Using NDVI time series to diagnose vegetation recovery after major earthquake based on dynamic time warping and lower bound distance

Major earthquake occurred in mountainous areas usually cause large number of landslides that lead to severe impact to local vegetation cover and growth. The negative influence to vegetation may last for many year and vegetation recovery may experience dynamic fluctuation. Existing methods for vegetation recovery diagnosis face difficulty in capturing the dynamic behaviours both within and between years that makes the interannual comparison impossible. This paper proposes a new method to diagnose regional vegetation recovery after a major earthquake by defining a difference measurement index (DMI) using MODIS NDVI time series at 8-day interval. This differs from many existing methods in its quantification of the difference between the studied time series and historical samples, by using a proposed algorithm consisting of lower bound distance and dynamic time warping. This algorithm can better differentiate vegetation disturbance from its natural fluctuation. Second, the method investigates relatively regional vegetation recovery via a dynamic index, the DMI. Vegetation conditions in different years can be compared with a historical benchmark and measured by DMI. This makes it possible to diagnose dynamic vegetation recovery and generate a series of interannual spatial distributions of regional vegetation state.