Ohio Valley NWS Offices - experiment ESTF

[BeauDodson]

The following offices will participate:

Jackson (JKL), Louisville (LMK), Paducah (PAH), St. Louis (LSX), Detroit/Pontiac (DTX), Grand Rapids (GRR), Gaylord (APX), Marquette (MQT), Goodland (GLD), Dodge City (DDC), Grand Junction (GJT).

NWS Central Region offices will produce collaborated, decisive, high temporal and spatial resolution short-term (0-36 hour) forecasts and frequent updates. Forecasts will match observations and expected synoptic and mesoscale weather trends to facilitate accurate short-term impact-based decision support for a myriad of users.

Implement hybrid schedule-driven─event-driven forecast updates. Scheduled updates are issued every 3 hours around 0230, 0530, 0830, 1130, 1430, 1730, 2030, and 2330 UTC. Event-driven updates are issued more frequently as the weather dictates.

Current observations are initialized into the digital database to match reality. Meteorologists then ensure the short-term forecast is relevant, representative, and scientifically-consistent with observations, forecaster thinking, mesoscale processes, conceptual models, and expected trends. Mesoscale detail and parameter gradients are incorporated into the forecast process.

The majority of changes at each update will be in the first 12 hours of the forecast, but the 12-36 hour forecast also will be revised proactively as needed.

Eleven NWS offices are participating in a testbed to assess and determine the best process to achieve project goals while maintaining a constantly relevant, accurate database.

----

Time period: February 15, 2012 through May 31, 2012

Improved short-term forecasts to facilitate greater user decision support. Issuance times may differ from current practice, but all forecasts will represent the most relevant, latest thinking.

Area Forecast Discussions (AFDs) will be issued for each tangible forecast update.

Short-term information on the National Digital Forecast Database will be more accurate and up-to-date in testbed county warning areas.

“Point and Click” forecasts on NWS testbed office websites will contain frequently revised data and detail. The “Activity Planner” on these websites will contain detailed, hourly tabular and graphical forecast information (meteorograms) for maximum customer utility.

Other forecast services (e.g., aviation/TAFs, AFM, PFM, HWO, hazardous weather headlines, fire weather, etc.) will be consistent with the ESTF.

User feedback is critical to the evaluation and success of the ESTF testbed process. Please forward the benefits, drawbacks, perception, and utility of ESTF to your operations to the National Weather Service - please find each offices email address to address your concerns/questions/thoughts.

ESTF_Announcement.pdf

[WestMichigan]

It will be interesting to see what happens. ILN was pretty good with the short term forecasts, GRR on the other hand almost never has any updates. I am looking forward to some additional updates from them when something is happening.

[GreatLakesWxMan]

Always sounds good in bureaucratic theory...we'll see how well we can implement this in practice. Neither tools nor forecasters are designed to forecast at the spatial/temporal scale being asked...and I won't even get into the statistical uselessness of hourly PoP forecasts (i.e., think about what the climatological hourly PoP is at your location).

[MGorse]

The NWS offices in Eastern Region have been doing ESTF for a few years now.

[OceanStWx]

Always sounds good in bureaucratic theory...we'll see how well we can implement this in practice. Neither tools nor forecasters are designed to forecast at the spatial/temporal scale being asked...and I won't even get into the statistical uselessness of hourly PoP forecasts (i.e., think about what the climatological hourly PoP is at your location).

We've been doing this for roughly a year now at GYX and this is definitely my biggest issue with it. The updates themselves are fine, and in benign weather usually just consists of keeping temps/sky on track. The hourly PoP is just an issue. On that time scale, we're really talking about a binary forecast. It either is or isn't raining/snowing. Yet as you say, guidance is not that good, and uncertainty increasing pretty quickly away from hour 0. If you try and treat the forecast as binary, when weather is rapidly changing this often requires more frequent updates than every 3 hours. When mixed ptypes are involved this can lead to a significant time sink trying to keep Wx and PoP consistent and accurate. If you don't treat the forecast as binary, you're not really using hourly PoP in a true sense, so why bother

Thankfully, QPF can still be 6 hourly at the forecasters discretion.