基于EPIC模型的区域水稻作物参数敏感性分析

发布时间：2018-03-23 13:49

本文选题：EPIC　切入点：水稻　出处：《浙江师范大学》2015年硕士论文

【摘要】：全球环境变化逐渐成为人们关注的热点和焦点问题,其中最为突出的是全球变暖趋势日益加剧。随着全球变暖问题的凸显,全球旱灾的发生频率和影响范围也在不断扩大。旱灾是世界上影响范围最广,造成农业损失最大的自然灾害之一,它严重影响着全球粮食生产和粮食安全。水稻作为一种易受干旱影响的作物,其生产面临巨大挑战。在水稻生产研究中,作物模型是一种常用的手段。但几乎所有的模型都很难具有普适性,在不同区域应用时都需要对模型的参数进行校准。对于参数繁杂的模型而言,对其中所有参数进行调整工作量巨大,难以实现。参数敏感性分析即是从众多参数中识别和选择关键的控制参数,筛选出引起模型结果不确定性的主要因素的一种有效手段。本文以EPIC作物模型为基础,对一定区域内水稻作物参数进行敏感性分析。并以我国南方水稻主产区为例对敏感性分析结果的空间差异性进行分析。具体步骤包括,从EPIC模型众多水稻作物参数中筛选出对模拟产量可能有一定影响的参数,并进一步确定这些参数的取值范围和分布形式,进而进行敏感性分析。本文选取了8个代表性地点进行敏感性分析,分析结果显示受不同地区环境影响作物参数的敏感性差异较大。为探究敏感性分析结果的空间差异性,本文以我国南方水稻主产区为例,对研究区内水稻种植点的作物参数进行敏感性分析,同时利用相关性分析进一步探究环境要素与敏感参数之间的关系。通过研究得到以下主要结论：(1)敏感性分析可以有效地筛选出对模型输出结果影响较大的参数,用户通过确定和调整这些参数完成模型校准。因此,敏感性分析在模型校准、模型简化等方面有较大的应用潜力。同时,EFAST作为一种简单、高效、准确的全局敏感性分析方法,可以在较少的样本量下完成全局敏感性计算。相较于一阶敏感性分析而言,全局敏感性全面考虑参数对模型结果的影响,以及参数之间相互作用对模型结果的影响。(2)本文通过对比8个不同地点的敏感性分析,结果发现其具有明显的空间差异性。以我国南方水稻区为例,总体敏感性较高的参数为WA、HI、TBS和TOP等4个参数。同时不同地点的最敏感参数不同,且各参数的敏感指数有区域性差异。(3)通过计算各参数的敏感指数与环境要素的相关性,结果表明TBS与施肥量相关性最高,WA与播种日期相关性最高,HI与日最高气温相关性最高,TOP与日最低温度相关性最高。从各环境要素来看,太阳辐射、日最高气温、日最低气温与TBS、WA、HI、TOP相关性较好,其中TBS与其成负相关,WA、 HI、TOP与其成正相关。
[Abstract]:Global environmental change has gradually become the focus of attention, among which the most prominent is the increasing trend of global warming. The frequency and scope of global drought is also expanding. Drought is one of the most widespread natural disasters in the world, causing the greatest agricultural losses, It has a serious impact on global food production and food security. Rice, as a drought-prone crop, faces enormous challenges in its production. Crop models are a common method, but almost all models are difficult to be universally applicable, and the parameters of the models need to be calibrated when they are applied in different regions. It is difficult to realize the adjustment of all the parameters. The parameter sensitivity analysis is to identify and select the key control parameters from many parameters. An effective method for screening out the main factors that lead to uncertainty of the model results. This paper is based on the EPIC crop model. The sensitivity analysis of rice crop parameters in a certain region is carried out. The spatial differences of the results of sensitivity analysis are analyzed by taking the main rice producing areas of southern China as an example. The concrete steps are as follows:. The parameters which may affect the simulated yield were screened from many rice crop parameters in the EPIC model, and the range and distribution of these parameters were further determined. Then sensitivity analysis was carried out. In this paper, 8 representative sites were selected for sensitivity analysis. The results showed that the sensitivity of crop parameters affected by environment in different regions was different. In order to explore the spatial difference of sensitivity analysis results, In this paper, the sensitivity analysis of crop parameters of rice planting sites in southern China was carried out by taking the main rice producing areas in the south of China as an example. At the same time, the correlation analysis is used to further explore the relationship between the environmental factors and the sensitive parameters. The following main conclusions can be drawn from the study: sensitivity analysis can effectively screen out the parameters that have great influence on the output results of the model. The user completes the model calibration by determining and adjusting these parameters. Therefore, sensitivity analysis has great application potential in model calibration and model simplification. At the same time, EFAST is a simple, efficient and accurate global sensitivity analysis method. The global sensitivity can be calculated with less sample size. Compared with the first-order sensitivity analysis, the global sensitivity takes into account the influence of the parameters on the model results. By comparing the sensitivity analysis of 8 different sites, it is found that there are obvious spatial differences. Take the rice region of southern China as an example, The most sensitive parameters are different in different locations, and the sensitivity index of each parameter has regional difference. The correlation between the sensitivity index of each parameter and the environmental factor is calculated by calculating the correlation between the sensitivity index of each parameter and the environmental factor. The results showed that the correlation between TBS and fertilization amount was the highest and the correlation between TBS and sowing date was the highest. The correlation between top and daily maximum temperature was the highest, and the correlation between top and daily minimum temperature was the highest. From the point of view of various environmental factors, solar radiation and daily maximum temperature were the highest. There is a good correlation between daily minimum temperature and TBS-WAHI-TOP, in which TBS is negatively correlated with TBSWA-TOP, and HITO-TOP is positively correlated with TBSWA-TOP.
【学位授予单位】：浙江师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：S511;S423

【参考文献】