Rule-based Decision Aid for the NWS WFO (poster)
Forecaster Event Awareness Tool (FEAT)
Paula McCaslin, Thomas LeFebvre, NOAA OAR ESRL Global Systems Division (GSD)
Introduction
Based on NWS forecaster requests, GSD developed a tool to pre-screen gridded forecasts in AWIPS II for interesting, impactful weather events. The Forecaster Event Awareness Tool (FEAT) looks for patterns in large batches of data that might normally go unnoticed. FEAT has many potential benefits for the operational forecaster.
How it works
Weather events described in equations create FEAT evaluation criteria. FEAT then systematically searches multiple forecast models for any events that match or exceed that preset criteria. Events are displayed in a color-coded a grid box for each forecast time,giving a visual summary of the events found and level of significance.
Figure: Forecast Events Awareness Tool (FEAT) application.
At a glance, the red grid boxes (right panel)
get your attention to a possible significant weather event.
What Can FEAT Do
FEAT can draw a forecaster’s attention to something that might otherwise go unnoticed. WFO’s can set custom criteria for local conditions, such as thresholds for Red Flag warnings or reduced visibility. FEAT notifies the forecaster about the predicted potential impact and its timing. FEAT alerts can relieve forecasters of some routine monitoring work, leaving them more time/attention for more critical decisions, and IDSS.
FEAT under the hood
Using FEAT within the AWIPS GFE, forecasters select an area of interest and define the alert criteria for a potential weather event.
Event Rules Definition and End Results
FEAT pre-screens for information outside of CAVE using the data application framework (DAF). FEAT events with a distinct color show what may be impactful.
(B)
- FEAT event rules are defined in a configuration file. Weather event rules, for example, can be:
- Dense Fog: Visibility <= .25
- High Wind: Wind >= 40.0 | Wind Gust >= 58.0
- While the event rule can remain constant, a change in the selected area (an airport, values above 10K feet, the CWA, etc.) will impact the percentage of grids that meet criteria for the event and thus the event significance.
- Also, when multiple model forecasts are in good agreement on a potential weather event, a higher weight is applied to the results.
Thus, FEAT can evaluate a weather event as: a) Percentage of grid points where criteria is met vs. total grid points, or b) Percentage of models in agreement
Figures Top: 0 -24%, or not an event for 100% of models;
Middle: 25-49% of grid points;
Lower: 50-75% of grid points.
Future Ideas
"How can we find interesting events over time in the midst of massive amounts of data to explore?" FEAT! The following are a few ideas from an enthusiastic forecaster:
- Look for gradients — Forecasters know where the weather is changing is most important. They suggest FEAT look for places with unusually high gradients in temperature, humidity or wind.
- Gravity Waves — Gravity waves are another gradient example. Use FEAT to scan data for gravity waves, a dangerous hard to forecast event (images).
- QPF over a standard time period — Models have different time resolutions, so need a clever way to sum up the QPF over a 24-hour time period and create FEAT event records based on the sum.
- Differences between model forecasts and the current forecast — This involves comparing the forecast elements (T, Td, Wind, etc.) to each model forecast and generating events based on the magnitude of the difference.
- Chance for freezing precip — This involves comparing the maximum wet bulb temperature forecast (something that's already being calculated) against the surface temperature forecast. Certain thresholds will generate various FEAT events.
Learning more
Here we see the moderate High Wind (event) related to the Dense Fog clearing out.
FEAT is a monitoring tool and might reveal new insights to help forecasters. Think possibilities.