Monitoring dry vegetation masses in semi-arid areas with MODIS SWIR bands

Monitoring the mass of herbaceous vegetation during the dry season in semi-arid areas is important for a number of domains in ecology, agronomy, or economy and remote sensing offers relevant spatial coverage and frequency to that end. Existing remote sensing studies dedicated to dry herbaceous vegetation detection are mainly motivated by the assessment of soil tillage intensity and soil residue management, risk of soil erosion, and risk of wildfire linked to the mass of dead fuel. Few studies so far have dealt with monitoring of straw and litter degradation during the dry season over large areas while they are important fodder for livestock sustainability. MODIS band combinations (NBAR collection 5) were tested against a set of field measurements carried out over 20 rangeland sites from 2004 to 2011 in the Sahel. The best empirical linear models were obtained for indices using MODIS bands in the shortwave infrared domain (Band 6 centered at 1.6 mu m, Band 7 centered at 2.1 mu m), in particular with the Soil Tillage Index (STI). STI explained 66% of the variance of dry masses (Mass = 3158(STI - 1.05), r(2) = 0.66, RMSE = 280 kg DM/ha, n = 232) for dry and intermediate season data. A regression is also proposed for year-round data (Mass = 3371(STI - 1.06), r(2) = 0.67, RMSE = 352 kg DM/ha, n = 536). The strong inter-site and inter-annual variabilitieswere well captured and the decay rate was found consistent with grazing intensity and fire occurrence. The results imply that the STI can be applied to monitor the mass of dry tissues in the Sahel and potentially in many semi-arid areas. (C) 2014 Elsevier Inc. All rights reserved.