Research on relationship between vegetation cover fraction and vegetation index based on flexible spatiotemporal data fusion model

The spatiotemporal data fusion model has been widely used to obtain high temporal and spatial resolution vegetation indexes and vegetation cover fractions, but its accuracy is often affected by low spatial resolution images e.g., moderate resolution imaging spectroradiometer (MODIS) images. This study design was based on the flexible spatiotemporal data fusion model (FSDAF), and investigated the effects of three different MODIS image pairs of the FSDAF model for vegetation cover fraction extraction in arid region of China. Furthermore, the linear and non-linear relationships between six vegetation indexes and vegetation cover fractions were investigated. The results showed that the retrieval accuracy of the vegetation cover fraction of FSDAF simulated images depended on the rate of change of the MODIS images in two periods. In addition, the accuracy of the image with a slight change was significantly higher than that of the image with a great difference. When vegetation indexes were used to simulate the vegetation cover fraction, the normalized difference vegetation index (NDVI) and green NDVI (GNDVI) linear fitting methods produced better results than other methods did, and provided the ideal results. The experimental results showed that the FSDAF model could be used to determine the retrieval of vegetation cover in arid area, and it had a good effect and applicability.