Modeling soil erosion risk in the Kashmir Valley, India: Spatiotemporal dynamics and future projections under CMIP6 climate and land use scenarios

Water-induced soil erosion threatens agricultural land and water resources across the Kashmir Himalaya, where steep terrain, expanding horticulture and climate variability interact under uncertain future conditions. This study combines the Revised Universal Soil Loss Equation (RUSLE), CA–Markov land-use projections and a four-model CMIP6 ensemble to produce spatially explicit, scenario-based estimates of average annual sheet and rill soil loss across the Kashmir Valley for 2005, 2010, 2015 and 2020 and for the future horizons 2030, 2050 and 2070 under SSP2–4.5, SSP3–7.0 and SSP5–8.5. Rainfall erosivity was derived from 41 years of monthly station records using the Modified Fournier Index. Historical domain-wide mean soil loss ranged from 8.57 to 14.31 t ha⁻¹ yr⁻¹ and did not change monotonically; the Extreme class (>50 t ha⁻¹ yr⁻¹) expanded from 186,415 ha in 2005–246,699 ha in 2020. For 2070, projected means are 8.12, 4.36 and 2.41 t ha⁻¹ yr⁻¹ under SSP2–4.5, SSP3–7.0 and SSP5–8.5, respectively, while the corresponding pixel maxima are 175.69, 185.65 and 187.63 t ha⁻¹ yr⁻¹ . A raster-based CA–Markov hindcast of 2020 from the 2005–2015 transitions yielded a Figure of Merit of 0.160 and Kappa of 0.475. Overall agreement (0.569) was below the no-change persistence benchmark (0.662); therefore, the future LULC outputs are interpreted as conditional scenarios with substantial allocation uncertainty. Recurrent high-value areas in the northeastern and southwestern highlands are candidates for field verification and targeted conservation.