Talking Points

1 Title slide with my name, the Hydrometeorological Prediction Center (HPC) and Environmental Modeling Center (EMC) website addresses, and the HPC Medium Range Desk telephone number.

2 The role of the HPC in the National Weather Service and for HPC users.

The HPC Medium Range Desk is co-located with short range HPC desks and the NCEP Environmental Modeling Center's (EMC) computer modelers. HPC conducts daily model briefings for EMC modelers and other NCEP staff.

3 The HPC Medium Range Desk is staffed by 2 day-shift forecasters everyday.

(The phone number is answered by short-range HPC staff during evening hours)

Forecaster #1 produces day 3-7 fronts and pressure forecasts by 2 pm ET and the

extended forecast discussion by 2:30 pm ET.

HPC also acts as the back-up National Hurricane Center (NHC) and

provides model interpretation and tropical system track guidance for days

4-7 via a daily noon ET hotline conference call.)

Forecaster #2 produces a day 1-7 Hawaiian sector model discussion by 1230UTC,

the day 4 & 5 portion of a HPC 5-day CONUS QPF by 1730 UTC, and a CONUS

day 3-7 nighttime minimum/daytime maximum and probability of precipitation

(PoP) forecasts by 2 pm ET.

HPC is also co-located & coordinates with the Marine Prediction Center (MPC)

4 The HPC website Medium Range Forecast Page with varied links

5/6 Outline of the HPC 12 Step medium range forecast methodology including:

- operational model characteristics, comparison, and interpretation

- mean height chart analysis and interpretation (S. Flood's presentation)

- ensemble forecasting

- importance of verification

Do you use a similar forecast methodology at your office or is the standard less structured

& highly forecaster dependent? Do you routinely use HPC guidance?

7 Comparison of the most used medium range numerical models

- the ECMWF is the highest resolution model but only runs at 12 UTC

8 Historical perspective: A Day 5 500 mb height forecast today has a much skill as a 36 hour forecast back in 1972! The improvement is due mostly to an increase in computer power/model resolution. Subsequent skill advancements will be tied more directly to the assimilation of better observed data into the models.

9 Comparison & implications of model monthly skill scores over the last 5 years

10 How far into the future do the models provide decent skill?

ECMWF 500 mb height average forecast skill valid day 6 is equal to the day 5.5

MRF or UKMET and only the day 5 NOGAPS. However..the 12 hour skill

advantage of the ECMWF over the MRF and UKMET is balanced during the

dayshift as available ECMWF guidance is 12 hours older (from 12 UTC the prior

day) compared to 00 UTC MRF or UKMET guidance from 00 UTC. Extended

guidance out to 126 hours from the 12 UTC AVN has the most skill of all the

models into the medium range forecast period during the dayshift.

- Model skill is highly dependent upon the weather pattern regime

11 Any model can provide the most skill forecast on a given day

12 What are the major causes of model error?

13 Comparison of the ensemble forecast models:

- the ECMWF offers the best guidance but is not commonly available.

- Web addresses for available ensemble models

14 Ensemble forecasting methodology and products

15 NCEP ensemble skill compared to the operational MRF

16-19 Examples of ensemble guidance available to HPC forecasters in real-time

20 What are Model Output Statistics (MOS)?

When can MOS be considered a good first guess for medium range forecasts?

21-23 Examples of MOS PoP biases and temperature errors for 1998-99

24-26 Preliminary skill comparison of the new 2000 MRF MOS vs. the older version.

- What may future months show & what is the potential weakness for 2000 MOS?

- What data was used to help derive the new MOS 2000 regression equations?

27 Disclaimer Public Service Announcement!

28 Apply HPC medium range forecasting methodology and theory to a January 12, 1999 case study for slides 26-38

- Are the models initialized correctly?

29/30 Compare varied model 500 mb guidance. The ECMWF and NOGAPS offer a much more amplified solution over the east-central U.S. than the MRF and UKMET. All models point to a major precipitation event for the NW U.S.

31 Compare varied model PMSL guidance

- Will there be major U.S. winter storms?

32 Blended model guidance is available to HPC forecasters in real-time from the MRF, AVN, ECMWF, UKMET, NOGAPS, and MRF ensembles.

33 How good is model run-to-run continuity?

Is good model continuity important?

34 NCEP ensemble "spaghetti" plot guidance shows the potential for more east- central U.S. trof digging than indicated by the MRF/UKMET. Does this mean we go with the more amplified ECMWF/NOGAPS solution today?

35 Even the MRF's own 250 mb forecast mass fields support a "diggy" pattern

36/37 Compare 00 UTC medium range model guidance to updated 12 UTC model guidance as available during the shift. How do the 12 UTC models influence your medium range forecast with respect to skill and product deadlines?

38 How well did the models verify?

39/40 What adjustments should be made to MOS and MRF QPF guidance. Why?

41/42 HPC Verification Scores:

- Historical mean yearly absolute temperature errors 1972 - 1999

- temperature errors are smaller during the warm season

- HPC guidance has more skill than MOS for temperatures and PoPs

43 Conclusions and the last of the cartoons!