Soil erosion is a critical challenge to the sustainability of soil and water resources, particularly in volcanic regions like the Telagawaja Sub-Watershed, Bali. The area is defined by andesitic-basalt rocks (Qvab), basaltic tuff (Qvbt), and ancient volcanic formations from the Tertiary period (Tomub), coupled with tropical climate conditions with an average rainfall of 219.21 mm per month, temperatures ranging from 20 to 27 °C, and intensive agriculture, making it highly prone to erosion. This study compares soil erosion predictions from conventional data sources, derived from field observations and laboratory tests, with remote sensing data obtained from multisensor satellites (e.g., CHIRPS, Sentinel-2, and DEM Alos Palsar), using the Universal Soil Loss Equation (USLE). Erosion estimates based on conventional data range between 0.37 and 4,657 t ha−1 yr−1, while remote sensing estimates vary from 0 to 49,384 t ha−1 yr−1. Both approaches highlight that very light erosion (<15 t ha−1 yr−1) dominates the region, covering 45.90% and 29.21% of the area, respectively. The comparison reveals a 21.03% agreement between methods, with 78.97% exhibiting differences in erosion classification. Conventional methods tend to produce more uniform outcomes, whereas remote sensing generates more spatially detailed, pixel-based maps, especially in areas with complex topography and vegetation variability. The study underscores the value of integrating both techniques for generating more precise erosion zonation maps, crucial for effective watershed management and soil conservation strategies.
