conservation support practices. The results show that 0 and 110.31 t year −1 are the minimum and maximum soil erosion, with the average rate of soil erosion is 17.30 t year −1 in the present study area. Soil erosion risk regions were divided into five categories: very slight, slight, moderate, severe and extremely severe soil erosion areal extent and area percentages is 229.17 km 2 (94.48%), 7.83 km 2 (3.23%), 4.25 km 2 (1.75%), 1.20 km 2 (0.50%), and 0.12 km 2 (0.05%), respectively. The Area Under Curve was indicated that RUSLE model had good performance (75.1%). This study demonstrates the utility of GIS and remote sensing for predicting soil erosion, allowing important information to be extracted for implementing soil conservation programs in the Bulango Ulu reservoir basin.


INTRODUCTION
One of the most urgent environmental phenomena facing countries worldwide is soil erosion (1,2).Its negative impact on the environment and agricultural productivity makes it a challenge that must be addressed immediately (3).Extensive research has shown that soil erosion has a negative impact on soil fertility, water quality, and biodiversity, resulting in significant economic losses and ecological damage (4).Governments and policymakers are increasingly recognizing the importance of this issue and taking steps to mitigate its consequences (5).Global soil erosion is estimated to be 75 billion tons per year, with the majority of cropland losing soil at a rate of 13 to 40 t ha -1 a -1 (2).Indonesia is one country with severe soil erosion, soil erosion is estimated to be 35-220 tons ha -1 per year, with an annual rate of growth of 7-14%, or 3-28 tons ha -1 .(6).Addressing soil erosion requires collaborative efforts on a global scale, emphasizing sustainable land management practices and the adoption of innovative technologies to safeguard the world's precious soil resources (1).The urgency to combat soil erosion remains a central focus for researchers, policymakers, and environmentalists alike, aiming to secure a more resilient and sustainable future for all nations (7).
The quantitative evaluation of soil erosion is critical in developing effective soil erosion control strategies, and using mathematical evaluation models is a useful tool for assessing this environmental hazard (8).By quantifying the extent and intensity of soil erosion, these evaluation models aid in understanding its impacts on soil health, agricultural productivity, and ecosystem integrity (9).Integrating data-driven approaches, like as the Universal Soil Loss Equation (USLE) or the Revised Universal Soil Loss Equation (RUSLE), allows for a comprehensive assessment of soil erosion dynamics (10).Researchers have demonstrated the usefulness of these models in guiding the formulation of sustainable land management practices and soil erosion control measures (2,(11)(12)(13)(14)(15)(16)(17).As soil erosion continues to be a global concern, the adoption of these mathematical tools holds significant promise for fostering resilient ecosystems and safeguarding precious soil resources.
Traditional methods of assessing soil erosion can be costly and time-consuming (15).However, the use of geographic information systems (GIS) to integrate existing soil erosion models, field data, and data provided by remote sensing technologies appears to be an asset for future research (18,19).Many scientific studies have been conducted in the last decade to characterize soil erosion in large areas using remote sensing and GIS technologies (20,21).These investigations indicated that these methodologies presented extremely good information concerning soil erosion zones, such as types of soil, lithological units, and coverage of vegetation, at a low cost and with a high degree of accuracy (20).The RUSLEbased assessment showed that the estimated total annual potential soil erosion can be estimated using remote sensing and GIS data.Generating accurate soil erosion maps in GIS environment is very important to locate the areas with high soil erosion and to develop adequate soil erosion prevention techniques (21).

Study Area
The focus of the study location is the District of Bulango Ulu, Bone Bolango, Gorontalo, Indonesia, which is located in the northeastern part of Gorontalo.The mountain basin considered for this research is in the Bolango River basin and it is one of the main branches.Its upstream, the Bulango Ulu Reservoir is being built on an area of 110.9 hectares in Owata Village.It is the only reservoir project in Gorontalo which is part of a series of irrigation reservoirs planned in Indonesia.This reservoir is also expected to reduce flooding by 84.62% in the downstream part of Gorontalo City.The Bulango Ulu Reservoir basin covers approximately 242.561 km 2 (24,256.1 ha) of land area, between latitude 0° 0° 34' 58.44"N -0° 46' 28.4556"N and longitude 123° 5 ' 36.9528"E-123° 16 ' 18.966"E (Fig. 1).The topography is mostly undulating, with elevations ranging from 48 to 1,908 meters above sea level..The range annual temperature is between 26.76 and 25.23 °C with a mean relative humidity is 80.67%.The basin receives 954.49-2857.33 mm of rainfall per year.The Bulango Ulu Reservoir Basin's dominant soil types are nitisols, ferrasols, and cambisols.The forest is the primary land use in the Bulango Ulu Reservoir Basin.The current study area's significant LULC types are settlements, shrub lands, plantations, water bodies, and cultivated land.

Data
Soil erosion assessment and modelling require the integration of diverse data from multiple sources, each with varying formats and scales.In order to achieve this goal, the current study successfully combined spatial datasets with on-site field survey data, allowing for a more comprehensive understanding of the phenomenon.Table 1 carefully documents and presents the different sources of spatial data used in this study, allowing for transparency and reproducibility in future research efforts.A soil erosion model was conducted on a GIS platform using ArcGIS 10.4 and Arc Hydro 10.4 (ESRI).ASTER Global Digital Elevation Maps (GDEM) was utilized to define the basin boundary and to derive the drainage lines.It derived the topographic factor and the support practice factor with overlaid by land use map.LANDSAT 8 digital data were utilized to estimate the vegetation parameters.The texture soil derived the erodibility factor and interpolation by Inverse Distance Weighting (IDW).The digital dataset had a spatial resolution of 30 m 30 m, which was similar with the LANDSAT image.

The Revised Universal Soil Loss Equation (RUSLE)
RUSLE model, a well acknowledged and frequently used tool for quantifying soil erosion, was used to estimate the geographical mapping of the soil erosion process.RUSLE takes into account several crucial factors that influence soil erosion, including rainfall erosivity, soil erodibility, slope length and steepness, vegetation cover, and land management practices.By integrating these variables, the model generates valuable insights into the potential rates and patterns of soil erosion across different landscapes.This approach not only aids in identifying vulnerable areas but also offers valuable guidance for implementing targeted soil erosion control measures and sustainable land management strategies.The reliance on the RUSLE model as a means of spatial estimation underscores its significance as a practical and effective tool in combating soil erosion and preserving the integrity of our precious soil resources.Renard,et al. in (22) define the RUSLE approach as follows: (1) where: A is the annual average of soil erosion rate factor (t ha −1 yr −1 ); R is the rainfall erosivity factor (MJ mm ha −1 h −1 yr −1 ); K is the soil erodibility factor (t ha MJ −1 mm −1 ); LS is the dimensionless slope length and steepness factor; C is the dimensionless crop management factor (ranging between 0 and 1) and P is the dimensionless conservation support practice factor (ranging between 0 and 1).
Soil erosion is caused by certain factors, it is critical to identify these factors.As a result, one of the most important aspects of this research was the creation of a spatial database for the effective factors.Conditioning factors were chosen based on previous research, data availability, and geographical knowledge (23,24).The following factors were taken into account in the current study: rainfall erosivity factor, soil erodibility factor, topographic factor, cover management factor, and support practice factor.

The Erosivity Rainfall
The effect of rainfall on soil erosion is the rainfall erosivity (R).It reflects the effect of rainfall intensity on soil erosion, and requires detailed, continuous rainfall data for its calculation (15).The R factor is commonly determined from rainfall intensity if such data are available.In the present study, monthly rainfall data for 10 years (2013-2022) were utilized to determine the R factor using the following Eq.( 2) developed by Wischmeier and Smith (1978) in (18).(2) where R is the rainfall erosivity factor (MJ mm ha -1 h -1 year -1 ), P m is the monthly rainfall (mm), and P a is the annual rainfall (mm).

The Soil Erodibility Factor
The soil erodibility factor (K) quantifies a soil's susceptibility to soil erosion by rainfall and runoff based on properties like texture, organic matter, structure, and slope, providing a numerical measure of how easily soil particles can be detached and transported, aiding in soil erosion prediction and soil conservation efforts (25).It is a measure of the susceptibility of soil erosion and the rate of runoff, as measured under the standard unit plot (19).The K factor equation based on Renard et al. (1997)   exp 0.01 ln where D g is geometric mean particle diameter.The resulting K value is reported in United States customary units of short ton ꞏ ac ꞏ h/ (100 ft ꞏshort ton ꞏ ac ꞏ in).It converted to t ha MJ −1 mm −1 with multiplied by 0.1317 (26).

The Topographic Factor
The topographic factor (LS) consists of two sub-factors: a slope gradient factor (S) and a slope-length factor (L), both of which are generated from ASTER GDEM (25).The L factor represents the impact of slope length on soil erosion.It measures the distance from the origin of overland flow to the point where the slope either starts to decrease or where runoff enters a well-defined channel, as longer slopes often result in higher soil erosion potential due to increased runoff and velocity, making it a critical factor for soil erosion risk assessment and soil conservation planning.As a result, as the slope length increases, so does the soil erosion per unit area.The S factor quantifies the impact of slope steepness on soil erosion, with steepness exerting a more significant influence on soil erosion than slope length.Steeper slopes lead to higher soil erosion potential due to increased runoff velocity, making the S factor a critical parameter in assessing soil erosion risk and guiding soil conservation efforts (15).In soil erosion modelling, the slope-length and gradient parameter are critical for calculating overland flow (surface runoff).The topographic factor is computed using Eq (5 and 8) (Moore and Burch (1986) in Golijanin et al., 2022).
  where λ is slope length (m), θ is the slope (radians), and m is related to the ratio  of rill soil erosion (caused by flow) to inter-rill soil erosion (principally caused by raindrop impact).

The Cover Management Factor
The cover management factor (C) in soil erosion modelling is predominantly determined by the percentage of vegetation cover and its growth duration.It also considers additional factors like mulch cover, crop residues, and tillage practices.The C factor reflects how effectively the vegetation and ground cover protect against soil erosion, making it an essential component for estimating soil erosion and designing soil erosion control strategies.The C factor decreases from 1 to 0 depending on vegetation cover and cropping management systems implemented to mitigate soil erosion (18).The C factor values seem to be inversely correlated with NDVI (Normalized Difference Vegetation Index) values.NDVI is a measure of vegetation cover and health.The lowest NDVI values (indicative of poor vegetation cover) correspond to the highest C factor values, implying that areas with less vegetation cover have a greater potential for soil erosion.
NIR-RED band 5-band 4 NDVI= = NIR+RED band 5+band 4 (10) where x = 2 and y = 1, which are the parameters indicating the shape of the relative curve between NDVI and the C factor, NIR is reflection in the near-infrared spectrum (band 5), and RED is reflection in the red range of the spectrum (band 4).NDVI assumes values from -1 to +1, with the highest values attributed to areas with greater vegetation cover.

The Conservation Support Practice Factor
The conservation support practice factor (P) in soil erosion modelling establishes a connection between the soil erosion ratio associated with a specific support practice and the soil erosion caused by both upslope and downslope tillage.It quantifies the impact of practices that modify the runoff quantity and rate, leading to reduced soil erosion.P factors are assigned to different land use/cover types and slopes based on land use/cover-type maps, aiding in assessing the effectiveness of conservation practices in mitigating soil erosion across various land use scenarios (Table 2).The P factor for forest with less litter is 0.005.

Soil Erosion Factor Spatial Distribution
Each soil erosion factor was transformed into a raster image with WGS84 coordinates and a 30m×30m resolution, using the ArcGIS platform.These five components were then combined and quantified to create a comprehensive soil erosion rate map for the Bulango Ulu Reservoir Basin, as depicted in Figure 2, offering a visual representation of soil erosion distribution in the area.
The study area covered one grid area in 0° 45' 0"N and Longitude 123° 15' 0"E with a grid size of ½° x ½°.Therefore, it only used annual rainfall data on one grid.The annual rainfall for the years 2013-2022 was found to be in the range of 909.7 -2478.0mm respectively.Using Eq. ( 1), the average R factor was observed to be 1042.7 MJ mm ha -1 h -1 year -1 .The highest value (159.2MJ mm ha -1 h -1 year -1 ) of R factor was observed on January when the total rainfall was 159.2 mm.The lowest value (9.5 MJ mm ha -1 h -1 year -1 ) of R factor was observed to be on August when the total rainfall was 75.9 mm (Table 3).The kinetic energy needed for soil to separate from aggregate and eventually flow downstream is provided by the R factor.Rainfall intensity and seasonality have been taken into account within the R factor.High detachment rates in soil are a result of brief, intense rainfall events.High-intensity rainfall happens during the monsoon season in tropical climates.Long dry seasons and abrupt, frequently occurring, intense rainfall events cause high rates of soil detachment.The environment of the current study area is continuously changing between dry and wet periods, which causes relative variations in moisture stress and affects the process of soil erosion (25).The LS factor in this context ranges from 0 to 10.78, with the most elevated values found in regions characterized by steep valleys, as illustrated in Figure 2. Higher LS factor values were found in the middle and eastern areas.This spatial pattern aligns with the expectation that areas with more pronounced slopes will have higher potential for soil erosion.The phenomena suggests that soil erosion increases with slope length and steepness (21).It is in line with the fundamental principles of soil erosion.Longer slope lengths and steeper slopes contribute to increased water runoff and greater erosive forces, leading to higher soil detachment and transport (27).Soil erosion is proportional to slope length rather than slope steepness.A comprehensive examination of all soil erosion factors found that the LS factor had a considerable impact on estimating total soil erosion in the study area.
The K factor erodibility of the basin ranged between 0.41 and 1.51 t ha MJ −1 mm −1 , with a mean value of 0.74 t ha MJ −1 mm −1 .This indicates the range of erodibility across different soil types within the basin.A higher K factor value suggests that the soil is more susceptible to soil erosion, while a lower value indicates greater resistance to soil erosion.The K factor map reveals that elevated values are predominantly concentrated in the upstream regions and to a lesser extent in the southern areas, indicating a higher risk of soil erosion in these locations.Conversely, lower K factor values prevail in a significant portion of the area, especially in the western and central sections of the study area, as shown in Figure 2.This association with lower K factor values in those areas may be attributed to specific soil characteristics like texture or higher clay content, which generally reduce erodibility and make the soil less prone to soil erosion (19).
The magnitude and the spatial distribution of C factor show values between 0.05 and 1.00, with mean value of 0.26.This suggests that different land cover types have varying degrees of impact on soil erosion.The spatial distribution of these C factor values across different land cover types likely reflects the varying degrees of soil protection or vulnerability in different areas (28,29).The highest (poor land cover management) almost coincide with the lowest NDVI values, (0.01 -0.20), while the bare land has a high C factor (0.30).Similarly, the plantation areas have a C factor of (0.26).Different land cover types have different C factor values.Poor land cover management, as reflected by low NDVI values, has the highest C factor (indicating high vulnerability to soil erosion).Bare land and plantation areas also have relatively high C factor values, implying higher soil erosion potential compared to other land cover types (16).
Taking into account the conservation support practice factor, the P factors were identified (as shown in Figure 2).The P factors in the range of 0.05 to 1.00 signify the effectiveness of conservation support practices in mitigating soil erosion.Lower P factors indicate successful conservation efforts, while higher P factors indicate insufficient or absent conservation measures, particularly in forests, natural vegetation, and major settlements in the basin.The highest P factors correspond to cropland with relatively inadequate conservation practices in the downstream area, reflecting a higher risk of soil erosion in those regions.

Soil Erosion Rate Assessment
This study employed a combination of remote sensing, GIS methods, and RUSLE model to assess the scale and spatial pattern of soil erosion in the research area.It involved the calculation of five key soil erosion risk factors, namely rainfall erosivity (R factor), soil erodibility (K factor), topography (LS factor), land cover and management (C factor), and conservation support practices (P factor), which collectively enabled a comprehensive evaluation of soil erosion vulnerability across the study region.
Based on the soil erosion rate distribution map of the Bulango Ulu Reservoir Basin, this study used the soil risk classification method by Olii, M & Ichsan in (30), and considered variations in soil erosion risk classification across different spatial scales, such as global, national, and regional.Consequently, the study categorized soil erosion risk into five distinct levels (very slight (< 15 tꞏha -1 ꞏyear -1 ), slight (15-60 tꞏha -1 ꞏyear -1 ), moderate (60-180 tꞏha - 1 ꞏyear -1 ), severe (180-480 tꞏha -1 ꞏyear -1 ), and extremely severe (> 480 tꞏha -1 ꞏyear -1 )), as depicted in Figure 3, offering a refined assessment of soil erosion risk within the study area.The average rate of soil erosion is 17.30 t year −1 .Very slight, slight, moderate, severe and extremely severe soil erosion areal extent and area percentages in the current research area is 229.17 km 2 (94.48%), 7.83 km 2 (3.23%), 4.25 km 2 (1.75%), 1.20 km 2 (0.50%), and 0.12 km 2 (0.05%), respectively (Table 4).The final soil erosion model has generated Figure 3, which illustrates the annual soil erosion map of the study area, aiding in the identification of regions with an elevated susceptibility to soil erosion.The study's analysis pinpointed regions with elevated risk to soil erosion, specifically concentrating in the middle portions of the study area in the form of shrubs (60.50%) and dryland agriculture (11.11%).Table 5 shows that forests contribute 23.11% to soil erosion in the Bulango Ulu Reservoir basin.In particular, besides forests dominating this basin for 87.04% of the total basin area, the presence of steep slopes contributed to the risk of these locations, as such terrain accelerates the movement of water and increases erosive forces (31).Furthermore, the soil conditions prevalent in these regions demonstrated a lack of resilience against soil erosion processes, exacerbating the risk (32).Adding to the complexity, the forest cover in these identified areas was characterized by notably low density, failing to provide the protective cover required to mitigate soil erosion.This interplay of factors underscores the intricate nature of soil erosion dynamics, where seemingly protective elements such as forests can, under specific circumstances, contribute to heightened vulnerability.The recognition of these high-risk zones emphasizes the necessity of targeted conservation strategies, even within forested landscapes, to address the multifaceted factors that amplify soil erosion risk.

Validation
The accuracy of the soil erosion risk results was verified using success rate and prediction rate curves, with validation involving a comparison between soil erosion locations and the generated risk maps (33).The Area Under Curve (AUC) is a metric used to gauge the accuracy of a final map in predicting soil erosion risk (34).AUC values range from 0.5 to 1.0, with a value closer to 1 indicating higher model accuracy.AUC values can be categorized as follows: 0.5-0.6 (poor), 0.6-0.7 (average), 0.7-0.8(good), 0.8-0.9(very good), and 0.9-1 (excellent) (35,36).This metric offers a clear and standardized way to assess the performance of soil erosion risk prediction models, with higher AUC values signifying greater accuracy in identifying areas at risk of soil erosion.The validation process combined independent field survey data and high-resolution satellite imagery from sources like Google Earth and SAS Planet.In this study, 150 sample locations were used, comprising 100 samples for soil erosion risk areas and 50 for non-soil erosion risk areas.The results, as shown in Table 6 and Fig. 4, indicated a good rate of 0.751 (75.1%), suggesting a relatively accurate predictive model for soil erosion risk assessment.

CONCLUSION
This study employed the RUSLE, complemented by Remote Sensing and GIS, to compute soil erosion rates in the Bulango Ulu reservoir basin.RUSLE was chosen for its simplicity, ease of understanding, minimal data requirements, and utilization of readily available inputs.The assessment of soil erosion enabled the classification of the basin into five soil erosion risk categories, with a mean estimated soil erosion rate of 20 tonꞏha −1 ꞏyear −1 , with the five soil erosion risk classes, ranging from 0 and 110.31 t year −1 .Areas of 1.20 km 2 (0.5%) and 0.12 km 2 (0.05%) were classed as severe and extremely severe soil erosion.The accuracy of produced maps by these models was validated and compared using AUC, which showed the accuracy of the maps created by RUSLE models was 75.1% (good performance).These areas in the Bulango Ulu reservoir basin should therefore be prioritized for conservation to extend the lifetime of the reservoir or at least in accordance with the planned lifetime of the reservoir.In conclusion, this study emphasizes the efficiency of using Geographic Information Systems (GIS) and Remote Sensing (RS) as accessible and costeffective techniques for modelling soil erosion.These methodologies enable the evaluation of soil erosion potential and risk in the context of the Bulango Ulu reservoir basin, delivering useful insights for conservation and management initiatives in the area.

Table 1 .
Datasets used for soil erosion

Table 4 .
Risk level of soil erosion

Table 5 .
Soil erosion rate based on land use

Table 6 .
Values derived from a matrix of confusion for assessing the soil erosion risk model performance.
Fig. 4. The AUC (The Area Under Curve)