A New Spatial Downscaling Algorithm to Generate A Long-Term Global 250-m Resolution NDVI Product

Satellite-based Normalized Difference Vegetation Index (NDVI) time-series data are useful for monitoring the changes of vegetation ecosystems in the context of global climate change. However, there are currently no ideal NDVI datasets that reconcile long-term series with high spatial resolution. Here, we have developed a simple and novel data downscaling algorithm based on the coefficient of variation (CV) statistics, which combines the detailed spatial features of MODIS data with the long-term temporal information of AVHRR data. The proposed data fusion method helps generate a global monthly NDVI database that has a 250 m-resolution and covers the long period of 1982−2018. We evaluated the accuracy of the fused data against MODIS NDVI using the metrics of Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Pearson’s correlation coefficients (R). Validation suggests a high performance of the downscaling algorithm and a high accuracy of the new NDVI database. We further applied the downscaled data to monitor NDVI changes of various vegetation types and in areas having high vegetation heterogeneity, and we obtained stable results similar to MODIS data. The whole data downscaling and validation processes were completed on the Google Earth Engine platform, and here we provide a code for users to easily get the data for any part of the world. The downscaled global-scale NDVI time series has high potential in monitoring the temporal and spatial dynamics of the terrestrial ecosystems under changing environments.