Multi-scale dynamics and environmental controls on soil respiration in the loess hilly region in China.

Accurate estimation of the spatial and temporal soil carbon dioxide (CO2) flux is vital for understanding the impacts of climate change on the soil carbon cycle. However, despite the expansion and refinement of weather stations, few statistical models have predicted soil respiration (SR) using meteorological data, particularly in deep soils. Thus, this study explored whether meteorological data can effectively predict SR at depths of 0-10, 10-50, and 50-100 cm, and examined the time-frequency patterns between SR and meteorological factors during 2005-2022 in the loess hilly region in China. The results showed that: (1) Within the three soil depths, the coefficient of determination ranged from 0.69 to 0.71, the root mean squared error varied from 0.10 to 0.17, and the Nash-Sutcliffe efficiency coefficient varied from 0.68 to 0.71. This finding indicated that the model based on dewpoint temperature demonstrated satisfactory reproducibility in predicting SR. (2) Cumulative CO2 efflux from 2005 to 2022 recorded 10,050.4 g·m-2 at 0-100 cm soil depths. Among these depths, the relative contribution of CO2 release was highest at 0-10 cm soil depth, reaching 72 %, followed by 10-50 cm and 50-100 cm, accounting for 17 % and 11 %, respectively; (3) Wavelet analysis revealed that SR exhibited discernible seasonal periodicity. Low temperature displayed the strongest intermittent correlation with SR in the 0-10, 10-50, and 50-100 cm soil depths, respectively. Those results show that the model based on dewpoint temperature characterizes the temporal variation in SR and demonstrates that low temperature significantly impacts SR.