Lab Exercise, COMET Case Study 013

"Southern California Floods and Florida Tornadoes"
23 February 1998

ETA LAB
Jon Mittelstadt and Michael Staudenmaier

This case focuses on the storm system that affected Southern California on 23 February 1998. The data used for this were presented as a laboratory exercise during the COMAP NWP Symposium held at COMET in September 1998.

Goals:

1. Research the performance of three versions of the Eta model on one particular case in Southern California.

2. Investigate potential model biases.

3. Examine the need for high-resolution modeling in complex terrain.

4. Discuss how much time to spend on verifying model initializations.

5. Discuss the benefits/problems with knowing model design and gain knowledge of the Eta model design and how it may impact biases.

Schedule:

I. Before lab: Fill out bias worksheet	(15 minutes)
II. Intro and scenario briefing	(10 minutes)
III. Fill in tables and answer questions	(30 minutes)
IV. Break	(10 minutes)
V. Answer discussion questions	(15 minutes)
VI. Share results and discuss	(40 minutes)

List in the left-hand column any biases, strengths, and/or weaknesses of the Eta model that are used at your office, or that you yourself have seen. In the right-hand column list a basis for these biases, for example, local-office study, NCEP statistics, forecaster experience, etc. Then fill in the Eta model version(s) that apply. At the bottom of the page, list your geographical region.

BIAS/STRENGTH/ WEAKNESSES	BASIS	What version(s) of Eta model?

GEOGRAPHICAL REGION: _______________________________________

Background:

Eta Model Background for the time of this event, 23 February 1998.

NCEP was running three versions of the Eta model. The Eta-32 (32-km resolution "Early Eta") ran at 0000 UTC and at 1200 UTC, available on an 80-km display grid. The Eta-29 ("Meso-Eta") ran at 0300 UTC and 1500 UTC, available on a 40-km display grid. The Eta-10 ("even more Meso-Eta"?) ran over a western United States domain at 0300 UTC and was available on a 10-km display grid.
The Eta-32 had (and still has) 45 vertical levels, the Eta-29 had 50, and the Eta-10 had 60 vertical layers.
The Eta-32 had (and still has) a 12-hour 3DVAR EDAS "spin-up" period (involving four 3-hour forecasts). The Eta-29 used a 3-hour period. The Eta-10 did not have a spin-up period; its initial analysis started from the Eta-29 0300 UTC analysis interpolated to the 10-km grid and then the new 3DVAR system was used to assimilate several additional data sources.
Boundary conditions for the Eta-32 and Eta-29 came from the AVN model. Boundary conditions for the Eta-10 model came from the Eta-29.
The model physics of the three versions were essentially the same. However, one exception was that the Betts-Miller-Janic convective scheme in the Eta-10 used different thresholds designed to produce more precipitation over areas of higher terrain.

Group Assignments:

Group #	Parameter to investigate
1.	Min and max surface temperatures over 6-hour periods. Use 2 meter temps.
2.	Maximum wind speed and direction at surface over 6-hour periods. You will have to estimate these values. Use 2-meter winds.
3.	6-hour total QPF.
4.	6-hour convective QPF.
5.	24-hour total and convective QPF.
6.	CAPE over 6-hour periods.
7.	500 and 850 mb RH at 1200 UTC 23 Feb 98 and at 0000 UTC 24 February 98 for Vandenburg AFB.
8.	500 and 850 mb wind speed and direction at 1200 UTC 23 Feb 98 and at 0000 UTC 24 February 98 for Vandenburg AFB.
9.	Spin-up problems with any parameter.

Procedure:

A. Fill in the table on the last page with the model forecast values and observed values (if any) for your research parameter. To get started, use GARP to view the data from 23feb98.

B. Examine the model forecast and provide brief answers for the following questions. Examine any model field and observation that you desire.

C. Create a few gif images using GARP and prepare a 3-minute presentation of your results.

Questions:

1. How well did the Eta models initialize your research parameter?

2. What advantages and/or disadvantages were provided by the 40-km display grid compared to the 80-km grid?

3. What advantages and/or disadvantages were provided by the 10-km Eta model?

4. What role did your parameter play in the model’s onset and intensity of QPF? (groups 3-5 answer this question - What were the main factors determining where the model QPF you investigated was located?)

Please write your group number here ____________

Locations	Elevation	Lat, Lon	Description
MWS (Mt. Wilson, CA)	4800 feet	34.2,-118.1	Mountain
87Q (Cambria, CA)	100 feet	35.7,-121.2	Coastal
WJF (Lancaster, CA)	2400 feet	34.7,-118.2	Valley

Table for Groups # 1-4 and 6

Forecast @ Location & Period	Eta-32/80 model	Eta-29/40 model	Eta-10/10 model	Reality (if known)
87Q 12-18z 23feb98
WJF 12-18z 23feb98
MWS 12-18z 23feb98
87Q 18z/23- 00z/24feb
WJF 18z/23-00z/24feb
MWS 18z/23-00z/24feb

Table for Group #5

Forecast @ Location & Period	Eta-32/80 model	Eta-29/40 model	Eta-10/10 (03z-03z)	reality (if known)
WJF 00z/23-00z/24feb
MWS 00z/23-00z/24feb
87Q 00z/23-00z/24feb

Table for Group #7 and #8

Forecast @ Location & Period	Eta-32/80 model	Eta-29/40 model	Eta-10/10 model	reality (if known)
Vandenburg AFB 12z 23feb98
Vandenburg AFB 00z 24feb98

Table for Group #9

Parameter	Eta-32/80 model	Eta-29/40 model	Eta-10/10 model	reality (if known)

Discussion Questions:

1. What should be done at the local office to prove/disprove biases? At NCEP? At universities?

2. What do you do when models disagree and you can’t find a basis for rejecting one?

3. How can a SOO keep office knowledge up to date when models change rapidly?

4. How much time should be spent looking at initializations? On verifying 3-hr and/or 6-hr forecasts?

5. What other improvements are needed in the data the NWS receives from NCEP?

Case Study 013