B21G-0550
Limitations of Vegetation Indices For Detecting Pasture Degradation: A Case Study of Montane Pastoral Systems in Central Asia

Tuesday, 15 December 2015
Poster Hall (Moscone South)
Ian Murray Sember Eddy, Forest Sciences Centre, Vancouver, BC, Canada
Abstract:
Grazing is the most extensive land use on Earth. Widespread consequences of overgrazing pastures include long-term decreases in plant biomass and limited recovery of vegetation. Remotely-sensed vegetation indices linked to biomass (e.g. NDVI) are routinely used to monitor pasture health over broad areas to track pasture degradation and recovery over time. Unfortunately, overgrazing can impact vegetation in various other ways not easily evaluated using satellite imagery, such as by altering species composition. Furthermore, the response of vegetation to grazing may be influenced by underlying terrain and topographic gradients.

We examined multi-decadal trends in pasture condition in Kyrgyzstan, a country where pasture degradation is of serious concern. Using a chronosequence of Moderate-Resolution Imaging Spectroradiometer (MODIS) imagery, we compared fifteen-year trends in NDVI with contemporary field-based measurements of pasture health in thirty 1-km 2 sites. Multivariate regression was used to discern the relationship between long-term NDVI trends and pasture health in pastures of differing terrain (areas of varying topographic wetness index and solar insolation). Preliminary results suggest that pasture degradation can be correlated with either positive or negative changes in NDVI depending upon the topographic position of the pasture. Furthermore, terrain characteristics explained a considerable portion of the observed variance in NDVI trends across the region. Improving our understanding of grazing impacts in montane systems is critical given their vulnerability to impending climate change.