S-Space College of Agriculture and Life Sciences (농업생명과학대학) Dept. of Plant Science (식물생산과학부) Theses (Master's Degree_식물생산과학부)
The sensitivity analysis of cultivar parameters for CERES-Maize model using long term weather data
- 농업생명과학대학 식물생산과학부
- Issue Date
- 서울대학교 대학원
- sensitivity analysis; cultivar parameters; maize; ANOVA; DSSAT; CERES-Maize; long term weather data
- 학위논문 (석사)-- 서울대학교 대학원 : 식물생산과학부, 2016. 8. 김광수.
- The uncertainty of model arises from input data, equation, and parameters. The sensitivity analysis would be used to assess the degree of uncertainty, which would facilitate improvement of a crop model. Little efforts have been made to assess impact of cultivar parameters under a long term climate condition, which would be useful for examining reliability of climate change impact assessment studies using crop models. In this study, we focused on sensitivity analysis of cultivar parameters using 30 years of weather data.
Six cultivar parameters in CERES-maize model were subjected to sensitivity analysis. Weather data observed for 30 years and single soil parameter set at a site in Suwon area used in sensitivity analysis. The total number of simulation was 3.9 x 108 to accommodate the range of parameters and simulation conditions based on a complete factorial design. Sensitivity of parameters were analyzed in terms of ear weight, biomass, and yield.
It was found that P5, P1, and G3 were influential parameters in ear weight, biomass, and yield, respectively. The error term were mainly derived from climate and unknown error. Particularly, the interaction resulted from climate and planting date made a variation of sensitivity index to influential parameters (P1, P5, and G3). Climate change condition made a sensitivity index of influential parameters to be changed in decadal analysis. For example, in ear weight, the rank of impact of parameters in 1980s was P1 > P5 > G3 but changed in 2000s was P5 > P1 > G3. The result of sensitivity analysis using OAT was similar with that of ANOVA. For example, the rank of OAT was P5 > P1 > G3 > G2 > P2 > PHINT in ear weight.
Our results indicated that cultivar parameters associated with thermal time had greater sensitivity than other parameters in simulation of maize growth and yield. To reduce the uncertainty derived by temperature, it is recommended that the experiment should be performed under ambient temperature conditions to make sure accurate cultivar parameter values in estimation process. The minimum temperature and daily temperature difference would cause change of sensitivity index in decadal analysis on sensitivity index. It would be better using long term weather data to figure out the response of parameters under climate change condition. The result of sensitivity analysis could be differed by experimental conditions
the version of DSSAT, sensitivity analysis method, parameter range, climate data, crop management data, sample size, and soil conditions. Quantifying the impact of simulation conditions for conducting sensitivity analysis would help understand accurate response of parameters.