A flexible method to determine the sensitive physical parameter combination for soil carbon under five plant types. Issue 8 (18th August 2017)

• Record Type: Journal Article
• Authors: Sun, Guodong; Mu, Mu
• Is Part Of: Ecosphere. Volume 8:Issue 8(2017)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2017-08-18
• Subjects: conditional nonlinear optimal perturbation related to parameter (CNOP‐P) -- model parameter errors -- sensitivity of parameter combination -- soil carbon -- uncertainty analysis; Ecology -- Periodicals; Ecology; Periodicals; 577.05
• DOI: 10.1002/ecs2.1920
• Languages: English
• ISSNs: 2150-8925

Abstract

Abstract: Model parameter errors are a critical source of uncertainty in simulations of soil carbon. It is important to determine which parameter errors should be reduced to improve the simulation ability of soil carbon. It is essential to determine the most sensitive parameter combination. In this study, a new flexible approach is employed to identify the most sensitive parameter combination based on the conditional nonlinear optimal perturbation related to parameter (CNOP‐P) using the Lund‐Potsdam‐Jena model. The CNOP‐P is a type of parameter error that triggers the upper bound of uncertainty in a numerical simulation and prediction, for which a nonlinear effect can be reported. To compare the sensitivity of parameter combinations, the one‐at‐a‐time (OAT) approach is also applied to determine the sensitivity of each parameter. Using the CNOP‐P approach, parameter errors can really cause large simulation errors for soil carbon throughout an entire study region (2674.69 g C/m 2 ). Results show the most sensitive parameter combinations of soil carbon are different for five plant functional types (PFTs). For all cases, f air (fraction of decomposed litter emitted as CO2 to the atmosphere) and f inter (fraction of soil‐bound decomposed litter entering intermediate soil carbon pool) comprise the most sensitive parameter combination for boreal needle‐leaved evergreen trees, boreal needle‐leaved summergreen trees, temperate broadleaved summergreen trees, temperate broadleaved evergreen trees, and C3 perennial grass. For C3 perennial grass under semi‐arid conditions, the hydrology process is also critical in the uncertainty of simulated soil carbon. The most sensitive parameter combination using the new flexible approach is different to that of the top rank of sensitivity of each parameter when employing the CNOP‐P and OAT methods, which implies that a nonlinear effect of the parameter combination is crucial for determining the sensitive parameter combination. The sensitivity of the parameter combination is dependent on the PFTs and climate condition. Among factors causing the uncertainty of simulated soil carbon, the fast‐decomposing soil carbon pool plays the main contribution. It is considered that the physical process involved in the fast‐decomposing soil carbon pool needs to be improved as a priority compared to other physical processes to enhance the simulation ability of soil carbon.