【作者】 杨争；

【导师】 丁丽霞；

【作者基本信息】 浙江农林大学 ， 森林经理学， 2012， 硕士

【摘要】 山核桃是浙江省临安市的重要经济树种，是当地农民的主要收入来源。及时准确了解山核桃种植面积、长势、产量信息，对于政策的制定、价格的宏观调控等都具有非常重要的意义。目前山核桃遥感监测研究在我国并不成熟的情况，本文应用遥感技术，结合样地调查，利用山核桃冠层光谱特征和主要农学参数、气象因子建立遥感估产模型，主要研究内容如下：1.提取山核桃面积本文以浙江省临安市为研究区，利用冬夏两个不同时相的CBERS-02卫星数据与实地调查资料，采用主成分分析与监督分类相结合的方法，提取浙江省临安市的山核桃空间分布及面积信息。2．建立单产模型分析不同生育期植被指数(NDVI、PVI、RVI、DVI)、气象因子与山核桃产量的相关性，确定开花期的NDVI为最佳估产遥感因子，与山核桃产量有显著性相关的气温、降雨量等18个气象因子中，有10个为十分显著的气象因子，其中7月的旬平均降水量对产量的影响最大。采用相关分析和线性回归的方法，建立以植被指数NDVI和气象因子为变量的复合遥感估产模型，并用实际统计的产量数据作为参照数据，采用相关系数评价和根均方差（RMSE）评价方法对建立的模型进行精度分析，筛选最佳单产模型。通过研究，本文主要得到以下结论：1.根据山核桃冬夏两季的林相差异，利用遥感变化检测方法提取山核桃空间分布及面积信息，提取的精度为83.71%，Kappa系数为0.81。提取山核桃总面积为3.11万公顷，与统计数据对比分析，总面积误差为1.3%。该方法是可行的，满足山核桃遥感信息提取的需要。2.经精度分析知，遥感因子植被指数与气象因子获取的多元线性回归模型精度高，其拟合R2为0.851，检验R2为0.820，F值为52.536，残差RMSE为7.78，模型易于使用，且稳定性较好，是山核桃估产的有效工具。更多还原

【Abstract】 Carya cathayensis is an important economic species in Lin’an of Zhejiang Province, is the mainsource of income for local farmers. So it was greatly significant important to understand the plantingarea, growth and yield for the policy development, macro-control of price. The remote sensing modelfor Carya cathayensis yield estimation was to be established in Lin’an, Zhejiang Province based oncanopy spectrum and main agriculture parameter. The main achievements are as follows:1. Planting areaTwo scenes of the CBERS-02data in the summer and winter and field survey data were dealedwith principal component analysis and supervised classification methods, the spatial distribution andarea information of Carya cathayensis in Linan of Zhejiang Province were extracted.2. Yield estimation modelAfter analysis of the relevance of vegetation index (NDVI, PVI, RVI, DVI), meteorologicalfactors and Carya cathayensis yield in the Carya cathayensis different growing periods, NDVI is thebest estimation of the remote sensing factor in the growing period,flowering period is the bestestimation of the growing period,among18meteorological factors,10meteorological factors has thegood relevance with the Carya cathayensis yield,average precipitation in July ten-day period have thegood relevance with the Carya cathayensis yield.In the optimum period of Carya cathayensis yieldestimation,established the Carya cathayensis prediction model.With a regression analysis of thevegetation index NDVI、meteorological factors and Carya cathayensis yield.Comparison with analysisof statistical data,using the correlation coefficient evaluation and the root mean square error(RMSE)evaluation method to build the model accuracy。Through research,the following conclusions of this paper:1.The overall classification accuracy of83.71%,Carya cathayensis planting area of3.11millionhectares. Comparison with analysis of statistical data,the total precision error of1.3%.The resultsshowed that using the differences of winter and summer images,the extraction of area information isfeasible by remote sensing-based dynamic monitoring.2.The regeression analysis of vegetation index and meteorological factors,attained themulti-dimensional linear regression model is better,predicting values of0.851,the accuracy of fittingof0.820,F of52.536,RMSE of7.78,the maximum relative prediction error can be up to14.78%,and the smallest relative predicition error is5.67%.The results showed that the model is easy to use,andthe model stability is better,which can be used an effective tool for prediction of Carya cathayensisproduction.更多还原