Variational Continuous Assimilation of TMI and SSM/I Rain Rates: Impact on GEOS-3 Hurricane Analyses and Forecasts

Abstract

This study describes a 1D variational continuous assimilation (VCA) algorithm for assimilating tropical rainfall data using moisture/temperature time-tendency corrections as the control variable to offset model deficiencies. For rainfall assimilation, model errors are of special concern since model-predicted precipitation is based on parameterized moist physics, which can have substantial systematic errors. The authors examine whether a VCA scheme using the forecast model as a weak constraint offers an effective pathway to precipitation assimilation. The particular scheme investigated employs a precipitation observation operator based on a 6-h integration of a column model of moist physics from the Goddard Earth Observing System (GEOS) global data assimilation system (DAS). In earlier studies, a simplified version of this scheme was tested, and improved monthly mean analyses and better short-range forecast skills were obtained. This paper describes the full implementation of the 1DVCA scheme using background and observation error statistics and examines its impact on GEOS analyses and forecasts of prominent tropical weather systems such as hurricanes. Assimilation experiments with and without rainfall data for Hurricanes Bonnie and Floyd show that assimilating 6-h Tropical Rainfall Measuring Mission (TRMM) Microwave Imager (TMI) and Special Sensor Microwave Imager (SSM/I) surface rain accumulations leads to more realistic analyzed storm features and better 5-day storm track prediction and precipitation forecasts. These results demonstrate the importance of addressing model deficiencies in moisture time tendency in order to make effective use of precipitation information in data assimilation.