Super-resolution mapping of coastline with remotely sensed data and geostatistics
结合遥感数据与地统计学方法的海岸线超分辨率制图

The soft classification results of coarse-resolution data and the coarse & fine resolution variograms were derived, using the Landsat satellite data of the modern Yellow River Delta coast. Treating the fine-resolution indicator vriogram as the prior model of spatial structure of the study area, the super-resolution images of land & ocean class were generated using data exploratory analysis, indicator cokriging (ICK), and sequential indicator co-simulation (SIcS) techniques. Then the spatial distribution features of the coastline were extracted by the contouring method. Taken human interpretation output as the benchmark, the coastline mapping results derived from geostatistics showed better quality than that derived from traditional hard-classification methods. Super-resolution mapping techniques based on geostatistics can properly illustrate spatial distribution of coastline at fine scale; meanwhile maintain the class fraction values and the spatial structures of the original coarse-resolution data. By the form of super-resolution mapping of coastline, the potential of the geostatistical techniques in integrating multi-source and multi-scale spatial information has been demonstrated.