Research: Identifying and Understanding Displacement Biases in Numerical Forecasts of Elevated Convective Systems
This was a collaborative research project, supported by UCAR/COMET, to work with researchers and forecasters at the NOAA Hydrometeorological Prediction Center (HPC). In some cases of heavy rainfall occurring with elevated mesoscale convective systems (MCSs), such as the TL/AS type discussed here, numerical models forecast a spatial pattern of rainfall that is generally correct, but is displaced too far to the north. One example is shown below.
Our research found that operational numerical models using parameterized convection did indeed have a significant "northward bias," although preliminary results show that this bias is reduced when using models with explicitly predicted convection. Furthermore, the reasons for the bias can vary from case to case: in some events, the model prediction of the location of the surface boundary was itself displaced to the north, but in other cases convective dynamics such as cold pools were responsible for the displacement errors. This research made up much of graduate student Charles Yost's M.S. thesis, and is in the process of being prepared for submission to a peer-reviewed journal.