An advantage of using the modeled soil temperature and moisture data for soil respiration empirical modeling

Soil respiration (SR) is one of the main pathways for ecosystem respiration and carbon losses. To conduct the SR estimations on the global scale, the SR modeling becomes a viable alternative for the direct SR measurements. One way to conduct SR modeling is to apply an ensemble of empirical models for selecting better condition-specific models. This approach is applied to simulate the data of the 25-year SR monitoring in forest and grassland ecosystems situated at the same bio-climate zone on the similar Entic Podzol (Arenic). We investigated the effects of the weather- and site-specific conditions as well as parameterizations by different sources of temperature and moisture data, namely: Tsoil, Tair, Prec are the directly measured soil/air temperature and precipitation and Tsoil_m, Msoil_m, Tlit_m, and Mlit_m are the SCLISS-modeled soil/litter temperature and moisture.

We found that for the forest soil with higher soil-organic-carbon (SOC) storage and good sun or wind protection due to tree canopy, it is better to use the models dependent on Tsoil and SOC, while for the grassland soil with the direct solar heating it is better to use Raich-Hashimoto type models with the quadratic temperature dependency. The condition-specific models always showed better results in the slope, determination coefficient, mean bias error, and root mean square error than the models parameterized for all conditions at once. When the direct Tsoil&Msoil measurements are not available for the SR model parameterizations, good suitability is shown by the Tlit_m&Mlit_m parameterizations confirming the important role of litter layer as a source of SR in both forest and grassland. The application of Tlit_m&Mlit_m for parameterizations allows for an expansion of using empirical SR models customized for different soils to the global scale.

The study was funded by State assignment of Federal Research Center of the Russian Academy of Sciences (#122040500037-6 and #122111000095-8).