Addressing Uncertainity in Global Soil Respiration Estimates By Quantiying Sources of Bias.

Soil respiration (Rs) and its response to temperature are key to quantifying terrestrial carbon stocks and predicting the feedbacks to climate change. Currently, estimates of global Rs range from 68 to 98 Pg C yr^-1, resulting in considerable uncertainty in carbon fluxes. Temperature and precipitation based models were parameterized based on a monthly global soil respiration database to identify a more constrained range of Rs. We estimated Rs from three variations of climate-driven models and different climate data input scenarios. Our global annual mean Rs estimates range from 66.62 Pg to 100.72 Pg. Of the three models, the second order exponential with hyperbolic precipitation function model was the most conservative, while the first order exponential model provided the largest estimates. The four sources of uncertainty contributed between -4.37 to 12.35 Pg C yr^-1. Using annual averaging temperature represent monthly temperature (Jensen’s inequality) underestimated global Rs by -4.37 Pg, while using average precipitation to represent monthly precipitation overestimated Rs by 4.85 Pg. Compared to other models, using hyperbolic precipitation models caused 9.85 Pg of uncertainty in Rs. Parameterized models on yearly timescale Rs samples rather than monthly scales caused 7.93 Pg of uncertainty. Lastly, including a threshold, or not, caused 12.35 Pg of uncertainty in Rs. We tested for a threshold and found that in the monthly timescale data, compared to the first-order, the second-order exponential models was the better fit. After accounting for these sources of uncertainty within the estimates, we find that the second-order exponential Rs model with monthly timescale data estimate (80.99 Pg C yr^-1) provides the most accurate estimate of global scale Rs and is closer to Rs estimated by carbon flux accounting (81.65 Pg C yr^-1). These results help to constrain ecosystem models and aid researchers in quantifying the uncertainty associated with quantifying global carbon cycling.