THE IMPACT OF ASSIMILATING SATELLITE-DERIVED PRECIPITATION RATES ON NUMERICAL SIMULATIONS OF THE ERICA IOP-4 CYCLONE

The present study uses a regional-scale numerical model to test the im pact of dynamically assimilating satellite-derived precipitation rates on the numerical simulations of one of the deepest extratropical cycl ones to develop south of 40-degrees-N in this century. This cyclone ev ent occurred during the Experiment on Rapidly Intensifying Cyclones ov er the Atlantic (ERICA) intensive observing period 4 and has been sele cted because of the strength of the cyclone and the availability of th e special ERICA data in addition to the Special Sensor Microwave/Image r (SSM/I) and Geostationary Operational Environmental Satellite (GOES) infrared (IR) satellite data. The unique methodology developed herein to synthesize the SSM/1 and GOES IR satellite data produces precipita tion estimates that have realistic spatial and temporal structure. The assimilation of satellite-derived precipitation is accomplished by sc aling the internally generated model profiles of total latent heating. At points where the model is not producing precipitation, the vertica l distribution of total latent heating given by satellite precipitatio n is specified from instantaneous model-based profiles at adjacent poi nts using a search algorithm. The technique does not assume a priori t hat the satellite-estimated precipitation corresponds to either convec tive or stratiform model precipitation, and uses heating profiles that am consistent with the model's parameterization of either type of pre cipitation since they are not specified from externally based paraboli c or other structure functions. Several simulations are performed with and without satellite data assimilation at varying horizontal and ver tical model resolutions. The results from the 80-km 40-layer control a nd assimilation runs demonstrate that the assimilation of satellite pr ecipitation 1) does not introduce noise into the simulations at any ti me during or after the data assimilation period, 2) forces the model t o reproduce the magnitude and distribution of satellite precipitation, and 3) improves the simulated central mean sea level pressure (MSLP) minima slightly, frontal positions, and, to a greater extent, the low- level vertical-motion patterns when compared with subjective analyses and satellite imagery. The model retains the information introduced by the assimilation of satellite-derived precipitation 8.5 h after the e nd of the data assimilation period. An increase in the vertical and ho rizontal model resolution further reduces the errors in simulating the MSLP minima but does not consistently improve the cyclone position er rors in the assimilation runs. Either the exclusion of the search algo rithm, the doubling of satellite precipitation, or an eastward shift o f satellite precipitation by 400 km increases the MSLP and position er rors; therefore, the impact of assimilating satellite precipitation de pends on model resolution, the use of the search algorithm, and the ma gnitude and position of satellite precipitation. The increase in horiz ontal resolution generates the largest reduction in MSLP errors, while the shifting of satellite precipitation generates the largest increas e in MSLP errors. The results confirm the findings of earlier studies that the impact of assimilating satellite precipitation on the subsequ ent simulations is less sensitive to errors in magnitude rather than t o the distribution of satellite-derived precipitation and depends on t he relative accuracy with which the model simulates the cyclone in the control run. Despite the fact that this study focuses on a single cas e, it does demonstrate the promise of using combined infrared and micr owave satellite precipitation estimates to produce sustained positive impacts in mesoscale model forecasts of midlatitude cyclogenesis over data-sparse oceanic regions.