Time series global sensitivity analysis of genetic parameters of CERES-maize model under water stresses at different growth stages

Abstract

Sensitivity analysis (SA) is used to identify the effects of crop model input parameters on model results. Previous studies have indicated that the CERES-Maize model was difficult to calibrate under the different water stress conditions. A genetic parameters time-series sensitivity analysis is needed to guide parameter optimization. The objectives of this study were to: (i) comprehensively quantify the genetic parameters in CERES-Maize based on the extended Fourier amplitude sensitivity test. The sensitivity of CERES-Maize output variables was analyzed under different water stress conditions; (ii) determine the sensitivities of output variables to genetic parameters during the growth period. The results demonstrated that output variable sensitivity varied in response to different water stress conditions. The total sensitivity index (TSI) and time-dependent TSI of crop parameters were more sensitive than the first sensitivity index (FOSI) and time-dependent FOSI of crop genetic parameters. The sensitivities of two years (2013 and 2014) based on FOSI and time-series FOSI were consistent; some differences existed between simulations based on TSI and time-series TSI. Under different water management conditions, the sensitivity of phyllochron interval (PHINT) to biomass decreased earlier and faster when drought occurred in the early growth period (D1 and D2). The time series of sensitivity index was consistent with the CK treatment when drought happened in the later growth periods (D3 and D2). The two parameters of PHINT and thermal time from emergence to end of juvenile (P1) were most sensitive to leaf area index (LAI) when drought occurred in the early growth periods (D1 and D2). In addition to PHINT and P1, other parameters also had sensitivities for LAI when drought occurred in later growth periods (D3 and D4). Future studies should focus on the response of dynamic output variable to soil parameters and weather conditions over the growing season in order to calibrate and apply the CERES-Maize model.