Estimation of dry vegetation cover and mass from MODIS data: Verification by roughness length and sand saltation threshold

Abstract

Monitoring of dry vegetation in arid and semi-arid regions is crucial in environmental investigations such as assessments of wind erosion risk, livestock forage, and fire risk. However, there has been little research into estimating the coverage and mass of dry (non-photosynthetic) vegetation. In our study area at Tsogt-Ovoo (TsO) in the northern Gobi Desert, Mongolia, we used MODIS data to compare observed dry vegetation cover (VCd_obs) and mass (Md_obs) with values estimated from the Normalized Difference Vegetation Index (NDVI) and six other potential Short Wavelength InfraRed (SWIR) dry vegetation indices. Of the seven vegetation indices, the Soil Tillage Index (STI) and Normalized Difference Tillage Index (NDTI), both of which are calculated from SWIR bands 6 and 7, provided good estimates of VCd_obs and Md_obs. Our comparisons of observed roughness length (z0) and threshold friction velocity (u*t) for sand saltation at stony and non-stony sites with dry vegetation cover determined from STI verified its suitability for estimating VCd(STI). At both sites, the observed z0and u*tincreased with increasing estimated VCd(STI). Based on their relationships with STI, we predicted values of z0and u*tin the absence of dry vegetation that were in accordance with published data. The threshold wind speeds at 10 m height derived from u*tat both the stony and non-stony sites were reasonably close to the value that is commonly used in dust models. Thus, we verified that STI is a suitable index for estimating the amount of dry vegetation in the Gobi Desert and, potentially, in other arid and semi-arid regions. Although the relationships of z0and u*twith STI require further validation, it appears that the use of STI can improve future wind erosion risk assessments in arid and semi-arid environments.