基于EFAST的CROPGRO-Tomato模型参数全局敏感性分析

为了定量讨论番茄生长模型(DSSAT-CROPGRO-Tomato)中各参数对模拟输出结果的影响,运用扩展傅里叶幅度检验(EFAST)法,对影响番茄物候期、生长及生产等3类模型输出的各参数敏感性进行了研究,重点探讨了模型输入中品种参数、气象及土壤参数变化对模型输出结果的影响。结果表明:影响番茄开花期、坐果期和成熟期等物候期最敏感的品种参数和气象参数分别为初花期积温、日最低温度,而土壤参数对番茄物候期的影响可以忽略。影响番茄叶面积指数和冠层高度最敏感的品种参数为初花期积温,而影响番茄根、茎生长最敏感的品种参数为干物质分配比例,日最高温度和土壤田间持水率分别是影响番茄根、茎、叶生长最敏感的气象和土壤参数。影响番茄干物质量和总产量最敏感的品种参数为初花期积温,而影响番茄收获数量和单果质量的品种参数为最大单籽粒质量,日最高温度和土壤田间持水率也分别是影响番茄生产最敏感的气象和土壤参数。

Global Sensitivity Analysis of CROPGRO-Tomato Model Parameters Based on EFAST Method

Sensitivity analysis plays an essential role in the processes of crop model localization. It is important to quantify the sensitivity of input parameters of model for crop model calibration and application. In order to quantify the effects of input parameters variation on the sensitivity of CROPGRO-Tomato output variables, the extended Fourier amplitude sensitivity test method was applied to perform global sensitivity analysis. The most sensitive parameters for anthesis, fruit set and maturity date were the photothermal days in the early flowering period (EMFL) among the cultivar parameters, and the average minimum air temperature among climate parameters, while the phenology of tomato was not affected by soil parameters in the model. Among the cultivar parameters, the most sensitive parameters for leaf area index and canopy height was EMFL, while the most sensitive parameters for growth of root and stem was the maximum fraction of daily growth. The growth of root, stem and leaf was affected sensitively by the average maximum air temperature (TMXD) and the soil field capacity (SDUL) among climate and soil parameters, respectively. The most sensitivity parameters for the tops weight at maturity and fresh fruit weight was EMFL, while the most sensitivity parameter for number of fruits at maturity and single fruit weight was maximum weight per seed among cultivar parameters. TMXD and SDUL were also the most sensitivity parameters for tomato production among climate and soil parameters. The research result indicated that EFAST method was a feasible way for CROPGRO-Tomato model calibration and application.