Assimilation of All-sky Infrared Brightness Temperatures and Atmospheric Motion Vectors in Tropical Cyclone Forecasting

Potential impacts of assimilating satellite infrared brightness temperatures, together with atmospheric motion vectors retrieved from them, have been assessed through a series of convection-permitting observing system simulation experiments (OSSEs) for Hurricane Karl (2010). We investigated the potential impacts of assimilating data from the Advanced Himawari Imager (AHI) on the Himawari-8 Japanese Geostationary Meteorological Satellite (Himawari-8) which launched in October 2014 and the Advanced Baseline Imager (ABI) on the Geostationary Operational Environmental Satellite (GOES-R) which will be launched in 2016. Our focus on the observations from these two satellites was a result both of their hemispherical coverage, including the tropical oceans where there is generally a dearth of observations but where tropical cyclone genesis and development occur, and of their high temporal and spatial resolution data. Direct assimilation of satellite infrared brightness temperatures, especially from cloudy regions, is challenging given their strong nonlinear relationships to the underlying model fields. For this study we coupled the Community Radiative Transfer Model (CRTM) to the ensemble Kalman filter (EnKF) data assimilation system developed at the Pennsylvania State University (PSU) and built around the Weather Research and Forecasting model (WRF). This new framework, together with our assimilation strategies which included “superobbing” and data quality control, enabled us to effectively assimilate infrared brightness temperatures and atmospheric motion vectors in the analysis and forecasting of Karl. In this study, we have investigated the potentials of assimilating the satellite-measured and -retrieved variables for the detailed structure of tropical cyclones within the regions where only satellites are available. We found that assimilation of infrared brightness temperatures improved model representation of Karl's primary rainbands, eye, and individual convective clouds, as well as clear-sky regions through 6-hour assimilation, and even further captured through 24-hour assimilation (Figure 1). Not only were thermodynamic variables such as temperature, moisture and hydrometeors better constrained, but analysis and forecast errors in the wind fields were reduced as well. Additional assimilation of satellite-retrieved atmospheric motion vectors led to further improvements in both the thermodynamic and dynamic fields. Overall, we found that assimilation of satellite infrared brightness temperatures and satellite-retrieved atmospheric motion vectors were complementary, indicating the potential benefit of combining the two, as well as additional measurements and retrieved products, in both the analysis and forecasting of tropical cyclones. These results are especially relevant to those regions for which other data sources are lacking but in which tropical cyclones are born and live.