积雪是地球表面最为活跃的自然要素之一，积雪强烈影响着地表能量交换、水循环和辐射平衡。在天气预报、气候模型、水文模型应用和全球气候变化研究中，积雪作为其中主要的参数之一，具有重要的意义。在许多山区，季节性积雪是主要的水资源来源。积雪为世界上许多半干旱地区提供的农业和供水所需水源占到了相当大的比例，在这些地区，积雪的变化会造成巨大的经济和社会效应。因此，常规的积雪信息监测已经成为水资源管理与利用和全球气候监测的优先级业务。不同平台、不同类型的卫星传感器为积雪判识提供了多种数据源。极轨卫星传感器数据的空间分辨率相对较高，但由于每日观测次数有限，受云影响较为严重，多天合成积雪产品虽然最大化减少了云的影响，但无法满足日尺度上地表能量和水分交换过程等研究。被动微波遥感数据虽然具有全天时、全天候、不受云等天气情况影响的特点，但数据的空间分辨率通常较低（20-50 km），而且受地表异质性影响较为严重，积雪判识的精度较低。近年来，我国南方大范围降雪等极端天气事件，对积雪信息的准确获取提出了更高的要求。南方地区降雪多为湿雪和浅雪，目前的被动微波遥感积雪算法仍无较好的判识效果；此外，南方地区在降雪期间，通常同时受大片云的影响，凸显了云对光学遥感数据的影响。另一方面，随着我国卫星遥感技术不断发展，国产卫星遥感数据在自然灾害、天气预报、地表参数反演等研究中具有重要应用潜力。如何充分利用国产卫星遥感数据，并发挥多源遥感卫星数据的优势，及时、准确获取高精度的积雪信息是当前研究的重点和难点。本文研究即在上述背景下产生，本文的主要研究内容概括如下：（1）发展基于风云二号静止气象卫星双星（FY-2D和FY-2E）融合的中国地区雪盖判识算法。分别考虑有雪地表（森林和非森林地区）、无雪地表和云的光谱特征，使用雪盖指数、亮度温度差异等判识因子建立积雪判识算法。充分利用静止气象卫星数据高时间分辨率的特点，结合时间序列变化信息，提出多时相云检测方法，减小云雪误判率。此外，使用每日多时相雪盖图合成、时空滤波以及被动微波遥感数据（风云三号B星微波成像仪）辅助判识等方法进一步减小云对雪盖图的影响，最后生成风云气象卫星融合雪盖图。研究充分利用了国产风云气象卫星数据，发挥了静止气象卫星数据和被动微波遥感数据在积雪判识上的优势。经2010年和2011年冬季的全国气象站点观测数据验证，结果表明，风云气象卫星融合雪盖图总体精度约为91.28%。（2）通过对风云气象卫星融合雪盖图的精度验证与分析发现，青藏高原地区的积雪判识精度较低且波动性较大，结合青藏高原地区独特的积雪特征（浅雪、斑块状雪），本文利用2007-2010年冬季气象站点观测数据和高分辨率Landsat TM影像数据验证覆盖青藏高原地区的MODIS、IMS每日雪盖产品和AMSR-E每日雪水当量产品，分别从“点”和“面”尺度详细评价积雪产品的精度。研究表明，MODIS雪盖产品总体精度高于IMS雪盖产品，IMS雪盖产品存在较为明显的高估雪盖的现象，MODIS雪盖产品受云影响则较为严重。此外，对比分析2010和2011年冬季的MODIS、IMS雪盖产品和本文发展的风云气象卫星融合雪盖图，结果表明，风云气象卫星融合雪盖图的总体精度高于IMS雪盖产品。此外， AMSR-E雪水当量产品的雪水当量值高估现象较为严重，算法有待进一步改进。研究还表明，卫星遥感雪盖产品的精度受下垫面地表类型、雪深范围、雪盖图破碎度等影响。（3）本文所发展的风云二号静止气象卫星雪盖图为中分辨率，混合像元问题需要考虑，因此，本文进一步以青藏高原为研究区，发展基于风云二号E星（FY-2E）经验关系积雪覆盖度反演算法。借鉴极轨星雪盖指数反演积雪覆盖度方法，考虑FY-2E静止星数据特点，利用FY-2E可见光波段和中红外波段数据，分别建立差值、比值和归一化差值雪盖指数，并分析三个指数与积雪覆盖度之间的关系，建立经验关系积雪覆盖度反演算法。此外，本文使用人工神经网络算法反演风云二号E星积雪覆盖度，并对基于雪盖指数的经验关系反演算法和人工神经网络算法反演结果进行对比分析与验证。结果表明，基于差值雪盖指数算法和人工神经网络反演方法具有较高的精度，其中，基于差值雪盖指数的积雪覆盖度反演R2（决定系数）约为0.61，RMSE（均方根误差）约为0.17；基于人工神经网络方法的积雪覆盖度反演R2约为0.66，RMSE约为0.17。（4）在风云二号双星融合雪盖判识算法的基础上，本文进一步利用国内外极轨轨道、静止轨道卫星数据和积雪产品，探讨多源传感器雪盖图融合方法，提出基于风云气象卫星数据为主的中国地区多源传感器联合积雪判识算法。本文分别对多源静止气象卫星数据、多源被动微波遥感数据进行对比分析与相对交叉定标，使多源数据能联合用于积雪判识研究，在此基础上，主要对各类型雪盖图进行精度分析与对比，详细评价各类型雪盖图在中国地区的表现和优劣势，并以此探讨和建立雪盖图融合算法。结合气象站点观测数据和当前国际上主要的卫星遥感积雪产品对多源传感器联合积雪判识结果进行精度验证和对比分析，结果表明，本研究所得的多源传感器雪盖图精度约为90.72%（站点验证），总体精度接近甚至高于国际IMS雪盖产品。综上所述，本研究以我国风云气象卫星数据为基础，重点发展风云二号静止气象卫星双星融合雪盖判识算法、经验关系积雪覆盖度反演算法和多源传感器联合积雪判识算法，提高国产卫星遥感数据在中国地区的积雪判识精度，有助于更准确、及时、有效地监测积雪信息。

Snow is one of the most active natural elements in the Earth’s surface. Surface energy exchange, water cycle and radiation balance were affected by the snow cover. In addition, snow cover is an important parameter in weather forecasting, climate modeling, hydrological models and global climate change researches. Seasonal snow cover is the main water resource in many mountainous areas. In semi-arid regions, snow cover plays important role in agriculture and water supply, whose variations would cause huge economic and social effects. Therefore, daily snow cover monitoring has become a priority service in water resource management and global climate monitoring.There are different platforms and types of satellite sensors could be utilized to obtain snow cover information. Polar-orbiting satellite data presented high spatial resolution, however, it was serious affected by cloud obscuration. Although multiple day’s composited snow products could reduce the cloud obscuration, it can not satisfy the daily scale surface energy and water exchange researches. On the other hand, passive microwave data is without the effect of cloud, but the spatial resolution is low (20-50 km). Due to the effect of surface heterogeneity, the snow detection accuracy is lower than optical data. Recent years, extreme weather, such as snowfall events in southern China put forward higher requirements of more accurate, timely snow cover information monitoring. In southern China, the snow always present wet or shallow, which is hardly detected by the current passive microwave snow algorithms. In addition, during the snowfall events, the regions usually affected by large cloud, which could affected the snow cover monitoring by optical remote sensing data. On the other hand, although there are more and more Chinese satellite data could be used, the researches and applications of Chinese satellite remote sensing data are still insufficient, such as natural disasters, weather prediction and surface parameters retrieval. How to make full use of different types of satellite remote sensing data, including Chinese satellites data, and obtain timely and accurate snow cover information is important and difficult in current researches.The main purpose of this paper is to improve the accuracy of snow cover mapping over China. The main contents of this paper can be summarized as follows,（1）A new snow cover algorithm based on Fengyun-2 (FY-2D and FY-2E) geostationary satellites was developed. Spectral characteristics of snow-covered surface (non-forest and forest regions, respectively), snow-free surface and cloud were considered in the development of snow cover algorithm. Snow index, brightness temperature difference and other indices were used in the snow cover algorithm. In addition, taking advantages of high temporal resolution, a multi-temporal cloud detection method was proposed to reduce the classification errors between snow-covered surface and cloud. In order to reduce the effect of cloud, combination of multi-temporal snow cover maps, spatial-temporal filtering and the supplement of passive microwave snow cover maps were used. In this study, taking advantages of geostationary and passive microwave data, more accurate snow cover map was created based on Chinese meteorological satellites data. Meteorological stations observation in 2010 and 2011 winter seasons were utilized to validate the accuracy of snow cover maps. Results indicated that the snow cover maps proposed in this study show high overall accuracy (~91.28%).（2）More comprehensive evaluation of snow products over the Tibetan Plateau was carried out. Compared with other snow-covered regions in China, snow accuracy over the Tibetan Plateau was lower and presented higher variations. Snow cover over the Tibetan Plateau always presented as shallow and pathy, hence, several international snow products were evaluated and analyzed in this study. Meteorological stations observation, Landsat-5 TM data were utilized to evaluate the performance of daily MODIS snow cover products, IMS snow cover products and AMSR-E SWE products in 2007-2009 winter seasons. Results indicated that MODIS snow products presented higher overall accuracy than IMS snow products. IMS snow product overestimated the snow cover, while the MODIS snow products were serious affected by the cloud. Compared with Fengyun satellite snow cover map proposed in this study in 2010 and 2011 winter seasons, IMS also presented lower overall accuracy. In addition, AMSR-E SWE product showed over-estimation of the SWE value, which indicated that the algorithm still need to be improved. The accuracy of satellite snow products is related to land-cover types, snow depth and the fragmentation of snow maps. （3）Fengyun snow cover maps proposed in this study was medium spatial resolution, hence, the mixed pixel problem should be considered. Empirical fractional snow cover algorithms based on FY-2E geostationary satellite VISSR data (VIS and IR4 channels) were developed over the Tibetan Plateau. Landsat TM and ETM+ data were utilized to obtain the true value of fractional snow cover. Ratio Snow Index (RSI), Difference Snow Index (DSI) and Normalized Difference Snow Index (NDSI) based on FY-2E VISSR data were compared and utilized to retrieve the fractional snow cover. In addition, we used Artificial Neural Network (ANN) method to retrieve the fractional snow cover. Results indicated that the DSI (R2=0.61, RMSE=0.17) and ANN (R2=0.66, RMSE=0.17) methods presented better performances than the others.（4）Multi-sensor snow cover mapping and snow depth retrieval method was proposed. Different types of satellite data were compared and utilized to obtain 5km snow cover map and 25km snow depth map over China. The snow cover map is completely without the impact of cloud, while the snow depth map could full cover the whole China in daily basis. In order to use multi-sensor data more accurate, calibration of different geostationary satellites data, calibration of different passive microwave data were carried out respectively. The performance of different snow products were analyzed and compared, then a multi-sensor fusion method was proposed. In addition, meteorological stations and current major snow products were utilized to validate and compare the multi-sensor snow maps. Overall accuracy of the multi-sensor snow cover map proposed in this study was 90.72%, which was similar and even higher than IMS snow cover products.In summary, snow cover algorithm, fractional snow cover algorithm and multi-sensor snow mapping method were proposed based on Chinese Fengyun satellites. This study improves the overall accuracy of satellite remote sensing snow mapping over China, and thus contribute to a more accurate, timely and effective monitoring of snow cover information.