Analysis on the Representation of the Station Observation Based on the Land Surface Spatial Heterogeneity

The temperature based (T-based) validation method is the only direct validation method for evaluating the remote-sensing LST product, which needs ground-based LST measurements with a pixel representativeness. Therefore, in the process of using T-based method, the observation representation of the stations directly determines whether the validation result is reasonable. The stations located at a more homogeneous area will have a more representative observation for remote-sensing pixel, and in contrast, the station located at a heterogeneous will have a less representative observation. Therefore, in this section, we have implemented the land surface heterogeneous status analysis for the stations to assess the representation of stations. Under the same conditions, the heterogeneity of LST in the same region is mainly affected by the surface coverage of the underlying surface and the condition of vegetation growth. Therefore, this chapter uses the NDVI inversion date from the 16-day data of Landsat 8 and land use types data were selected to analyze the surface heterogeneity of the stations. In this research, the variogram method was used for heterogeneity analysis of the station-based observation and the degree of dispersion within the spot is measured by the coefficient of variation (CV). Based on the surface heterogeneity analysis, the representativeness of the stations-based observations was comprehensive grade, which sets the foundation for selecting the observations with represented to validate the remote sensing-based LST.