【作者】 陈巧；

【导师】 胡庭兴； 陈永富；

【作者基本信息】 四川农业大学 ， 森林经理学， 2003， 硕士

【摘要】 本文讨论了各种遥感监测土壤水分方法的适用范围和特点，并对各种方法进行了评价。基于以往方法的不足，在借鉴前人工作经验的基础上，本研究首次尝试应用了单时相乘积法监测土壤水分。该方法综合考虑地表温度和地表反射特征，用单时相1通道反射率与4通道亮温的乘积值直接与土壤相对湿度建模，并将该方法的计算结果与表观热惯量法和植被供水指数法进行对比分析，得出以下结论： 1．利用2002年4—7月的NOAA AVHRR 1.1km分辨率的遥感资料以及对应时段全国250多个气象观测站0—100cm的土壤湿度资料，分不同时段、不同植被条件、不同土层深度、不同方法建立统计模型，探导土壤湿度状况。该遥感资料覆盖了我国中轴线以东的大部分省、市、自治区，幅原辽阔，气候、植被类型多样。 2．分别用直线、对数、指数模型，分低、中、高植被覆盖和不同土层深度，对1、4通道乘积值、表观热惯量和植被供水指数分别与土壤相对湿度进行拟合，结果表明：单时相乘积法的直线模型能较好地估算低、中、高三种植被覆盖情况下10cm-70cm深度的土壤含水量，相对误差在20％以内，为直接预估深层次土壤水分提供了新思路。表观热惯量法的指数模型能较好地估算土壤表层(10cm-20cm)相对含水量，较适合于低植被覆盖或裸地的情况；植被供水指数法的指数模型能较好地估算20cm-50cm深度的土壤相对含水量，较适合于高植被覆盖的情况； 3．单时相乘积法的拟合精度优于表观热惯量法和植被供水指数法，不仅提高了在低植被和高植被覆盖的反演精度，而且还创造性地为中植被覆盖下的土壤水分反演提供了新方法。运用单时相乘积法反演土壤湿度，比在低植被条件下运用热惯量法在10cm、20cm的相对误差分别下降了4％、3％，平均下降了3.5％，比在中植被条件下运用植被供水指数法在10cm、20cm、50cm的相对误差分别下降了1.5％、3.5％、2％，平均下降了2.4％，比在高植被条件下运用植被供水指数法在10cm、20cm、50cm的相对误差分别下降了1.3％、1.4％、0.3％，平均下降了1％。 4．由于单时相乘积法只需要单时相卫星资料，且适合任意植被覆盖情况下、大范围的土壤湿度监测，解决了热惯量法因需要获得连续12h内两次晴空资料以及作物缺水指数法和农田蒸散双层模型等方法因依靠大量地面气象资料而不能解决的时效性和连续性问题。 5.经可替代性分析后，利用单时相乘积法建立了4月至7月份的通用模型，其模型为:10e血:卜113.69一1.44Q;20em:W=113.13一1.36Q;50em:W=110.74-1.23Q，经严格的统计学检验，证明其具有一定的通用性，并用此模型反演了5月31日、7月30日0一20cm的平均相对土壤湿度，取得了较好的效果。 综上所述，单时相乘积法适合于我国中东部地区4一7月任意植被覆盖下、大范围的土壤湿度实时监测。该方法解决了以往不能连续监测土壤含水量、监测范围小、所需气象资料太多的问题，获取资料方便，计算简便，能够较好地保证土壤湿度监测的实时性和实效性。更多还原

【Abstract】 In the research work, by analyzing the NOAA/AVHRR (resolution is 1. 1km) data(from April to August in 2000) and soil humidity(0-100cm) of the corresponding period of NOAA/AVHRR data observed by more than 250 meteorologic stations , the statistical models have been made out between the NOAA/AVHRR data and the soil moisture index(W) according to different periods, different soil deepness, different surface cover attributes, different methods and different regressive functions. And the NOAA/AVHRR data cover most provinces, cities and municipalities of China that include large scale and many kinds of climates and vegetations.This study discusses the applying scopes and the characteristics of the apparent soil inertia method and the vegetation supply water index method, and then values the two methods. Because of the shortages of the two methods, and on the basis of the former investigator’ s study, single phase product method is put forward in the study first. The method considering surface temperature and reflection attributes and was built between the value of product of channel 1 and channel 4 of single phase NOAA/AVHRR and the soil moisture index, and then the result of single phase product method was compared with the result of the apparent soil inertia method and vegetation supply water index method. The conclusions are as follows:1 ) L.inear, logarithmic and exponential empirical models were applied to develop the relation ship between soil moisture index (0-100cm) and apparent soil inertia and between soil moisture index (0-100cm) and vegetation supply water index and between soil moisture index (0-100cm) and the value of the product of channel 1 and channel 4 of single phase NOAA/AVHRR respectively. The result indicates that the logarithmic models of apparent soil inertia method has better performance in estimating soil surface(10-20cm) moisture index than others for bare soil or lowly covered soil. And the logarithmic models of vegetation supply water index method has better performance in estimating the deepness of 20~50cm of soil moisture than others for highly covered soil. And the linear model of single phase product method has better performance in estimating the deepness of 10-100cm of soil moisture than others for any level covered soil, and the mean relative error is controlled within 20% on the whole, it offers a new idea for using variable estimate deep soil moisture directly.2) The result indicated that the precision of the model built by single phase product method is better than the models built by apparent soil inertia method and by vegetation supply water index method. It not only improve the precision for lowly soil and highly soil, but also offers a new method for partly covered soil creationaryly. The retrieved values of single phase product method is 3% and 4%lower than that of apparent soil inertia method and the mean is reduced 3. 5% for lowly covered soil at the deepness of 10cm and 20cm respectively; it is 1. 5%, 3. 5% and 2% lower than that of vegetation supply water method and the mean was reduced 2.4% at the deepness of 10cm, 20cm and 50cm for the partly covered soil respectively; it is 1. 3%, 1. 4% and 0.3% lower than that of vegetation supply water method and the mean is reduced 1% at the deepness of 10cm, 20cm and 50cm for the highly covered soil respectively.3) It can resolve those questions about efficiency and continuity which can’ tbe resolved by apparent soil inertia method because of its ordered two-phases cloudless remote sensing data in 12 hours and by Crop Water Stress Index method and so on because of its ordered too many ground weather data. In addition it could be used to monitor soil moisture in large scale with any level covered soil.4) Currency model for April to August was built by using single phase product method after analyzing the replaceable between each model. The currency model was as follows: 10cm: W=113.69-1.44Q; 20cm: ff=113.13-1.36Q; 50cm: 1=110.74-1.230, all of them were qualified in terms of statistic test and had certain degree更多还原