农作物单产模型研究
文献类型：学位论文
| 作者 | 徐新刚 |
| 学位类别 | 博士 |
| 答辩日期 | 2007 |
| 授予单位 | 中国科学院水利部成都山地灾害与环境研究所 |
| 授予地点 | 成都 |
| 导师 | 周万村 ; 吴炳方 |
| 关键词 | 作物单产模型 农业气象模型指标集 组合模型 权重非负 遥感光合模型 产量空间分布 |
| 学位专业 | 自然地理学 |
| 中文摘要 | 农作物单产模型是农作物估产技术体系的主要组成部分。当前用于农作物估产的作物单产模型有许多，但在应用时多以单个模型进行作物单产的估算，很少考虑组合多个单产模型进行单产的综合估算，所用的模型也多为统计经验模型，一般只得到作物单产的数值信息，难以获取作物单产的空间分布信息。论文在“中国农情遥感速报系统”单产估算技术研究基础上，结合当前作物单产估算模型中存在的一些问题，研究分析了常规统计经验单产模型的组合估算问题，并借此讨论了相关模型的区域适宜性；同时也探讨了如何利用高低分辨率遥感数据各自优势构建机理型遥感光合模型提取作物产量空间分布信息的技术方法。首先，论文在农作物单产区划全国范围尺度上，建立农业气象模型指标集，在此基础上，以农业气象模型、遥感指数模型和气象遥感模型等统计经验模型为单项模型，研究构建组合模型的问题，并根据组合模型的权重分配情况对三者进行区域适宜性评价分析；然后在区域尺度上，以海河流域北京地区的主要粮食作物单产为研究对象，探讨运用机理性遥感光合模型，综合利用高低分辨率遥感数据各自空间和时间分辨率互补优势来提取作物单产空间分布信息的技术方法。论文的主要研究内容与成果如下：（1）全国单产区划尺度上的农业气象模型指标集的建立。在全国单产区划尺度上，以各区划主要粮食作物单产估算为研究对象，采用主成分分析法建立了农业气象模型的作物单产估算指标集，该指标集明确了对不同单产区划中的相应主要粮食作物进行单产估算时应该选择哪些物候期，用什么气象因子来建立农业气象模型的应用问题。通过主成分分析提取的农业气象模型的指标集表明：在影响作物产量的气象因素中，第一主导因素为温度，主要是平均气温和最高气温，其中均温对作物产量影响具有普遍性意义；第二主导因素则表现为水分，主要为降水和相对湿度。在影响作物产量的主导物候期因子中，影响作物产量的关键物候多表现为作物由营养生长开始到作物生殖生长阶段的物候期。（2）在全国单产区划基础上，提出以多个单产模型构建组合模型的方法。以“误差平方和最小，权重非负，且权重和为1”为目标约束函数，采用迭代算法计算组合模型权重，构建了以农业气象模型、遥感指数模型和气象遥感模型为单项模型的各种二元和三元组合估算模型。分析结果表明：组合模型可充分利用各单项单产估算模型提供的有用信息，得到的SSE（误差平方和）明显小于单个模型的SSE，估算相关系数R2也明显提高，表明组合模型的估算精度高于单个单产估算模型的精度。（3）通过分析组合模型的权重分配情况，在全国单产区划基础上建立起农业气象模型、遥感指数模型和气象遥感模型三个统计单产模型的区域适宜性体系。分析结果表明：气象遥感模型对各单产区划主要粮食作物的单产估算具有最普遍的区域适宜性，较之相应区域的农业气象模型和遥感指数模型，气象遥感模型估算具有更高的精度。（4）构建机理性遥感光合模型估算作物单产。论文在Monith干物质量生产效率模型 基础上，引入光能利用率模型和两个地面能量平衡模型SEBS(Surface Energy Balance System)与SEBAL(Surface Energy Balance Algorithm for Land)构建具有一定机理性的作物产量估算模型。结果表明：遥感光合模型不仅可获取作物产量数据信息，而且还能得到作物单产的空间分布信息图；通过该模型对海河流域北京地区的2006年主要粮食作物进行单产估算，结果表明估算与实测相关性R2达到0.771，表明本文提出的遥感光合模型具有一定的可比精度。（5）在运用遥感光合模型提取作物产量空间分布信息时，论文设计了一种利用TM和MODIS遥感影像各自空间和时间分辨率优势进行时间外推计算生长季作物干物质量，进而估算作物产量的方法，为今后综合利用高低分辨率遥感影像提取作物产量提供一定的方法参考。 |
| 英文摘要 | The estimation of crop yield was one main component of the crop production estimation. Nowadays, there were all kinds of yield estimating models, but scarcely took the combined models into account, most of models were statistically experiential models, by which only crop yield value could usually be obtained, and seldom got the spatial distribution information of crop yield.Considering the above problems, based on crop yield estimating technologies of China Crop Watch System with Remote Sensing, the dissertation had explored the combined model to estimate crop yield, discussed the zonal suitability of the models which were used to construct the combined models, and made the research how to use remote sensing data coupling high spatial resolution with high temporal resolution to establish the light utility model with remote sensing in order to obtain the spatial distribution information of crop yield. On one hand, the dissertation built the index collection for the agrometeorological model at Chinese crop zone scale, and then constructed the combined models by integrateing agrometeorological model, spectra-VI model with agromet-spectra model, furthermore, analyzed the weights of the combined models in order to assess the zonal suitability of the three models for different crop yield estimation in each zone. On the other hand, at region scale, the dissertation explored the method of applying the light utility model with remote sensing to get the spatial distribution of crop yield by combining spatial resolution advantage of TM with temporal resolution advantage of MODIS.The main content and conclusions were as following:（1）Index collection of agrometeorological model was build for each crop yield zone. The dissertation adopted PCA(principal component analysis) to analyze the dominantly phonological and meteorological factors effecting crop yield for different crops in each crop yield zone, and set up the index collection of agrometeorological model, which could resolve the problem of which phonological and meteorological factors should be efficiently selected to construct agrometeorological models. Moreover, the index collection indicated that, among the agrometeorological factors effecting crop yield, temperature, mainly average and maximum temperature, were the first dominant factors, specially, average temperature had universally effect on crop yield.; the second dominant was water, usually rainfall and relative air humidity. Among the phonological factors, the dominant factors effecting crop yield were mostly the ones during the phase from the start of crop nutritional growth to the end of crop procreative growth.（2）Studied the combined models at Chinese crop zone scale. Having on both minimum of SSE(Sum of Squared Error) and positive weight as the constraint function, and calculating the weight by certain iterative operation, the dissertation built all kinds of combined models with agrometeorological model, spectra-VI model and agromet-spectra model, the result showed that the combined models could make the best of the useful information of single models, and had more higher accuracy than single models, with decreasing SSE and improving R2.（3）Based on analyzing the weight of combined models, the framework of the zonal suitability for the three models, agrometeorological model, spectra-VI model and agromet-spectra model had been established. The result showed that the agromet-spectra model had the largest suitability, and was suitable to estimate crop yield in each zone. In addition, compared with the agrometeorological model and spectra-VI model, the agromet-spectra model had more high accuracy.（4）Explored the method of establishing the light utility model based on remote sensing to obtain the spatial distribution of crop yield. Integrating the light utility efficiency model, energy balance model with Monith’s dry matter production model, the dissertation constructed the light utility model with remote sensing, which could not only get crop yield value, but also obtain the spatial distribution of crop yield. Aapplying the model to estimate crop yield on Beijing cropland in 2006 showed that the correlation coefficiency between estimating yield and measured yield was 0.771, which indicated that the light utility model based on remote sensing had higher accuracy when adopted to estimate crop yield.（5）The research in the dissertation designed a method of making use of both spatial resolution advantage of TM and temporal resolution advantage of MODIS to compute crop yield, which could provide certain use for reference how to efficiently integrate remote sensing data with different resolutions to obtain crop yield information. |
| 学科主题 | 农业区划 |
| 语种 | 中文 |
| 公开日期 | 2010-10-21 |
| 分类号 | TP3;TP7 |
| 源URL | [http://ir.imde.ac.cn/handle/131551/2264]  |
| 专题 | 成都山地灾害与环境研究所_成都山地所知识仓储（2009年以前） |
| 推荐引用方式 GB/T 7714 | 徐新刚. 农作物单产模型研究[D]. 成都. 中国科学院水利部成都山地灾害与环境研究所. 2007. |
入库方式： OAI收割
来源：成都山地灾害与环境研究所
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。