In 2014, a serious of research tasks has been carried out,including collecting data,establishing norms,developing inversion algorithms,integrating production system,generating trial products,exploring product applications through data assimilation.We collected totally 12 kinds of raw satellite data products,observational data and reanalysis products. The amount of these collected data is over 700TB.In the algorithm R&D,the algorithms of 12 GLASS products have already been completed overall.Based on the collection of existing albedo product,we re-calculated the global background field of albedo,increased the spatial resolution of the background from the originally 5km to 1km. And we improved the algorithm of broadband emissivity in vegetation covered area.For AVHRR and MODIS data,separate split-window algorithms were developed for estimating land surface temperature.We also developed the upward long-wave radiation inversion algorithm based on the global cloudless skies.For the algorithm of all-wave net radiation, we collected relevant algorithms,developed the improved algorithm after evaluating exisiting algorithms validated and analyzed the trial product.The general idea of the fraction vegetation cover (FVC) algorithm is to train and test neural network algorithm based on the construction of the high spatial resolution global vegetation coverage sample data,and finally generating the FVC product algorithm.Five traditional surface latent heat flux algorithms were used in ET algorithm,based on more than 300 global land surface observation data,and by a fusion method according to the Bayes model,eventually, a Evapotranspiration product algorithm was formed.Based on Bayesian model,with the 8 current widely used international light efficiency models,a gross primary productivity(GPP) product algorithm was developed ultimately. In the system integration,all inversion algorithms have been coded,and integrated into the high-performance production systems.In the trial production, we have assessed the quality of all products and validated them preliminarily.We have also developed a data assimilation method with multi-source data,and explored the water cycle issue over northern China based on our remote sensing data products,and improved the forecasting ability of the model.%该研究2014年开展了一系统工作，分别包括数据收集、制定规范、算法研发、系统集成、产品试生产和同化应用等几方面。在数据收集方面，共收集了12种卫星原始数据产品、辐射能量观测数据以及再分析产品，共收集数据量超过700TB。在算法研发方面，总体上完成了12个GLASS产品的算法发展。基于收集的已有反照率产品重新统计了全球反照率背景场，把背景场的空间分辨率从原先的5km提高到1km；实现了植被覆盖区域宽波度发射率算法发展；针对AVHRR和MODIS数据，地表温度算法发展了各自的分裂窗算法；发展了基于混合算法的全球晴空上行长波辐射反演算法；净辐射产品的算法已完成算法收集、研发、验证及产品试生产等各项任务；FVC产品算法的总体思路为在全球高空间分辨率植被覆盖度样本数据建设的基础上，训练和检验神经网络算法，最终形成FVC产品生产算法；ET算法采用五种传统的地表潜热通量算法，根据收集的全球陆表300多通量观测数据，采用贝叶斯模型平均方法融合这五种算法，最终形成潜热通量产品生产算法；采用贝叶斯多算法集成方法，集合目前国际上应用广泛的8个光能利用率模型，最终形成GPP产品生产算法。在系统集成方面，对算法进行程序化与系统化，并集成到高性能生产系统。在产品试生产方面，初步完成12个产品试生产与初步验证。在同化应用方面，发展了多源陆面遥感产品的数据同化方法，基于我们生成的遥感数据产品，探讨了北方水循环问题，提高了模型的预报能力。
展开▼
