Regional soil erosion assessment from remote sensing data in rehabilitated high density canopy forests of southern China

• Vegetation restoration degrees determined from LAI and VFC are used to assess soil erosion under high density canopy.
• The model USLE is used to extrapolate vegetation indices threshold from plot scale to regional scale.
• Vegetation restoration degree determined from LAI is more sensitive to subcanopy soil erosion than from VFC.
• The remote sensing based method can be used for operational erosion monitoring in southern China.

Soil erosion is the most severe environmental problem in the red soil region of southern China. Erosion has been significantly reduced over the last 30 years through the deployment of a massive government re-forestation program. Nevertheless, soil erosion is still severe in some areas, and an efficient method for assessing soil erosion under secondary forest canopy is needed. Traditionally the vegetation indices derived from remotely sensed imagery have been used for identifying eroded areas. However, under high density forest canopy (HDFC) their applicability suffers due to biomass light absorption, which varies by canopy structure. A mapping method was developed by integrating remote sensing parameters to identify eroded areas. The remotely sensed vegetation indices, vegetation fractional canopy (VFC) and leaf area index (LAI) were calibrated based on soil erosion measurements from previous runoff plot experiments and extrapolated from USLE modeling to a regional scale. Using vegetation restoration degree (VRD), based on VFC (VRDVFC = VFC/VITVFC) and LAI (VRDLAI = LAI/VITLAI), soil erosion under HDFC was identified. Results indicate that the threshold value of VFC (VITVFC) and LAI (VITLAI) ranged from 0.45 to 0.60, and from 1.3 to 2.7, respectively. Secondary forest associated with VRDVFC > 100% or VRDLAI > 100% occupied 75.8% and 37.8% of the total study area, respectively. About 50% of the area distinguished as being eroded by LAI, was mapped as having no obvious erosion by VFC. LAI based mapping had a precision of 96.7% according to field validation. The eroded areas were primarily distributed in locations with elevation between 300 to 500 m and slope angles below 25°. The present method for distinguishing soil erosion under HDFC by combining VRDVFC and VRDLAI is effective and can be used for operational erosion monitoring in the red soil region of southern China.