Fractional vegetation cover (FVC) is one of the most critical indicators for herbaceous wetland vegetation status, degradation and desertification process simulations. The
dense in time series and high spatial resolution of FVC is required for wetland ecosystem monitoring because of its heterogeneous landscapes and rapid spatio-temporal variations. However, due to the tradeoff in satellite sensor designs, it is hard to acquire both high temporal and spatial satellite images for FVC estimation for wetland ecosystem. In this paper, the dense in time series HJ composites at 30-m spatial resolution and UAV platform were used for the estimation of time series FVC in Zoige wetland area. Considering the spatial variability of soil backgrounds for peat wetland area, an improved adaptive endmember selection linear spectral mixture (LSMM) model was proposed in this paper. The results revealed that the proposed method can provide a higher estimation accuracy than the background invariant LSMM model, and the time
series FVC estimation result is helpful to reflect both the spatial pattern and temporal variation characteristics of heterogeneous wetland regions.