Ensemble Sensitivity Analysis Into Operations: Approach and Forecaster Feedback

Meteorological forecasters constantly seek innovative approaches to improve weather predictions and provide timely, accurate information for decision-making. Our previous work has developed an ensemble clustering approach, which has been proven effective at distilling an ensemble forecast down to its most prevalent forecast scenarios. The clustering is now available to all National Weather Service (NWS) forecasters through the Dynamic-based Ensemble Scenarios for IDSS (DESI) software. Still, forecasters desire more information on what early features in the flow lead to these divergent forecast outcomes. Ensemble sensitivity analysis (ESA) provides this by quantifying the influences of early forecast conditions on later forecast events, providing valuable insights into the origins of the forecast uncertainty. In this context, we are developing an ESA framework that will hopefully become operational.

This presentation will highlight our latest R2O efforts with the ESA tool. This includes a new operational webpage at the Weather Prediction Center (WPC) that shows ESA results for three domains across CONUS (East, Central, and West) and Alaska. The approach uses 100 ensemble members from the GEFS, Canadian, and ECMWF at 0000 and 1200 UTC. The initial effort has focused on the mass field (500 hPa heights and sea-level pressure) out to day 9, but the approach has been recently expanded to include 6-h and 24-h precipitation from these ensembles. The final goal of this project is to apply the approach to mesoscale precipitation with time-lagged high resolution ensembles to better understand uncertainty with various high impact weather phenomena (snowbands, organized convection, lake-effect snow, etc.).

Two case studies of winter storms will be used to show the latest capability of the ESA tool in the context of enhancing situational awareness of various clusters (scenarios). Forecaster feedback is critical to the development of this tool and training. We will highlight some of the comments received by NWS forecasters using our ESA webpage in operations.