Abstract: Taking the Hetao irrigation region in Inner Mongolia as the research subject, with methods of classical statistics and geostatistics, we analyzed the spatial variability and scale effect of soil organic matter at different soil layers (0-20, 20-40, 40-70 and 70-100 cm) and sampling scales (1, 4 and 8 km). The results of classical statistics showed that the variation of mean organic matter contents increased with the increase of soil depth at all sampling scales; the coefficients of variations of organic matter contents at soil depths of 0-20 cm and 20-40 cm increased with increasing the sampling scale. Geostatistics indicated that there existed strong spatial autocorrelations in soil organic matter contents in different soil layers and at different sampling scales, and soil type was the dominant factor influencing its spatial distribution. There were certain anisotropic effects in the spatial patterns of organic matter content at different soil layers at all sampling scales. At small scale (1 km), this anisotropic effect presented a strip variation on eastwest direction at each soil layer. At moderate (4 km) and large scale (8 km), corresponding spatial variability was intensive in terms of eastwest and northwestsoutheast gradients at the top soil layer. The cross validation of ordinary Kriging applied on organic matter at the three scales showed that the root mean square errors were less than 1, indicating that the spatial variability of samples was overestimated. Our results highlight great significance for understanding the spatial distribution of soil organic matter, providing a scientific basis for sampling system design in agricultural technology research.

Key words: principal component analysis (PCA), Weinan City, remote sensing, ecological environment quality, remote sensing based ecological index (RSEI)