Soils apart from equilibrium – consequences for soil carbon balance modelling

Abstract

Abstract. Many projections of the soil carbon sink or source are based on kinetically defined carbon pool models. Parameters of these models are often determined in a way that the steady state of the model matches observed carbon stocks. The underlying simplifying assumption is that observed carbon stocks are near equilibrium. This assumption is challenged by observations of very old soils that do still accumulate carbon. In this modelling study we explored the consequences of the case where soils are apart from equilibrium. Calculation of equilibrium states of soils that are currently accumulating small amounts of carbon were performed using the Yasso model. It was found that already very small current accumulation rates cause big changes in theoretical equilibrium stocks, which can virtually approach infinity. We conclude that soils that have been disturbed several centuries ago are not in equilibrium but in a transient state because of the slowly ongoing accumulation of the slowest pool. A first consequence is that model calibrations to current carbon stocks that assume equilibrium state, overestimate the decay rate of the slowest pool. A second consequence is that spinup runs (simulations until equilibrium) overestimate stocks of recently disturbed sites. In order to account for these consequences, we propose a transient correction. This correction prescribes a lower decay rate of the slowest pool and accounts for disturbances in the past by decreasing the spinup-run predicted stocks to match an independent estimate of current soil carbon stocks. Application of this transient correction at a Central European beech forest site with a typical disturbance history resulted in an additional carbon fixation of 5.7±1.5 tC/ha within 100 years. Carbon storage capacity of disturbed forest soils is potentially much higher than currently assumed. Simulations that do not adequately account for the transient state of soil carbon stocks neglect a considerable amount of current carbon accumulation.

Bayesian calibration of a soil organic carbon model using Δ<sup>14</sup>C measurements of soil organic carbon and heterotrophic respiration as joint constraints

“…5b). This is in line with considerations by Wutzler and Reichstein (2007) who found that for soils that have not reached (and are below) their equilibrium stock, model calibration to the current carbon stock overestimates the decomposition rate of the slowest pool. They propose a transient correction which prescribes a lower decomposition rate for the old pool.…”

“…However, Schmidt et al (2011) proposed that the 14 C content of respired CO 2 and leached dissolved organic carbon could be used as additional constraints in model-data comparisons. Wutzler and Reichstein (2007) have shown that the commonly used equilibrium or steady-state assumption of many SOC models may lead to biased estimates of SOC turnover times. For a soil with SOC stocks below equilibrium, a calibration of turnover times assuming SOC stocks at equilibrium would yield too-fast turnover time estimates.…”

“…For a soil with SOC stocks below equilibrium, a calibration of turnover times assuming SOC stocks at equilibrium would yield too-fast turnover time estimates. In their modeling study Wutzler and Reichstein (2007) proposed a transient correction for decay rates to account for possible disturbances in the past. In the model-data comparison framework we propose, we tackled this issue from a different perspective by introducing and calibrating parameters relaxing the steady-state assumption.…”

“…The most common way to deal with initialization problems of conceptual and nonmeasurable SOC pools is to perform spin-up runs of the model in an undisturbed environment. Then estimates about initial SOC pools are retrieved based on a reconstructed disturbance history (Falloon and Smith, 2000;Wutzler and Reichstein, 2007;Yeluripati et al, 2009). Due to its simplicity the steady-state equations for the ICBM can still be derived relatively easily: (start − tlag (L,R) ) was taken as the initial 14 C signature of litter input, where start denotes the starting year of simulations.…”

Data-mining analysis of the global distribution of soil carbon in observational databases and Earth system models

“…Furthermore, SOC storage occurs via organic matter accumulation over decades and even millennia. Thus, prior NPP, land fires, and land-use changes may still affect current SOC values (Carvalhais et al, 2008;Wutzler and Reichstein, 2007). Land cover is another important factor, and incorporating the hydrology and resulting carbon dynamics in wetlands may lead to important improvements in ESMs.…”