Assessing soil erosion in understory-eroded forestlands based on the three-dimensional index of stratified vegetation volume

Introduction Traditional vegetation indicators such as fractional vegetation cover (FVC) and NDVI capture only horizontal extent and fail to distinguish the erosion–reduction functions of canopy, understory, and litter layers, often overestimating soil conservation in forests with closed canopies but degraded near‑surface vegetation. Methods We developed an empirical model to assess the soil erosion index (SEI) based on Stratified Vegetation Volume (cover × leaf area index or litter depth) across three vertical layers. SEI was measured from 17 standard runoff plots in a typical red soil erosion area of southern China and extrapolated to 46 natural forestland sites using a two‑variable empirical model based on bulk density and gravel content (cumulative grade accuracy: 91.1% within ± 1 erosion grade). Results Using extrapolated SEI as dependent variable and vegetation volumes of the three layers as independent variables, the three‑variable regression model performed best ( R 2 = 0.650, RMSE = 0.104). Regression coefficients followed a clear hierarchy: understory (−0.186) > litter (−0.046) > canopy (−0.040), indicating that understory vegetation contributed approximately 4.6 times more to erosion reduction than the canopy. The two‑indicator model (understory + litter) achieved the same grade classification accuracy (96.3%) as the three‑indicator model, offering a simplified pathway for rapid assessment when resources are limited. Discussion These results establish near‑surface vegetation as the core functional layer for forest soil conservation and provide a quantitative tool for diagnosing forest soil health and guiding restoration prioritization in degraded subtropical forests.