Land condition diagnosis based on multi-resolution analysis and wavelet transform

Many change detection methods enable researchers analyze remotely sensed images and detect land degradation. However, these time series are strongly influenced by seasonal climatic variations, and some change detection methods cannot accurately detect changes within them. In order to detect various changes in different time scales in time series data, such as abrupt, seasonal, and gradual changes, as well as the noise generated by factors such as geometric errors and cloud effects, multi-resolution analysis (MRA) approach is strongly recommended. Wavelet transform is one of the most effective methods in multi-resolution analysis due to its convenience and effectiveness in decomposing non-stationary signals into variations in different temporal and spatial scales. The purpose of the present research is to detect changes within NDVI time series in different time scales to support analysis of land condition using MRA based on wavelet transform. Intra-and inter-annual vegetation dynamics signals are selected through MRA to discover vegetation changes in these two time scales. Then the inter-annual part is used to detect land condition indicated by inter-annual vegetation changes.