Adaptive local kriging (ALK) to retrieve the slant range surface motion maps of Wenchuan earthquake

Kriging technique is widely used for raster data interpolation, such as digital elevation models and precipitation maps. It depends on the spatial distribution of sampled values and the variogram model which fit an empirical global semivariogram of entire image. However, such a variogram model may not suitable for datasets with strong trends of the spatial distribution, such as differential interferometric synthetic aperture radar (DInSAR) data. Here we propose a modified kriging method, adaptive local kriging (ALK), for retrieve of the missing data in the decoherence zone of the ALOS PALSAR DInSAR data of 2008 Wenchuan earthquake. In the ALK, dynamic linear variogram models fitting the local empirical semivariograms are used instead of a global variogram model for the whole image. The localized adaptive approach avoids drastic errors in the large decoherence gaps and the value prediction is thus significant improved as confirmed in comparison with the original DInSAR data.