Optimizing process-based models to predict current and future soil organic carbon stocks at high-resolution

Authors: PIERSON, DEREK - Idaho State University; LOHSE, KATHLEEN - Idaho State University; WIEDER, WILLIAM - National Center For Atmospheric Research (NCAR); PATTON, NICHOLAS - University Of Canterbury; FACER, JEREMY - Idaho State University; DE GRAAFF, MARIE-ANNE - Boise State University; GEORGIOU, KATERINA - Lawrence Livermore National Laboratory; SEYFRIED, MARK - Retired ARS Employee; Flerchinger, Gerald; WILL, RYAN - Boise State University

Submitted to: Scientific Reports
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 6/2/2022
Publication Date: 6/25/2022
Citation: Pierson, D., Lohse, K.A., Wieder, W.R., Patton, N.R., Facer, J., de Graaff, M., Georgiou, K., Seyfried, M., Flerchinger, G.N., Will, R. 2022. Optimizing process-based models to predict current and future soil organic carbon stocks at high-resolution. Scientific Reports. 12. Article 10824. https://doi.org/10.1038/s41598-022-14224-8.
DOI: https://doi.org/10.1038/s41598-022-14224-8

Interpretive Summary: Carbon stored in the soil helps mitigate greenhouse gases and improve soil health and productivity. There is a growing need to project soil carbon storage to address accelerating changes in land use and global climate. Scientists working with data from the Reynolds Creek Critical Zone Observatory and Experimental Watershed developed methods for mapping the amount of soil carbon stored across complex landscapes and produced a map of soil carbon storage across the Reynolds Creek watershed. Such a map for this and other landscapes is a critical step toward understanding carbon storage and its vulnerability to climate and land use changes, thereby informing land management decisions under a changing climate.

Technical Abstract: Soil carbon (C) management and mitigation policies are reliant on estimates of soil C stocks, especially at fine spatial scales. However, given enduring data limitations, statistical models used for such estimates are limited in their ability to predict the underlying composition and vulnerability of soil C to global change. Here we show that an optimized, process-based model is uniquely suited to fill this gap. We parameterize the MIcrobial-MIneral Carbon Stabilization (MIMICS) model with spatially-explicit data across the Reynolds Creek Experimental Watershed in SW Idaho, USA, and illustrate that data-constrained model parameterization can reduce uncertainty in total soil C stocks (r=0.82 for an independent dataset). We produce the first high-resolution (10 m^2) estimates of soil C stocks (including litter, microbial, particulate, and protected C pools) to a depth of 30 cm across the entire watershed (239 km^2), and predict their respective vulnerabilities to a suite of environmental disturbances. Generating high-resolution estimates of soil C stocks and measurable underlying pools, are a critical step towards understanding soil C storage and vulnerability, and informing land management under a changing climate.