Rainfall Forecast Validation Team

 

Team Leader:            Robert Rogers (NOAA/HRD)

 

Team Members:        Frank Marks (NOAA/HRD)

                                   

Collaborators:           Tim Marchok (NOAA/GFDL)

                                    Robert Tuleya (NOAA/NCEP/EMC/SAIC)

                                    Manuel Lonfat (RMS)

                                   

Team Objective:

 

            One of the most significant impacts of tropical cyclones is the copious amount of rainfall they often produce. Drowning from inland flooding in landfalling tropical cyclones is the leading cause of death from storms affecting the United States in the past 30 years. For this reason, the Tropical Prediction Center has stated that one of its highest priorities is to improve quantitative precipitation forecasting (QPF) for tropical cyclones. Dynamical numerical models provide one way of forecasting rainfall from tropical cyclones. While such models enable the depiction of the temporal and spatial evolution of tropical cyclones and their associated rain fields, they often exhibit errors related to inadequate initial conditions and model physics. Knowledge of these errors can aid the forecaster in interpreting numerical guidance of rainfall and adjusting their forecasts accordingly.

An accurate diagnosis of rainfall forecast errors requires a validation scheme that accurately measures the performance of the forecast system. However, no standard technique has been developed to validate rainfall forecasts from tropical cyclones. Most rainfall forecasts involving landfalling TC’s have been validated on fixed domains using standard validation metrics, such as bias and equitable threat score, that do not specifically account for the presence of the TC. As a result, many important features of the rainfall field that are related to the TC are overlooked. A QPF validation technique specific to TCs is therefore needed to better identify biases in the forecasts and lead to possible improvements in them.

Work in this team will be directed toward developing new techniques that address these issues. The validation schemes will be implemented using forecasts of landfalling tropical cyclones from the operational GFDL, GFS, and Eta models and the benchmark Rainfall CLIPER product against observed rain fields provided by the National Precipitation Validation Unit (NPVU) dataset. Another aspect of this work that is advancing is the quantification of the impact of vertical shear on the distribution of rainfall and its incorporation into a parametric model, the benchmark rainfall climatology and persistence model (R-CLIPER).

 

Accomplishments:

·      Devised a technique for validating tropical cyclone rainfall by comparing the performance of the GFDL, GFS, and Eta models in several different rainfall attributes important in TC rainfall: pattern reproduction, mean and volume, extreme rainfall amounts, and sensitivity to track error;   

·      Completed validation of these models on all U.S. landfalling tropical cyclones from 1998 to 2004;

·      Completed quantification of the impact of vertical shear on the distribution of rainfall; 

·      Incorporated this impact into a modified R-CLIPER parametric rainfall model.  

 

Milestones:

 

• Compare modified version of R-CLIPER with the other models using the same validation techniques described above;
• Compile statistics summarizing performance of all models (GFDL, GFS, Eta, R-CLIPER, shear-modified R-CLIPER) based on the four parameters mentioned above;
• Finish manuscript summarizing these comparisons, submit to Weather and Forecasting.
• Finish manuscript describing parameterization of shear impacts on rainfall fields, submit to Monthly Weather Review;
• Consider other elements quantify and include in parametric model (e.g., topography).