论文部分内容阅读
Long term precipitation with high spatio-temporal and precision can improve our understanding on the basin-scale hydrology,thereby increasing the ability of land surface hydrological processes and the state of the simulation.Traditional ground-based observations,remote sensing,and regional climate modeling,however,difficult to provide high spatio-temporal precipitation data over complex terrain,especially in basin scale.Data assimilation techniques such as 4-D Var and PODEn3DVar are proposed to be used in regional climate model to assimilate the ground observation data,active microwave precipitation products and dual-polar radar data into a weather research and forecasting (WRF) model to achieve dynamical downscaling.This paper is focus on the estimation and simulation of precipitation data with high spatio-temporal for hydrological research over complex terrain based on regional climate model and data assimilation techniques,to realize the following objectives: (1) to achieve dynamical downscaling by assimilating satellite precipitation products,ground observation and radar observations in multiple nested structure of the regional climate model; (2) the development of effective multi-source observational data assimilation system in a regional climate model; (3) to establish a reasonable evaluation system for precipitation product.Dopplar Radar Doppler-radar radial velocity and reflectivity are simultaneously assimilated into WRF model by a proper orthogonal-decomposition-based ensemble,three-dimensional variational assimilation method (referred to as PODEn3DVar),which therefore forms the PODEn3DVar-based radar assimilation system (referred to as WRF-PODEn3DVar).Results from real data assimilation experiments with the WRF model show that WRF-PODEn3DVar simulation yields better rainfall forecasting than radar retrieval,and radar retrieval is better than the standard WRF-3DVar assimilation,probably because of the flow-dependence character embedded in the WRF-PODEn3DVar.TRMM precipitation data are assimilated into the outer domain in WRF model by 4-D Var method.Results from real data assimilation experiments with WRF model indicate that TRMM precipitation assimilation with 4-D Var simulation yields better rainfall forecasting in inner domain than that simulated by WRF model without remote sensing assimilation.In addition to above data assimilation into WRF model,the next step,we will try to assimilate FY-series and GPM precipitation data into WRF model separately.After finishing all data assimilation one by one,we will try to assimilate the combining of all these data into WRF model to improve the precision of precipitation simulation.