随着全球气候显著变暖，干旱发生的频率和强度不断增强，干旱地区的扩大与干旱化程度日趋严重，干旱化趋势已成为全球关注的问题。干旱、缺水已严重妨碍了国家经济、社会发展以及人民生产生活。传统旱情监测方法是对气象站点观测资料统计分析，获得用于评估旱情的干旱指标。但气象站点分布离散，其在空间上的监测精度受制于气象站点的分布密度。遥感作为一种新型的对地观测综合性技术，具有时效性强、覆盖范围广、客观准确及成本低等特点。在干旱监测中引入遥感技术，可使得旱情监测结果更为精细和准确。然而，大多遥感干旱指数被设计用于反映地表综合的干旱程度，很少有针对特定作物类型，比如：玉米作物。另外，由于干旱指标大都建立在特定的地域和时间范围内，有其相应的时空尺度，单个干旱指标很难达到时空上普遍适用的条件。针对上述问题，本文以玉米作物旱情监测为目标，以时序遥感影像和气象观测站降雨数据为基础，建立了一种顾及作物物候特征的多干旱指数联合监测方法。即利用检测的玉米作物物候信息校正以遥感为数据源的植被状态指数（VCI），利用结构推理的方法融合修正的VCI指数和标准降雨指数（SPI），实现对玉米作物旱情的联合监测。因此，本文主要的研究工作和成果表现为：1）遥感影像不规则兴趣区的分维估计算法针对玉米作物耕地呈块区（不规则兴趣区）在遥感影像上分布的特点，利用分形乘积的原理，设计了一种降维-差分计盒维数法（Dimensionality-Reduction based Differential Box-Counting algorithm，DR-DBC），以实现对遥感影像不规则兴趣区的分维估计。2）分维与玉米作物物候的相关性检验针对玉米作物生育期遥感时序影像分维变化的特点，建立了分维时序峰值与玉米作物物候的对应关系，并利用地面调查数据作了实验验证。通过构建一系列的对比因子和对比指标，对分维鲁棒性进行了检验，验证了该指标用于表征玉米作物发育状态的可行性。3）玉米作物物候的动态估计算法针对实际应用中对玉米作物物候信息现时性需求，考虑以NDVI均值、分维值和有效积温为数据输入，行政区划为目标单元，构建以隐形马尔可夫模型（Hidden Markov Model，HMM）为基础的玉米作物物候信息动态估计框架。通过实例验证，并与常规逐像元物候检测方法对比分析，验证了该方法的有效性。4）物候调节植被指数（PA-VCI），及其与SPI指数互相关性检验针对VCI指数以日历年为时间基准，忽略了作物本身物候变化的缺陷，提出了一种物候调节植被状态指数（Phenology Adjusted Vegetation Condition Index，PA-VCI）。通过与常规VCI指数对比分析和实例验证，验证了利用作物物候信息修正VCI指数的必要性，并检验了PA-VCI指数与SPI指数的互相关性。5）PA-VCI指数与SPI指数的融合算法考虑到对遥感干旱指数和气象干旱指数优劣互补的必要性，提出利用结构推理的方法对PA-VCI指数和月时间尺度SPI指数进行融合处理，实现对玉米作物旱情的监测，从而为防灾减灾、国民经济和社会的可持续发展提供决策支持。

With the global climate warming, and the increasing of drought frequency and intensity, the arid regions are expanding and its drought level has become increasingly serious. The drying trend has became a global concern. Drought and water scarcity have seriously hindered the national economic and social development, as well as people's production and living conditions. The traditional drought monitoring methods are most statistical analysis based on the data from meteorological stations, to gain an index of drought for evaluating drought. However, because the distribution of meteorological stations is discrete, the monitoring precision in space will be limitted by the distribution density of meteorological stations. As a new comprehensive technology of earth observation, remote sensing possesses the characteristics of fine timeliness, wide coverage, strong objectivity and accuracy as well as low cost. Introducing the remote sensing technology in drought monitoring will bring more precise and accurate drought monitoring results. However, most existing remote sensing drought indices are designed to reflect the macroscopic drought level of the earth surface, and few are designed for a specific crop such as corn crop. In addition, due to the drought indices are mostly based on specific geographic and time range, possessing its corresponding scale, a single drought index can’t be universally applicable in space and time.In view of the problems mentioned above, this paper aims at drought monitoring of corn crop, and establishes a joint monitoring method of multi-drought indices, regarding crop phenological characteristics, basing on temporal remote sensing images and precipitation data from meteorological stations, namely to modify the Vegetation Condition Index(VCI) based on remote sensing data with the help of the detected progress stage information of the corn crop and to achieve joint monitoring of the corn crop drought with the method of structure inference, integrated with the modified VCI as well as the Standard Precipitation Index (SPI). Therefore, the main research work and achievements of this paper show as follows:1) Fractal dimension estimation for irregular region of interest (ROI) of remote sensing images Concerning the distribution characteristics of corn crops cultivated in block area (irregular ROI) in the remote sensing images, the paper utilizes the principle of fractal product to design a Dimensionality-Reduction based Differential Box-Counting algorithm (DR-DBC), in order to achieve the fractal dimension estimation for irregular ROI.2) Correlation between fractal dimension and corn progress stage According to the characteristics of fractal dimension change of remote sensing temporal images in corn crop growth period, the paper establishes the corresponding relations between fractal dimension time series and corn crop development stage, and carries out the experimental verification with the ground survey data. The paper also examines the robustness of fractal dimension and verifies the feasibility of applying these indicators in characterization of corn crop growth state through building a series of contrast factors and indicators.3) Real-time estimation of corn crop development stageFor the real-time requirements for corn crop phonological information in practical application, the paper considers taking the mean NDVI, fractal dimension value and the accumulated growth degree days (AGDDs) as data input, and taking the administrative divisions as target units, to build a real-time estimation framework of the percentage of corn crop growth stages based on Hidden Markov Model (HMM). The paper also verifies the effectiveness of the method through example verification and comparing with the conventional pixel by pixel, detection method.4) Phenology Adjusted Vegetation Condition Index (PA-VCI), and correlation test of PA-VCI and SPI index Because VCI takes the calendar day as time reference, it ignores the crop phenological changes. Aiming at solving this defect, the paper puts forward a kind of Phenology Adjusted Vegetation Condition Index (PA-VCI). By comparing with the VCI index and carrying out some example verification, the paper validates the necessity of modifying VCI index by using crop phenological information, and tests the correlation of PA-VCI and SPI.5) Fusion between PA-VCI and SPI indexConsidering the complementary necessity of drought index from remote sensing and meteorological stations, the papers puts forward the fusion processing of PA-VCI index and monthly-scale SPI index by using the method of structue inference, to monitor the corn crop drought. It can provide some decision support for disaster prevention and the sustainable development for the national economy and society.