Spatial distribution pattern of soil organic matter in the wind erosion region of northeastern China based on the cokriging method

Abstract

Soil wind erosion triggers complex particle migration from the source to downwind regions. During this process, soil organic matter (SOM), primarily adsorbed onto particle surfaces, exhibits distinct spatial differentiation across affected areas. Knowledge about the spatial distribution of SOM and its impact factors in wind erosion regions is essential for a deep understanding of global carbon cycling. In the wind erosion region of northeastern China, a crucial base for livestock and commercial grain production, profound changes in land quality driven by climate change and human activities have intricately altered SOM spatial patterns. However, detailed studies on the spatial variability of surface SOM in this region remain scarce. A total of 374 surface soil samples from this region were analyzed for SOM and soil pH. Combined with spatial data sets including particle-size distribution, soil moisture (SM), air temperature (T), and the normalized difference vegetation index (NDVI), the regional spatial distribution patterns of SOM were revealed and mapped using the ordinary kriging (OK) and cokriging (CK) approaches. The results indicated strong variations of SOM in space, with corresponding coefficients of variation exceeding 35 % across the entire region and each land use. According to the semivariogram analysis, SOM demonstrates moderate spatial dependency in the study region, exhibiting spatially positive correlations with clay content (CLAY), SM, and NDVI, and negative correlations with pH, soil sand content (SAND), and T. Among the various impact factors, incorporating SAND as an auxiliary variable in CK interpolation achieved the greatest improvement in the accuracy of SOM spatial distribution simulation compared to the OK model, increasing the explained variance from 35 % to 75 %. This simulation showed a general downward trend of SOM from the northeast to the southwest of the study region, with 65.24 % of the area displaying SOM ranging from 10-30 g·kg⁻¹. The values of SOM greater than 30 g·kg⁻¹, comprising 19.43 % of the total region, were primarily located in the humid and semi-humid meadow steppes and forests. In contrast, areas with SOM below 10 g·kg⁻¹ accounted for only 15.33 %, mainly situated in the arid desert steppe and the Horqin sandy lands. These findings provide a scientific basis for understanding regional SOM loss and developing efficient measures for sand prevention and control.