|DANGAL, SHREE - Woods Hole Research Center|
|SCHWALM, CHRISTOPHER - Woods Hole Research Center|
|SANDERMAN, JONATHAN - Woods Hole Research Center|
Submitted to: Journal of Advances in Modeling Earth Systems
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/22/2022
Publication Date: 4/5/2022
Citation: Dangal, S.R., Schwalm, C., Cavigelli, M.A., Gollany, H.T., Jin, V.L., Sanderman, J. 2022. Improving soil carbon estimates by linking conceptual pools against measurable carbon fractions in the DAYCENT Model Version 4.5. Journal of Advances in Modeling Earth Systems. https://doi.org/10.1029/2021MS002622.
Interpretive Summary: Models help us understand climate change but a major challenge in these models is accurately representing soil processes that affect soil organic carbon (SOC) storage. Current models often use assumed, instead of measured, values for different types of SOC, each of which have different rates of decomposition and stability. Woods Hole researchers updated the DAYCENT model with measured SOC data obtained from several long-term USDA-ARS research locations, then compared the modified model (DCmod) with the default model (DCdef) to evaluate how SOC responded to current and future scenarios of climate change in the US Great Plains region where agriculture is a major land use. The DCmod improved predictions of current SOC compared to DCdef and suggested that soils in managed grasslands and croplands today have lost 4% and 40%, respectively, of the original SOC in native Great Plains soils. Losses of SOC were predicted under all future modeling scenarios, with much greater losses predicted by DCmod than by DCdef. These results will be of great interest to other scientists and to policy makers striving to adapt agricultural and climate change policies in a rapidly changing world.
Technical Abstract: Terrestrial soil organic carbon (SOC) dynamics play an important but uncertain role in the global carbon (C) cycle. Current modeling efforts to quantify terrestrial SOC dynamics in response to global environmental changes do not accurately represent the size, spatial distribution and flux of C from the soil. Here, we modified the DAYCENT (daily version of the Century) biogeochemical model to improve our understanding of SOC dynamics in response to land use change under a changing climate in the US Great Plains ecoregion by utilizing C fraction data to a depth of 20cm. We parameterized and calibrated the conceptual pools of the DAYCENT model against C fraction data to quantify the SOC dynamics during the historical (1895-2005) period, and project its response to Intergovernmental Panel on Climate Change (IPCC) 5th Assessment Report (AR5) RCP8.5 (business as usual with high GHG emission), and RCP4.5 (stabilization of radiative forcing by 2100) climate scenarios and IPCC 4th Assessment Report (AR4) A2 land cover scenario through 2100. Results showed that matching the conceptual pools against C fraction data using the modified DAYCENT model (DCmod), led to better initialization of equilibrium model pool stocks and distribution at long-term research sites compared to the default DAYCENT model (DCdef). Regional simulation using the DCmod demonstrated higher SOC stocks for both croplands (34.86 vs 26.17 Mg C ha-1) and grasslands (54.05 vs 40.82 Mg C ha-1), compared to the default model for the contemporary period (2001-2005 average) which better matched observationally constrained data-driven maps of current SOC stock distributions. When these SOC stocks were compared to the simulated baseline SOC (1895-1899 average), the DCmod showed large absolute losses for both croplands (17.62 vs 10.60 Mg C ha-1) and grasslands (2.51 vs 1.06 Mg C ha-1) during 1895-2005. The larger absolute losses also translated into larger relative losses with croplands and grasslands losing 33% and 4% of the baseline SOC, respectively. Projection of the SOC dynamics to land cover change (IPCC AR4 A2 scenario) under the IPCC AR5 RCP8.5 climate scenario showed absolute SOC loss of 8.44 and 10.30 Mg C ha-1 for grasslands and croplands, respectively using the DCmod compared to the DCdef (6.55 and 7.85 Mg C ha-1 for grasslands and croplands). Future absolute SOC loss also translated into large relative losses with DCmod losing SOC stocks by 29% and 16% of contemporary SOC for croplands and grasslands, respectively. The projected SOC loss using the DCmod was 31% and 29% higher for croplands and grasslands compared to DCdef model. Overall, our modeling study demonstrates that initializing SOC pools against C fraction data result in more accurate representation of pool sizes with larger absolute and relative SOC losses to agricultural intensification in the warming climate.