[A-CG39] Biogeochemical cycles in Land Ecosystem

[ACG39-10] Effects of soil properties in estimating soil respiration and methane absorption

*Hina Yamanuki¹, Kazuhito Ichii¹, Yuhei Yamamoto¹, Munemasa Teramoto², Sun Lifei³, Jun Koarashi⁴, Mariko Atarashi-Andoh⁴, Hirohiko Nagano⁵, Takashi Hirano⁶, Kentaro Takagi⁶, Sachinobu Ishida⁷, Masahiro Takagi⁸, Toshiaki Kondo⁹, Yoshiyuki Takahashi³, Naishen Liang³ (1.Center for Environmental Remote Sensing, Chiba University, 2.Arid Land Research Center, Tottori University, 3.National Institute for Environmental Studies, 4.Japan Atomic Energy Agency, 5.Niigata University, 6.Hokkaido University, 7.Hirosaki University, 8.University of Miyazaki, 9.Japan International Research Center for Agricultural Sciences)

Keywords:Soil respiration, Methane absorption, Soil properties, 14C, Machine learning

Soil exchanges greenhouse gases (GHGs) such as CO₂ and CH₄ with near-surface atmosphere, and these exchanges are one of the essential components of soil CO₂ and CH₄ cycles. In general, forest soils act as a CO₂ source via emission through soil respiration and CH₄ sink via absorption by the soil. Various efforts have been made to observe soil CO₂/CH₄ fluxes, analyze their mechanisms, and upscale them. However, due to insufficient observation and understanding, significant uncertainties remain in their spatio-temporal variations and their underlying mechanisms. For example, various efforts on upscaling soil respiration have been conducted using machine-learning models or semi-empirical models, and further improvements were expected by denser observation networks and/or adding parameters relating to soil characteristics.

In this study, we analyzed the effects of observed soil properties on explaining spatio-temporal variations in observed soil respiration and CH₄ absorption. With the largest soil respiration observation network across Asia developed and maintained by National Institute for Environmental Studies (NIES) and several observed soil properties such as ¹⁴C, organic matter properties, and mineral properties by Japan Atomic Energy Agency (JAEA), we estimated soil respiration and methane absorption across Japan based on random forest regression. Adding ¹⁴C data to the explanatory variables for CO₂ estimation improved the accuracy of the estimation compared to the model based on meteorological data only. Mineral properties data were important for CH₄ estimation, and the importance of mineral properties data as explanatory variables was greater than those of soil meteorological data such as temperature and soil moisture. We revealed that soil properties, in addition to meteorological data, are essential for estimating soil CO₂/CH₄ fluxes.