Grassland fractional vegetation cover analysis using small UVAs and MODIS——A case study in Gannan Prefecture

Gannan Prefecture was selected as the study area. Photos of different sized subplots of grasslands were taken with a camera (Canon) and small UAVs (unmanned aerial vehicles). Terra/MODIS MOD13Q1 vegetation indices were obtained from May to October 2015. Correlation analyses were conducted between vegetation indices enhanced vegetation index (EVI), normalized difference vegetation index (NDVI) and grassland fractional vegetation cover (FVC), and regression models were determined. Based on the accuracy of different regression models, the best estimation model for grassland FVC in Gannan Prefecture was selected, and grassland spatio-temporal characteristics of FVC during the growing season were analysed. The results showed that 1) small UAVs carrying cameras could be used to obtain digital photos containing data on grassland FVC; 2) EVI was more correlated with grassland FVC than NDVI, and the logarithmic model based on EVI was the most suitable model for monitoring grassland FVC in Gannan Prefecture, with model accuracy reaching 88.00%; 3) grassland FVC of flat meadows in 2015 reached the maximum in August, and the remaining grassland types reached the maximum in July; 4) grasslands in Gannan Prefecture were dominated by high and mid-grassland FVC levels, and these grasslands were mainly distributed in Maqu County, Luqu County, and Xiahe County and Hezuo City. Overall, the grassland FVC in the mid-western and south-western areas was better than that in the eastern area. Establishment of the best model for grassland FVC was beneficial not only in elucidating the spatio-temporal distribution and seasonal dynamic variation of grassland FVC, but also to the maintenance of stable development of grassland ecosystems in Gannan Prefecture.