Spatial variation of correlations between vertical soil water and evapotranspiration and their controlling factors in a semi-arid region

Abstract

A decrease in soil water content (SWC) has been observed at the global scale and has led to a reduction in evapotranspiration (ET). However, the trend in SWC in eastern Asia differed from the global trend, and this may be due to the overlapping of warm/wet seasons. Because of the limitation of deep soil water content data, especially in drylands, changes in SWC and the correlation between SWC (particularly deep SWC) and ET remain poorly understood. In this study, spatial distribution of correlation coefficient (r) between SWC and ET were calculated based on reanalysis data and remote sensing data, which were validated by in situ data (2004–2005) and sampling data (2016–2017). Firstly, the correlation between SWC and ET decreased with soil depth. The depth of the sharp decrease was 10 cm based on in situ monitoring data (2005–2008) and spatial sampling data (2016–2017). Secondly, the spatial correlation between SWC and ET varied with different vegetation cover types. The correlation increased from the western typical steppe to the eastern forest. The highest values of r were in a forest-covered area, because it had the highest water uptake capacity. However, the difference between r_10 and r_100 (D_r) showed a different spatial pattern, and the lowest negative D_r value was obtained from a forest–grass transition zone in which the enhanced vegetation index showed a significant increasing trend. The significant negative regression determination coefficient (R2 = 0.9, p < 0.01) between B_EVI (0–0.1) and D_r indicated that the narrowed D_r in the transition zone was due to the vegetation increase. This meant that more deep soil water was taken up into the atmosphere. Thus, the deep SWC could become unstable and deep soil could become drier, which would be unsustainable for the ecosystem.