Spatial distribution characteristics and scale effects of regional soil organic carbon

Abstract: Combined with the current research progress and shortcomings on spatial distribution characteristics and scale effects of soil properties, using a combination of variogram theory, spatial autocorrelation theory, multi-fractal theory and other methods from the aspects of the spatial variability of soil properties and the structure, the paper gradually revealed the spatial distribution characteristics of SOC (soil organic carbon) and its scale effects under 4 kinds of scales. The results showed that: Ratio of nugget to sill of SOC under 15, 25, 35 and 45 km scales were 61.78%, 42.65%, 49.98% and 39.34%, respectively, which showed a moderate spatial correlation. The spatial correlation from high to low was 45, 25, 35 and 15 km in turn. Except the 15 km scale, the spatial variability caused by random factors including representative measurement error and the micro-scale process, was less than the structure variance, namely the ratio of nugget to sill was less than 50%, the structural factors were dominant, and spatial variability from the random part showed a decreasing trend with the scale increasing. The variation function couldn’t be described with discrete characteristics (i.e. spatial negative correlation), which was also impossible to make a significant test for variable range. As the separation distance increased, the Moran index decreased from full positive to negative, then the positive and negative values alternately occurred, and eventually the values turned out to be negative totally. Nearer positive correlation distance represents the spatial correlation distance, which is the first cross point of positive and negative conversion under different scales, and were 1 607, 7 520, 8 649 and 9 053 m for 15, 25, 35, and 45 km scale. With the scale increasing, the spatial correlation distance increased, and compared to the change range in the semi-variation function, it became smaller with varying degree under different scales except 15 km scale. The order for standardized statistic under different scales was 15 km < 45 km <35 km < 25 km, which increased firstly and then decreased with the scale, and standard statistics under 4 scales were larger than 1.96. All random scale accounted for less than 5%, and on the whole each scale had good spatial structure. The multidimensional fractal parameter value under 25 km scale was the largest, indicating that the SOC content in this scale was mainly concentrated in a dense area. The magnitude of variation of Rayleigh dimension of SOC under 4 kinds of scales was consistent with that of width of multidimensional spectrum. The spatial distribution of SOC is a typical fractal dimension. Multifractal method is a tool for describing the spatial heterogeneity of SOC. It can reveal the scale variation characteristics of spatial heterogeneity of SOC. No matter what kind of scale, specific value of the measured value based on the Mkrige method agreed well with the predicted value, and the coverage ratio of specific value was above 85%. Spatial distribution characteristics of SOC were deeply revealed through combining variation function, spatial autocorrelation, multi-fractal and Mkrige method from spatial variability, spatial correlation and spatial structure in research area. The results can provide a method to the research on the spatial distribution characteristics of SOC in relatively flat agricultural region. However, due to the different combinations of artificial and natural factors, the spatial variability, correlation and structure will vary with scale change.