基于EFAST方法的AquaCrop作物模型参数全局敏感性分析

【目的】敏感性分析是作物模型本地化过程中的重要环节,对作物模型的校正与应用有重要的意义。【方法】本研究以国家精准农业示范研究基地2012—2013、2013—2014和2014—2015年冬小麦试验为研究对象,采用全局敏感性分析方法扩展傅里叶幅度检验法(Extended Fourier Amplitude Sensitivity Test,EFAST)对AquaCrop模型42个作物参数进行敏感性分析,以评估模型在北京地区的敏感参数。【结果】(1)对干生物量敏感作物参数是:水分和温度胁迫参数(生物量生产的最小生长度(stbio),引起冠层早衰的土壤水分消耗上限(psen))、生物量和产量参数(归一化水分生产力(wp))、蒸散参数(作物冠层形成后到衰老之前的作物系数(kcb))、作物冠层和物候发展参数(冠层生长系数(cgc),从播种到出苗时长(eme),最大冠层覆盖度(mcc),冠层衰老系数(cdc),从播种到成熟的时长(mat),产量形成过程中收获指数的建立长度(hilen))。其中stbio,kcb,wp和cgc 4个作物参数敏感性指数最大;(2)对冠层覆盖度最敏感的参数是:作物冠层和物候发展参数(cgc,mcc,每公顷株数(den),出苗率达到90%时的土壤覆盖度(ccs),mat和cdc)、根区发展参数(最大有效根深(rtx))、水分和温度胁迫参数(psen)、蒸散参数(kcb);(3)对产量最敏感的参数是作物冠层和物候发展参数(从播种到开花时长(flo),mat,cdc,hilen和从播种到开始衰老时长(sen))、水分和温度胁迫参数(psen)、生物量和产量参数(参考收获指数(hi)和wp)、蒸散参数(kcb)。【结论】利用EFAST方法对AquaCrop模型中的作物参数进行一阶和全局敏感分析,最大干物量的敏感性分析结果以及干生物量随时间变化的敏感性分析结果显示,敏感性参数的选择上差异不大,但排序上存在较大的差异,最大干生物量的敏感性分析不能分析作物参数对干生物量在整个生育期的影响,结果不全面;冠层覆盖度随时间变化的一阶和全局敏感性分析结果显示,在敏感参数的选择和排序上均有较好的一致性,全局敏感性分析中作物参数的敏感性指数更高,对冠层覆盖度的影响表现得更明显。本研究结果用于AquaCrop模型本地化,可提高该模型在北京地区的模拟效率和模拟精度。

Global Sensitivity Analysis of AquaCrop Crop Model Parameters Based on EFAST Method

【Objective】Sensitivity analysis is an important link in crop model localization, and it plays an important role inAquaCrop model calibration and application.【Method】In this study, in order to identify the sensitivity parameters, the 2012-2013, 2013-2014 and 2014-2015 winter wheat experiments were conducted in National Precision Agriculture Demonstration Research Base in Beijing, China, the Extended Fourier Amplitude Sensitivity Test (EFAST) method was used to carry out sensitivity analysis of 42 crop parameters of AquaCrop model.【Result】The sensitivity parameters were: (1) For dry biomass: water and temperature stress (minimum growing degrees required for full biomass production (stbio), upper threshold of soil water depletion factor for canopy senescence (psen)), biomass and yield production (water productivity normalized (wp)), transpiration (crop coefficient when canopy is complete but prior to senescence (kcb)), canopy and phaenological development (GGD-increase in canopy cover (cgc), GDD-from sowing to emergence (eme), maximum canopy cover in fraction soil cover (mcc), GGD-decrease in canopy cover (cdc), total length of crop cycle in growing degree-days (mat), building-up of harvest index during yield formation (hilen)). stbio, kcb, wp and cgc were the four most sensitive parameters; (2) For canopy cover: canopy and phaenological development (cgc, mcc, number of plants per hectare (den), soil surface covered by an individual seedling at 90% emergence (ccs), mat and cdc), root development (maximum effective rooting depth (rtx)), water and temperature stress (psen), transpiration (kcb); (3) For yield: canopy and phaenological development (GDD-from sowing to flowering (flo), mat, cdc, hilen and GDD-from sowing to start senescence (sen)), water and temperature stress (psen), biomass and yield production (reference harvest index (hi) and wp), transpiration (kcb).【Conclusion】The results of first order and total order sensitivity analysis for AquaCrop model of winter wheat maximum dry biomass and dry biomass time-varying showed that there was a little difference in the choice of sensitivity parameters, but many differences in the ranking. The sensitivity analysis of maximum dry biomass was not comprehensive, which could not analyze the effect of crop parameters on dry biomass during the whole growth period. The results of the first order and total order sensitivity analysis for AquaCrop model of winter wheat canopy cover time-varying showed that there was a good consistency in the selection and ranking of sensitive parameters. The values of total order sensitivity indices of crop parameters were higher than first order, and the influences on canopy cover were more obvious. This study provides guidelines for AquaCrop model calibration and application in Beijing, China, as well providing guidance to simplify the AquaCrop model and improveits precision, especially when many parameters are used.