Application of predictor variables to support regression kriging for the spatial distribution of soil organic carbon stocks in native temperate grasslands

Purpose As the main component of terrestrial carbon pool, soil organic carbon (SOC) is vital to soil fertility and biogeochemical cycle. Quantifying the spatial distribution of regional SOC stocks (SOCS) provides critical support for climate change and food security decisions. Our aim was to explore the optimal interpolation method to improve the accuracy of spatial prediction of SOCS in temperate grasslands. Materials and methods To support such research, we performed soil sampling to depths of 0 to 20 and 20 to 30 cm throughout the Hulun Buir grassland of Inner Mongolia. We compared prediction of the spatial patterns of SOCS using regression kriging (RK) and ordinary kriging (OK). We used topographic factors, climate variables, satellite data (the normalized-difference vegetation index (NDVI)), and soil texture as predictors in the RK method. Results and discussion SOCS was significantly positively correlated with precipitation, NDVI, topographic variables, and clay content, but negatively correlated with temperature and sand content. NDVI explained more than 40% of the SOCS spatial variation and was the dominant factor. Geostatistical analysis showed strong and moderate spatial dependence of SOCS in the 0-20- and 20-30-cm soil layers, respectively. The RK and OK soil pools to a depth of 30 cm were 607.28 and 559.46 Tg, respectively. Conclusion Compared with OK, the RK method improved the SOCS prediction accuracy by 20.4, 30.1, and 23.9% for soil depths of 0-20, 20-30, and 0-30 cm, respectively. Our findings suggest that OK may be acceptable where the environmental conditions are homogeneous, but that RK performs better in heterogeneous areas.

Automated summary

This study aims to improve the accuracy of spatial prediction of soil organic carbon (SOC) in temperate grasslands by exploring the optimal interpolation method. The results show that compared to ordinary kriging (OK), regression kriging (RK) performs better in heterogeneous areas, resulting in higher prediction accuracy. The normalized-difference vegetation index (NDVI) is identified as the dominant factor influencing the spatial variation of SOC.