遥感和GIS支持下的分布式融雪径流过程模拟研究

Study on the Distributed Snowmelt Runoff Process Based on RS and GIS

基于遥感(RS)和地理信息系统(GIS)技术,设计和构建了一个分布式融雪径流模型,整个分布式融雪径流过程的模拟计算基于能量平衡和水量平衡,由分布式栅格融雪过程、分布式栅格产流过程以及分布式栅格汇流过程组成,融雪以及产汇流过程全部基于栅格尺度,全面实现了融雪过程的分布式模拟。分布式栅格融雪过程中对"度日法"加以改进,引入了"单元时段"的概念,从而得到了"度分融雪模型";针对融雪过程中颇为复杂的冻融反复性难题,提出了旨在解释积雪冻融反复性物理机制的"冻融系数"的重要概念,对于准确把握融雪过程的物理机制具有重要意义。同时基于G IS开发了分布式融雪径流模拟系统,为分布式融雪径流模型的运行提供了平台和技术支持,二者均为融雪洪水预警决策支持系统的核心模块部分。基于由MODIS等遥感数据得到的积雪信息、地表温度等下垫面信息,基础地理信息数据如DEM及其空间分析数据和大量野外同步观测数据(积雪信息、气象数据),对典型研究区新疆军塘湖流域2006年春季典型融雪期(2006-03-06,11:00—2006-03-10,11:00)内的洪水过程进行了模拟,模拟结果精度较高,平均精度0.82,达到了融雪洪水预警预报的业务需求标准。

Snowmelt runoff is the main composition of water resources in middle and high latitude region.Research on the snowmelt runoff process are important in monitoring and foresting flood and drought disaster as well as managing water resources.In recent years the studies on distributed snowmelt runoff model are based on the application and local improvement of foreign mature models such as the SRM,have more limitations in their applicabilities.A Distributed Snowmelt Runoff Model based on RS and GIS has been designed and explored in this paper.The model consists of three parts,grid snowmelt model,grid snowmelt flow generation model and grid snowmelt flow concentration model.It was based on energy balance and water quantity balance.The grid snowmelt model was based on the energy balance.And the grid flow model was based on the water volume balance as well as the distributed grid runoff model,thus has formed a completely,distributed snowmelt runoff model which could apply to the actual service.Snowmelt and runoff process were all based on grid scale,all of the input/output data were distributed processed,the primary format were TIFF,so the model was provided with distributing character and had strict physical mechanism.An important concept of 'Unit Period of Time' was put forward in the paper,and thereby developed a 'Degree and Cent Snowmelt Runoff Model'from the traditional 'Degree-Day Method' which has widespread application nowadays,for it was not fit for the practical applications because of its glancing time-resolution,in the snowmelt period,the processes were usually very unabiding,so the time-step in day is not appropriate,but with the 'Degree and Cent Snowmelt Runoff Model',the computing time-stride of the snowmelt flood precaution decision support system could adjust itself,and to make use in the real-time computation and the forecast.It also brought forward another important concept of 'Freezing and Melting Coefficient',which was aimed to explain the snow freezing and melting repetitive physical mechanism and this concept is significant to exactly hold the snowmelt process,because the water melting of snow/ice usually refreeze in the afternoon.Besides,a Distributed Snowmelt Runoff Stimulating System by GIS has been developed,which is the platform and support of the running of the Distributed Snowmelt Runoff Model,has all of the basic functions of GIS,such as spatial analysis,hydrological analysis,dynamic simulation,and so on.Both the model and the system are the kernel modules of Snowmelt Flood Prewarning Decision Support System.At last,based on a great deal of remote sensing data,obtained distributed input data for the model,for example,distilled snow cover information with Unmixing Pixel Method,and land surface temperature(LST) with Split-window Arithmetic designed by Zhihao Qin,Kebiao Mao and so on,with our proper correction.Basic geo information data like DEM and its relevant data such as shope and asect,field simultaneous observation data mainly including weather data and hydrology data,were disposed with uniform spatial resolution(30m×30m in this paper) and the same data format(TIFF),stimulated the runoff process in representative snow melt period in 2006 of the typical study area,Juntanghu river basin,a close watershed,which located in the middle of the northern Tianshan Mountain,Xinjiang,China,the area of it is 833.57km2,with the Distributed Snowmelt Runoff Stimulating System,the time-step was 15 minutes.An average relative error of this model with the synchronous true measuring data was under 0.18,indicating that the model was able to meet the need of snowmelt flood alarming and foresting.