Experimental and Model Estimates of Respiration of the Forest Sod-Podzolic Soil in the Prioksko-Terrasny Nature Reserve

Abstract

Modeling the carbon cycle in forests is often restricted by modeling the main components, including emissions (respiration of soils and debris) and production (deposition of carbon in plants and soils). In this study we examine the applicability of various versions of the T&P model to estimate monthly, seasonal, and annual fluxes of CO2 from the sod-podzolic soil in the mixed forest of the Prioksko-Terrasny Nature Reserve, Moscow oblast. The model is parameterized and verified, and the accuracy is tested using a database of 20 years of monitoring CO2 emissions from soils and independent weather variables, including mean monthly air temperature and monthly precipitation. Numeric experiments show that all versions of the T&P model (both initial and parameterized by training sets at different temporal intervals) satisfactorily describe the long-term dynamics of mean monthly respiration of the sod-podzolic soil under forest cover (SRm). Parameterization of the T&P model with experimental data as a training set practically does not improve the quality of modeling in any of the test intervals. The use of weather data averaged over 20 years for the calculation of SRmod-mean and estimates of seasonal and annual soil fluxes of CO2 on their basis (SeSRmod-mean) in most cases overestimates the corresponding experimentally obtained values (SeSRexp). SeSRmod-mean for annual, summer, and winter soil CO2 fluxes are on average 4.5–6.7% higher than SeSRexp, and SeSRmod-mean for the warm season shows an overestimation of approximately 3%. The largest discrepancy of calculated estimates to experimental data is found for the spring period: the overestimation amounts to ~22%. Thus, the use of weather data averaged over 20 years has shown the applicability of an ensemble of versions of the T&P model for estimating seasonal and annual fluxes of CO2 from soil in a humid continental climate.