Observation‐based global soil heterotrophic respiration indicates underestimated turnover and sequestration of soil carbon by terrestrial ecosystem models
He, Yue et al. (2022), Observation‐based global soil heterotrophic respiration indicates underestimated turnover and sequestration of soil carbon by terrestrial ecosystem models, Dryad, Dataset, https://doi.org/10.5061/dryad.b2rbnzsj9
Soil heterotrophic respiration (Rh) refers to the flux of CO2 released from soil to atmosphere as a result of organic matter decomposition by soil microbes and fauna. As one of the major fluxes in the global carbon cycle, the estimation of global Rh still exists large uncertainties, which further limited our current understanding of the carbon accumulation in soils. Here, we applied a Random Forest algorithm to create a global dataset of soil Rh, by linking 761 field observations with both abiotic and biotic predictors. We estimated that the global Rh was 48.8 ± 0.9 Pg C yr-1 for 1982–2018, which was 16% less than the ensemble mean (58.6 ± 9.9 Pg C yr-1) of 16 terrestrial ecosystem models. By integrating our observational Rh with independent soil carbon stock datasets, we obtained a global mean soil carbon turnover time of 38.3 ± 11 yr. Using observation-based turnover times as a constraint, we found that terrestrial ecosystem models simulated faster carbon turnovers, leading to a 30% (74 Pg C) underestimation of terrestrial ecosystem carbon accumulation for the past century, which was especially pronounced at high latitudes. This underestimation is equivalent to 45% of the total carbon emissions (164 Pg C) caused by global land use change at the same time. Our analyses highlight the need to constrain ecosystem models using observation-based and locally adapted Rh values to obtain reliable predictions of the carbon sink capacity of terrestrial ecosystems.
This dataset provides a global gridded product at 0.5-degree resolution of predicted annual soil heterotrophic respiration (Rh) during 1982–2018. A Random Forest (RF) approach was used to derive the predicted Rh trained with 761 observations with 19 predictors (including climate, vegetation, soil biotic and abiotic variables). To improve the RF model accuracy, we developed a stratified 10-fold cross-validation, by grouping our dataset into three climate zone classes (i.e., tropical, temperate and boreal zones) and ensuring each class was approximately equally represented across each fold. The average predicted map across the RF model ensemble was used as the final product.
National Natural Science Foundation of China, Award: 41988101
National Key R&D Program of China, Award: 2019YFA0607304
National Natural Science Foundation of China, Award: 42022004
National Natural Science Foundation of China, Award: 41901085
Second Tibetan Plateau Scientific Expedition and Research Program, Award: 2019QZKK0606