Improving Forecasts During Heavy Precipitation Events: Model Biases and Numerical Experiments

North Carolina State University, Raleigh NC

Investigators

Abstract

Forecasting of heavy precipitation events remains one of the most difficult challenges in numerical weather prediction. The ability of numerical models to represent precipitation is complicated by the necessity to parameterize some precipitation while some is explicitly resolved. Furthermore, errors in precipitation forecasts have the potential to degrade model forecasts through latent heat release and other diabatic processes, especially during heavy precipitation. It is important to improve understanding of how errors in model precipitation forecasts might influence synoptic-scale forecasts. The problem of quantitative precipitation forecasting (QPF), in addition to significant socioeconomic relevance, holds important implications for the problem of atmospheric predictability. Numerical model QPF errors can promote forecast degradation via the influence of diabatic processes such as latent heat release on atmospheric dynamics and thermodynamics. A previous study by the Principal Investigator has documented errors in numerical forecasts during heavy precipitation, and demonstrated that these errors are consistent with model misrepresentation of latent heat release in the vicinity of a convective, cold-frontal rain band. The Principal Investigator proposes to accomplish the following specific objectives: 1) Development of a climatology of operational model forecast biases for specific synoptic scenarios in which model representation of convection may limit forecast accuracy; 2) Documentation of the physical basis for systematic biases in operational model behavior that relate to the representation of heavy precipitation; 3) Examination of the degree to which model representation of lower-tropospheric, diabatically generated potential vorticity (PV) anomalies are sensitive to model representation of precipitation processes (both grid-scale and sub-grid-scale); 4) Determination of which convective parameterization schemes yield the most realistic representation of diabatic PV modifications in both the lower and upper troposphere; 5) Exploration of modifications to model representations of precipitation that would improve forecasts of the dynamical feedbacks associated with latent heat release. Successful completion of this research could lead to better utilization of current numerical models by forecasters as well as lead to quantitative improvements in numerical models precipitation forecasts.