Premium
Modelling intercrop management impact on runoff and erosion in a continuous maize cropping system: Part I. Model description, global sensitivity analysis and Bayesian estimation of parameter identifiability
Author(s) -
LALOY E.,
BIELDERS C. L.
Publication year - 2009
Publication title -
european journal of soil science
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.244
H-Index - 111
eISSN - 1365-2389
pISSN - 1351-0754
DOI - 10.1111/j.1365-2389.2009.01187.x
Subject(s) - identifiability , surface runoff , soil science , environmental science , erosion , hydraulic conductivity , mathematics , sensitivity (control systems) , hydrology (agriculture) , statistics , soil water , geotechnical engineering , geology , ecology , engineering , paleontology , electronic engineering , biology
Summary In order to evaluate the long‐term impact of intercrop management on runoff and erosion in a continuous maize cropping system a new model was developed, based on the continuous plot‐scale runoff model of Laloy & Bielders (2008). This model is called the Continuous Runoff and Erosion Hillslope model with DYnamic soil Surface properties (CREHDYS). This paper details the erosion and crop growth and decay model components and presents a thorough global variance‐based sensitivity analysis (GSA) with regard to the erosion prediction, followed by a Bayesian parameter identifiability assessment. As compared with the classical local univariate sensitivity analysis, a GSA is able to deal with the typical non‐linearity of process‐based hydrological and erosion models. The most influential parameters were the Manning's roughness coefficient followed by the saturated hydraulic conductivity of wheel track cells and the median particle size of the material. The soil aggregate stability and soil‐cohesion parameters were found to be almost non‐influential. Most of the results of the parameter identifiability procedure were in close agreement with the GSA. Indeed, the parameter uncertainty seemed to be proportional to the degree of influence, with Manning's coefficient being the most precisely identified, whereas soil and wheel track cohesion parameters showed the largest uncertainty. Exceptions were the soil aggregate stability and the Green‐Ampt soil matric potential. The uncertainty associated with the former was surprisingly low given its low level of influence whereas the uncertainty associated with the latter is partially explained by its negative correlation with the soil saturated hydraulic conductivities of overland flow and wheel track cells.