【作者】 李为萍；

【导师】 史海滨；

【作者基本信息】 内蒙古农业大学 ， 农业水土工程， 2004， 硕士

【摘要】 精细农业日益成为高新农业技术领域研究的热门课题，而土壤水盐信息和作物信息的空间结构性研究是发展精细农业所必需的技术支撑。在此背景下，本文以内蒙古河套灌区节水灌溉试验地中新疆产康地105号油料向日葵生育指标及土壤水盐信息为研究对象，系统研究两者的空间结构性及其协同关系。在论文中将影响系数法进行补充引用到对土壤水盐及作物信息试验数据的特异值处理中。运用以交叉协方差函数形式表达的协同克立格法（Cokriging，简称CK法）进行两变量协同关系研究的同时，尝试进行了作物生育指标与土壤含水率、EC值三变量间协同关系的研究。在向日葵现蕾期（2003年夏）测定其株高、茎粗等生育指标及试验地土壤含水率、EC值。将株高、茎粗、土壤含水率及EC值视为区域化变量，运用地质统计学的基本理论对其进行空间结构性分析。将向日葵株高与茎粗的乘积记为综合性指标并视为主变量，选取土壤含水率与EC值为辅助变量，分别运用CK法与普通克立格法(Ordinary Kriging，简记为OK法)对向日葵综合性指标进行预测研究。通过土壤水盐与向日葵综合性指标三变量间的协同关系可得一估值样本，将土壤含水率与EC值同向日葵综合性指标分别建立协同关系可得两估值样本，同时利用OK法也可获得一估值样本。将所得到的4个估值样本与实测样本从不同角度进行系统的比较分析，结果表明CK法估值方差较OK法估值方差明显降低，且三变量CK法估值方差小于两变量CK法估值方差。运用Jacknife法对估值样本精度进行比较，得出OK法Jacknife值要小于CK法对应值。同时对实测样本及4个估值样本进行统计描述，表明CK法估值样本的方差、标准误差等统计特征数较OK法估值样本更接近实测样本。三变量CK法估值样本的变异系数最接近实测样本，而OK法估值样本的变异系数与实测样本的变异系数差值最大。表明增加辅助变量的个数，可以变相增加主变量的信息量从而可以更细微的刻画出主变量的空间结构性，同时本文的研究结果与Yates and Warrick(1987)研究结果相一致。更多还原

【Abstract】 The precision farming has been becoming the hot topic in the scope of agriculture technology, and the researches on the spatial variability of soil water-salt information and crop information are the technical support of implementing the precision farming. So in this thesis, the growing index of sunflower (Kang Di 105), soil water conten and soil EC are chosen as researching objectives, and the spatial variability of them and their correlation are researched. At the same time, the influencing coefficient method (ICM) is complemened and intrduced into the scope of soil water salt and crop information. The Cokriging is marked as CK and expressed by covariance function. The correlation of two variables is researched, and the coorelation of soil water content, soil EC and growing index of crop is also attempted researching.During the squaring stage, in 2003, the plant height and stem-diameter of sunflower have been measured. The soil water content, EC, the plant height and stem-diamete are regarded as regionalized variables, and the spatial variability of them are researched using the basic theory of geostatistics. Then product between the plant height and stem-diameter of sunflower is marked as the integrated index and regarded as primary variable. Choosing water conter and EC as secondary variable, the integrated index of sunflower is estimated using the three variables CK method that is in the form of covariance function. At the same time, the soil water content and EC is chosen as secondary variable respectively, the two estimation sets of the integrated index are obtained using the two variables CK method. Another estimation set is getted using Ordinary Kriging that can be markde as OK.The four estimation sets are analyzed and compared with the original set in different aspects. We can get the results that the estimation variance of CK is less than the estimation variance of OK, and estimation variance of three variables CK method is less than corresponding value of the two variables CK method. Using the method of Jacknife to compare the precision of the estimation sets, we find the Jacknife value of OK is less than the value of CK.Through statistically analyzing the four estimation sets and original set, we can get that the variance, standard deviation of the Cokriging estimation sets are more close to the corresponding values of the original set than the OK estimation set. But the coefficient of variation of the three variable CK is the most close to the coefficient of variation of the original set. So it <WP=4>indicates that increasing the number of secondary variable can increasing the information of the primary variable and can depict the spatial variability of the primary variable minutely. The results of the thesis are consistent with the results of Yates and Warrick (1987).更多还原