积雪的观测、模拟和多尺度遥感数据同化方法研究

The objective of this project is to develop a novel snow data assimilation scheme which can assimilate multi-resolution and multi-sensor remote sensing data such as passive microwave and optical remote sensing data. The Common Land Model (CoLM) is adopted as model operator to predict snow variables. The Ensemble Kalman Filter (EnKF) and Ensemble Multiscale Filter (EnMSF) will be used as data assimilation algorithm. In order to assess the CoLM performance at alpine meadow, firstly, continuous variation of seasonal snow will be observed at an integrated snow observation station located in Babaohe watershed. A set of high-quality snow dataset will be produced and can be used to develop and validate snow model and conduct assimilation experiments at alpine meadow. Different assimilation strategy will be designed and tested for multi-frequency passive microwave brightness temperature (such as AMSR-E) and MODIS snow cover area product, respectively. Finally, a multi-scale remote sensing data assimilation scheme will be developed for improving snow variables estimation via EnMSF algorithm. The project will help to understand and improve snow process parameterization in land surface model at alpine meadow, promote application of remote sensing in the field of land process modeling and hydrology, and provide an important approach for snow prediction in ungauged basins.

以黑河流域上游八宝河流域高寒草甸区布设的积雪综合观测场为基础，对雪深、雪水当量、土壤温湿廓线等相关变量进行长期连续观测，获得一套长时间序列的、可用于发展和验证积雪模型的、高质量的积雪观测数据集。分析和评价通用陆面过程模型（CoLM，Common Land Model）对高寒草甸区积雪变量（雪深、雪水当量、雪层密度、雪层温度等）的模拟精度，改进积雪参数化方案。发展针对MODIS积雪面积和被动微波亮度温度的积雪数据同化方案，利用集合多尺度滤波算法（Ensemble Multiscale Filter, EnMSF）实现多分辨率和多传感器遥感数据(光学和微波)的同化，提高积雪的模拟精度。本项目的开展，有助于认识和改善陆面过程模型对我国高寒草甸区积雪过程的模拟，促进和拓展遥感数据在陆面过程和水文领域的深入应用，为解决观测资料缺乏地区的积雪预报提供重要方法。

本项目重点开展积雪的观测、模拟和数据同化研究。在项目执行期间，分别在黑河上游垭口和北疆阿勒泰建立了积雪综合观测系统，获得一套长时间序列的、可用于发展和验证积雪模型的、高质量的积雪观测数据集。主要研究成果包括：（1）分析了黑河上游和北疆地区积雪的时空变化特征及驱动机制；（2）针对MODIS积雪面积产品的不足，发展了基于神经网络算法的山区积雪面积比例产品算法和基于多目标进化算法的MODIS积雪面积产品去云方法；（3）发展同化被动微波亮度温度、MODIS积雪面积产品、以及地面观测资料的积雪数据同化方法，构建了区域尺度积雪数据同化系统。目前已正式发表论文14篇，其中SCI论文6篇，EI论文1篇，国际会议论文2篇；培养研究生3名。