Process-based model estimation of soil carbon stocks and sensitivities at high resolutions
Abstract
Efforts to protect and manage soil carbon (C) are heavily reliant upon methods to estimate soil C stocks and sensitivity to environmental change. Here we present a new programmatic routine for producing high-resolution estimates of soil organic carbon (SOC) stocks using the MIcrobial-MIneral Carbon Stabilization (MIMICS) model. The central process involves harnessing common spatial data and machine-learning to provide a widely applicable method for generating detailed estimates of SOC stocks and sensitivities across complex landscapes. As a proof of concept, we used the routine to map 0-30 cm SOC stocks across the entirety of the Reynolds Creek Experimental Watershed (240 km^2) at a grid scale of 10 m^2 (validation r > 0.81; calibration r > 0.74). The explicit representations of soil C processing and accumulation in MIMICS also provide the routine with the capability to produce high-resolution spatial estimates of soil microbial C, particulate C, protected C and litter C stocks. By reducing the modelling scale, direct connections can now be improved between more nuanced field data (e.g. microbial and SOC fraction data) and the corresponding mathematical representations found in the process-based model. Further, in contrast to many digital SOC mapping methods, which often rely on direct statistical correlations, the process representations in MIMICS provide a platform to project SOC stock sensitivities to changing conditions. Together with the mapping capabilities in the routine, this allows for the production of high resolution maps of the relative sensitivity of SOC stocks, across complex landscape areas, to expected environmental changes, such as increases in soil temperature and gross primary productivity. With further development and cross-site validation, we foresee great potential for this programmatic routine to efficiently provide the detailed soil C information needed to support proactive land management and policy decisions.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2021
- Bibcode:
- 2021AGUFMEP43B..01P