| Global environmental change has gradually become the hotspot and focus of attention, one of the most prominent is the rising of the global warming trend. As the problem of global warming highlighted, the frequency and scope of drought are also constantly expanding. The drought is the world’s most widely influence, one of natural disaster which cause the biggest loss of agriculture, and it seriously affects the global food production and security. As a vulnerable crop, the production of the rice faces enormous challenges.In the study of rice production, crop model is a commonly used method. But almost all of the models are difficult to find the generality, all need parameters calibration when it is applied in different areas. For large models, to adjust all parameters is a big workload, and it is difficult to achieve. Parameter sensitivity analysis is to identify and choose the control parameters from many parameters, and it is an effective means to screen out the main factors of uncertainty of model results.In this paper, sensitivity analysis was carried out on the rice crop parameters within a certain area based on the EPIC crop model. And it takes the spatial heterogeneity of rice sensitivity analysis in the south of China as an example. Specific steps are as follows:selecting the parameters of EPIC model that may have effect on the analog output from many rice crop parameters. And further determining the scope and distribution form of these parameters, and analyzing the sensitivity. This article selected eight representative site for sensitivity analysis, the analysis results show that with the effect of different region environment, crop sensitivity had a big difference. To explore the spatial difference of sensitivity analysis results, based on rice in south China as an example, the article studies on sensitivity analysis on the crop parameters in the study area of rice planting point, at the same time, using the correlation analysis to further explore the relationship between the environmental elements and the sensitive parameters.This paper obtained the following main conclusions through research:(1) Parameters effected on the output to a great extent can be effectively filtered out using the model sensitivity analysis, users achieve the model calibration through the adjustment and determination of these parameters. Therefore, sensitivity analysis has great potential in application in the aspect of model calibration and model reduction. At the same time, EFAST as a simple, efficient, accurate global sensitivity analysis method, can complete the global sensitivity calculation use a few samples. Compared to the first-order sensitivity analysis, the global sensitivity analyze the parameter influencing on model results on a deeper level and the interaction between parameters impacting on model results.(2) It showed obvious spatial differences through contrasting sensitivity analysis results of 8 different locations. Taking the south rice production area in China as an example, the higher four overall sensitivity parameters were WA, HI, TBS and TOP. And the most sensitive parameters of different place were distinct, the sensitive parameter of each index has regional differences.(3) The paper calculates the correlation between the sensitive parameter of each index and environmental factors. The results show the strongest correlation between TBS and fertilization, WA and the sowing, HI and daily maximum temperature, TOP and daily minimum temperature. From the point of the environmental factors,the best correlations with the TBS, WA, HI, TOP are solar radiation, daily maximum temperature, and the daily minimum temperature. There’s a positive correlation between TBS and these factors. WA, HI and TOP negatively correlates with these factors. |
PDF Full Text Request