【作者】 张旭东；

【导师】 迟道才；

【作者基本信息】 沈阳农业大学 ， 农业水土工程， 2014， 博士

【摘要】 粮食和水问题是全球性的、国际性的问题,是关系到全球人类生存和社会安定的战略问题。东北三省是我国的粮食主产省区和商品粮基地,水稻作为国人的主要口粮其地位非常重要,其生产过程也是农业灌溉用水消耗的大户,目前,全球水资源紧张、行业用水竞争加剧,环境、生态、社会、经济用水矛盾突出,在此背景下,本文以遥感技术和微观试验相结合的方法对东北三省的水稻水分生产率(WP)进行了评估,围绕作物产量和水分消耗之间的关系进行探讨,为确保粮食安全和农业水资源的可持续利用提供理论依据。本论文根据29个农业气象试验站确定了东北三省水稻的生育期(即本研究的研究时段)及分区方法(以提高研究结果的精度),并分析了水稻产量的空间分布特征及规律；在分区的基础上,对东北三省综合利用TM数据、时间序列eMODIS数据、MOD12Q1土地覆盖类型数据、DEM数据等异源多时序数据,采用基于专家知识经验构建的决策树对2009年水稻的种植面积进行提取；采用新版MOD16算法,对不同分区利用MODIS产品(包括MOD12Q1土地覆盖类型、MOD15A2植被指数产品、MCD43B3反照率产品),结合GMOA的气象数据,对东北三省2009-2010年水稻的蒸发蒸腾量(ET)进行了反演；构建了水稻产量空间化模型,并依据该模型将水稻产量数据外推到像素级,得到了东北三省2008-2010年水稻的产量分布图；根据产量图和ET图,采用逐像元相比的方法得到了的该区域的水稻水分生产率图；在试验站布置不同尺度、不同水分条件的灌溉试验,利用同时段的灌溉试验结果对基于遥感的结果进行验证。本论文的主要成果及创新点有：1)提出了基于异源多时相遥感数据的作物种植面积决策树提取方法,适宜于大范围的水稻面积提取,且精度满足要求。2)以地级市行政区划为分析单元,构建的基于水稻生育期内最大NDVI和相应分析单元平均单产之间关系的水稻产量空间化模型,该模型能从整体上控制误差,通过NDVI将统计产量扩展到像素级水平,较好地解决了传统统计产量数据难以空间化的问题。3)基于MOD16遥感反演ET模型,首次对北纬40-50°之间大范围的水稻生育期耗水量、水分生产率进行了估算,提出了东北三省2008-2010年的水稻水分生产率图,水稻水分生产率的区间为1.26-2.58,平均值为1.890,其中最高值分布北纬44-46°之间。4)设置了盆栽、测坑、小区三个不同尺度不同水分控制条件下的水稻水分生产率试验,以其结果对基于遥感的水分生产率研究成果进行验证,并寻找两者共同存在的规律或特征,为基于卫星遥感的宏观方法和基于田间试验的微观手段在作物水分生产率方面的研究提供了新的有益尝试。更多还原

【Abstract】 The problem about food and water relates to human existence and social stability, it is a strategic and challenging issue to solve it across the globe. In Northeast China, the three provinces are production base and commercial center for major grain. Being one of the mainfood sources, rice plays a significant role in the agriculture production in China. However, the process of rice growth consumes amounts of irrigation water. At present, water storage across the globe and the usage contradiction among different industries are becoming increasingly intensified. In these contexts, we estimated the rice water productivity (WP) using the method of remote sensing combined with micro-test of field investigation. The relationship between crop yield and water consumption was discussed. The study is expected to provide theoretical guidance for the sustainable utilization of agricultural water resources and the stable production of rice in the Northeast China.This paper specified the rice growth periods (i.e., the periods of this research) and the partition method (in order to improve the accuracy of the results of this research) in the three provinces based on the29agro-meteorological experimental stations, and analyzed the spatial distribution of rice yield as well as its characteristics.Using multi-sensor and multi-temporal remote sensing data derived fromTM, time-series NDVI, MOD12Q1and DEM, the rice planting area in three northeast provinces of China was extracted based on expert Knowledge decision-tree and the division in2009. The evapotranspiration(ET) of rice for different partitions was estimated using the new MOD16algorithm with MODIS products(including MOD12Q1land cover types, MOD15A2vegetation index products, MCD43B3albedo products) combined with GMOA meteorological data from2009to2010in the three provinces. The paper proposed a rice spatialization model of yield which extrapolated the yield to the pixel level. Additionally, the rice distribution maps were showed from2008to2010. According to the production map and ET map, the method of pixels compared one by one was used to get the water productivity map in rice area. At last, the irrigation experiment results in different scale and water condition at the same time periods were compared with the results based on remote sensing for validation.Main achievements and innovations of this paper were as follow:1) a decision tree method using different sources of remote sensing data to extract crop planting area was proposed. This method is verified to have high precision;2) based on the administrative divisions prefecture level analysis, we put forward a space model according to the relationship between rice yield and the maximum NDVI in rice fertility period of the corresponding unit, which expanding statistics yield to pixel level and controlled the error in each unit. The model provides a good solution to the hard traditional problem of the spatialization of the statistical yield data.3) Based on remote sensing inversion MOD16ET model, the paper estimated water consumption and water productivity of rice in a wide range for the north latitude40°-50°in crop growth period for the first time. It concluded that the rice water productivity changed from1.26to2.58with a mean of1.890in the three provinces in2008-2010and the highest value was in the north latitude44°-46°which was reasonable compared with the experiment results. Comparison among the results of the same field showed that the rice water productivity of the three northeastern provinces was significantly higher than the south areas of China, and was higher than that in other rice-growing areas of the world.4) we designed three different scales-the pot (0.07m2), pit (5m2) and community (60m2) under different water conditions for experiment of rice water productivity, the results were similar to these based on remote sensing. This attempt to provide a new paradigm for cooperative research on crop water productivity combined macroscopic satellite remote sensing with microscopic field experiments.更多还原