Evaluating soil heterotrophic respiration simulated by microbially-explicit global models

Abstract:

Microbially-explicit models have been developed to improve understanding carbon dynamics and forecasting future climate change, however, their accuracy and uncertainties in simulating global-scale heterotrophic respiration (RH) spatio-temporal variabilities have not been comprehensively evaluated. Here, we used statistical global RH products, as well as 7,955 daily RH measurements from 258 studies across the globe, to evaluate the performance of biogeochemical models. We found that all models well predict the magnitude and trend of global annual RH, although differ by ~5 Pg C yr-1. Spatial RH latitutional variability was well simulated except in the northern middle latitudes (~50 ° N), where the biogeochemical models tend to overestimate RH fluxes compared to observational based statistical models. Many of these problems were driven by biases in model inputs, especially related to NPP and litterfall fluxes. Biogeochemical models overestimate at ~68% of observational sites, underestimate it at ~26% of sites, and provide accurate estimates at only ~6% of sites. Our results thus demonstrate that the next generation of biogeochemical models show promise, but need to be improved for realistic spatio-temporal variability of RH. Finally, we emphasize that high spatio-temporal resolution RH data are urgent for benchmarking and improving biogeochemical models.