Improving change vector analysis in Multi-temporal space to detect land cover changes by using cross-correlogram spectral matching algorithm

Traditional Change Vector Analysis in Multi-temporal space (TCVAM) can effectively extract land cover change information based on VI time series, and it has been one of the main methods to detect land cover change at large scale. However, the TCVAM may exaggerate the change information and mix the land cover conversion and land cover modification because of the oversensitivity to the changes of VI values. The paper proposes an Improved Change Vector Analysis in Multi-temporal space (ICVAM) based on cross-correlogram spectral matching algorithm and applies it in the Beijing-Tianjin-Tangshan urban agglomeration district, China, using MODIS_EVI time series data to test the performance of the ICVAM. The results demonstrated the improvement of the ICVAM compared to the TCVAM: overall accuracy increased by 10.80% and the kappa coefficient increased by 0.13. The ICVAM has great potential to be widely used for land cover change detection based on VI time series at large scale.