Impacts of data assimilation of simulated rain rates from the NASA HIRAD instrument on tropical cyclone precipitation forecasts in a mesoscale model

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Monday, 18 January 2010
Exhibit Hall B2 (GWCC)
Cerese M. Inglish, Florida State University, Tallahassee, FL; and T. N. Krishnamurti and T. L. Miller

Handout (372.9 kB)

Increasing the predictive capabilities of mesoscale modeling of tropical cyclone precipitation forecasting is crucial for a variety of hydrological, agricultural and urban interests that depend on accurate forewarning of precipitation. It is for this reason that validation of special instrumentation that accurately observes the heaviest rain rate areas in hurricanes is crucial for properly initializing mesoscale models. Through improvements in data assimilation techniques, it has been shown that improved spatial, temporal and quantitative estimates of precipitation used to initialize a mesoscale model can more accurately predict a tropical cyclone's intensity and precipitation patterns.

One instrument that will be advantageous to the improvement of precipitation forecasting is the National Aeronautics and Space Administration (NASA) Hurricane Imaging Radiometer (HIRAD). HIRAD is a multi-frequency interferometric radiometer currently in development that provides wide-swath high-resolution measurements of surface winds and precipitation in a hurricane. It will provide improved views of high wind gradients and heavy precipitation regions that were previously impaired with other instruments, such as the Step Frequency Microwave Radiometer (SFMR) aboard the typical hurricane hunters. This study validates the usefulness of HIRAD's capabilities by utilizing simulated high-resolution HIRAD rain rate observations in Hurricane Frances (2004) to initialize the Weather Research and Forecasting (WRF) Advanced Research WRF (ARW) mesoscale model. It reveals key impacts that data assimilation of rain rates into mesoscale model moisture initialization schemes have on forecasted tropical cyclone rain rates, spatial structure of precipitation, and cyclone intensity.

The study follows the basic design structure of an Observing System Simulation Experiment (OSSE) and employs the use of a high-resolution 10-day Fifth Generation Mesoscale Model (MM5) forecast to act as a Nature Run. The WRF-ARW model serves as its fraternal twin model, in which the data assimilation and forecasting occurs. This model has a slightly different resolution and physics compared to the nature run model, but operates in a concurrent storm time frame and obtains much of its initialization information from the nature model. Simulated HIRAD rain rate data is collected from the Nature Run and used in a rain rate initialization scheme developed at Florida State University. This initialization includes the radiometric rain in the model's initial state. The process occurs for 24 hours prior to the forecast start time during which the assimilated data and appropriate errors on the order of the instrument's observing capabilities are given to the model. A control experiment with no aircraft observations is also performed for comparative analysis. Results for all of the experiments conducted are assessed against the nature run in terms of standard error metrics such as Root Mean Square Error (RMSE), Anomaly Correlation and Equitable Threat Scores (ETS). This will elucidate the spatial and quantitative benefits on precipitation forecasting by assimilating observed HIRAD rain rates from tropical cyclones into mesoscale models.