The Direct Assimilation Technology And PreliminaryApplication Of FY-2F Imager Data

Direct assimilation of satellite radiance is one of the most heated research subject s nowadays. In this paper, with the Hurricane Weather Research and Forecasting (HWRF) model, FengYun-2F(FY-2F) imager data in infrared channels are assimilated directly to improve the forecast skills of typhoon in West Pacific. Since few studies are involved with geostationary satellite data assimilation, this paper focuses on the characters of FY-2F sensor and its data.Through the analysis of weighting functions, the infrared channel3(IR3) is verified to be a high-level channel and the other three channels are surface layer channel (IR1,2,4). The Community radiative transfer model (CRTM) is used to provide the simulated brightness temperature of FY-2F based on the output of HWRF and the simulation is compared with observation. It is found that, there is significant bias in the cloudy area, especially in the middle and low latitude where huge difference exists between simulation and observation for the monsoon cluster and cloud of tropical cyclone. While under the clear-sky condition, the observation of low-level channels are influenced by surface emission heavily. Because of the uncertainty of emission of land, the error of simulation over ocean is smaller than that over land. Meanwhile, there are big bias of the simulation in IR4at both daytime and nighttime. Therefore, a compounded cloud detection algorithm is proposed to remove all cloud-contaminated pixels and observations over land in IR1and IR2, all observations in IR4are also excluded. In addition, the data format is conversed to feed the simulation system.The interface program of FY-2F is added in Community Gridpoint Statistical Interpolation(GSI) including data reading and quality control so that the FY-2F observation can be assimilated directly. The deference between observation and simulated brightness temperature of background(0-B) are compared before and after bias correction and so are the observation and simulation of analysis field after assimilation(0-A), which is adjusted from the initiated condition after assimilation.It is shown that, the averaged values of0-B are all positive and decreased after bias correction. The decrease of0-B in IR3is the most remarkable, which is more than3.5K in average before correction and less than0.34K after correction. Meanwhile, the distribution of bias in each channel tends to be Gaussian distribution. After assimilation, the averaged values of0-A are less than0-B in general. That means the analyzed field are closer to observation. The progress of direct assimilation of satellite radiance mainly changes the humility and temperature. The varied range of specific humidity is less than lg/kg and the change of temperature is no more than0.6K in850hPa.To check the influence of direct assimilation of FY-2F radiance observation on typhoon track and intensity forecast, three experiments are designed:The control experiment is one without any assimilation of data; satellite experiment one(Sat_expl) is one in which only AMSU-A and HIRS data are assimilated; in satellite experiment two(Sat_exp2), three infrared channels of FY-2F observations are assimilated directly in the foundation of Sat_expl. Three typhoons are selected to be study objects. Each typhoon is simulated several times in6h cycle-run. And the initial conditions are modified by three-dimensional variational data assimilation system of GSI in Sat_expl and Sta_exp2. The result shows that, after direct assimilation of FY-2F imager data, the ratio of cases in which the track error decreases in24hour,48hour and72hour simulation are26/50,22/48,22/41respectively. And the ratio of cases with decreased simulated-intensity error are19/50,18/48and19/41.