Unveiling the Explanatory Power of Environmental Variables in Soil Organic Carbon Mapping: A Global–local Analysis Framework

Yujiao Wei,Yiyun Chen,Jiaxue Wang,Bo Wang,Peiheng Yu,Yongsheng Hong,Liandong Zhu

DOI: https://doi.org/10.1016/j.geoderma.2024.117011

IF: 6.1

2024-01-01

Geoderma

Abstract:Soil organic carbon (SOC) is a critical component that affects soil quality and global carbon cycling. Current SOC mapping approaches are based on the spatial stationarity relationship of SOC and soil formation processes. Nevertheless, the spatial pattern of SOC is the consequence of different soil-forming factors and processes that operate at different scales. In this work, we hypothesized that the covariation of environmental variables and SOC might differ spatially, and proposed a global (whole area) and local analysis framework that aimed to enhance our comprehension of the explanatory scale of environmental variables on SOC variation. This framework primarily incorporates Geographically Weighted correlation and the Multi-scale Geographically Weighted Regression (MGWR) model. With 216 farmland topsoil samples collected from the Qilu Lake watershed in Yunnan Province, China (area of 354 km2), we explored both global and local relationships between environmental variables and SOC to verify the feasibility of this framework. Results showed that the explanatory power of environmental variables on SOC variation is scale-dependent. Our analysis revealed that certain variables, which may explain local variations of SOC, are often overlooked due to their insignificant global correlation with SOC (p > 0.05). For example, in this case study, soil porosity and two landscape metrics that characterize the anthropogenic processes of land use patterns can effectively explain the local spatial variation of SOC. They improved the model performance of MGWR, but their global correlation with SOC is not significant. The proposed framework highlights the necessity of investigating the explanatory power of environmental variables on a global and local scale.