Local Data Analysis and the Mesoscale Model on the WFO-Advanced Workstation
Preprints, Twelfth International Conference on Interactive Information and Processing Systems for Meteorology, Oceanography, and Hydrology, January 28 - February 2, 1996, Atlanta, GA, USA, 216-219.
Paul Schultz
NOAA Forecast Systems Laboratory
Boulder Colorado
Corresponding author address: Paul Schultz, NOAA/FSL
R/E/FS1, 325 Broadway, Boulder CO 80303-3328.
E-mail paul.j.schultz@noaa.gov
Table of Contents
The NOAA Forecast Systems Laboratory (FSL) conducted two weather forecast exercises during 1995 to demonstrate the WFO-Advanced forecasting workstation. The workstation provided many data sets and display capabilities, including a local data analysis and mesoscale modeling function, the FSL Local Analysis and Prediction System (LAPS). Both the workstation and the forecast exercise are documented in several papers and presentations at this conference (see references below).
Each exercise was patterned after NWS forecast office operations, because the workstation is intended to satisfy the requirements of NWS forecasters during the AWIPS era. The first exercise was conducted during the month of August, ran five days a week during daylight hours, and drew its subjects mostly from the development-minded meteorologists at FSL. [The second exercise, which occurred after this writing, was conducted during October and November, ran around the clock, and drew its subjects from the rank and file of NWS field operations.]
The workstation provides access to local data sets, such as radars, mesonets, etc., in addition to the usual feed of centrally-provided data sets, such as satellites, model grids, RAOBs, etc. LAPS analyzes all the data on a 10-km grid covering a 600-km square centered on the forecast office (Fig.1), on 21 isobaric surfaces 50 mb apart. These analyses are used to initialize the mesoscale forecast model. For the August test, the model ran once per day, starting at midnight, out to 18 hrs (6 pm MDT). The model, a descendant of the Regional Atmospheric Modeling System (RAMS) developed at Colorado State University, runs on the same horizontal grid as the analysis component, but the vertical coordinate is sigma_z with 25 levels, so the vertical resolution is similar to the analyses. Substantial development was invested in configuring and streamlining the model to make it efficient enough to complete the 18-hr model runs in eight hours on an HP735 computer. Thus, a model run initialized with 0600 UTC data was available to the forecaster by 1500 UTC.
Figure 1. The LAPS analysis and model forecast domain. Elevation contour interval is 250 m. Solid outline indicates the Denver Warning and Forecast Office area of responsibility in the AWIPS era (not the same as the current area). This is also the area of interest for the FSL forecast exercise.
The exercise consisted of creating routine textual forecast products and weather discussions, plus severe weather watches and warnings, in NWS formats. None of these products were disseminated outside the laboratory. Forecasters worked for one week each. (The author was among the first forecasters in the exercise.) The "short-term" forecaster was concerned with the 0-12 h forecasts and severe weather warnings; the "long-term" forecaster was concerned with the period from 12 h to 48 or 60 h, depending on time of day. And a third forecaster, designated the "backup" forecaster, dealt mostly with nowcasting the convective situation in the morning hours.
Figure 1 shows the domain over which the analysis and prediction services were provided, and the area of forecast responsibility.
Table 1 shows the products generated every hour from the analyses. The specific analysis techniques are detailed in McGinley (1991), Albers (1995), Albers et al. (1995), and Birkenheuer (1991). In addition to the products listed in the table, many other fields such as moisture flux convergence and thermal advection are made available by calculating those fields "on the fly," that is, upon load request. Such fields load at the same speed as the pre-generated fields.
__________________________________________________
Table 1. Analysis and forecast products generated by LAPS.
3-D fields
temperature
height
u, v, and w wind components
specific humidity
cloud water (liquid, ice)
hydrometeor concentration (rain, snow)
radar reflectivity
2-D or surface fields
temperature wind
MSL pressure
1500-m pressure
dew point
relative humidity
potential temperature
equivalent potential
temperature
lifted index
positive buoyant energy
negative buoyant energy
cloud base height
cloud top height
helicity
precipitable water
1-hr precip accumulation
storm total precip accumulation
1-hr snow accumulation
storm total snow accumulation
radar reflectivity
__________________________________________________
The analyses from LAPS have been generated in real time and transmitted to the Denver NWS forecast office to support operations there since 1991.
The grid configuration and other aspects of the mesoscale model are listed in Table 2. This setup is very similar to that used for FSL's daily model runs, which have been generated since late 1993 (Snook et al. 1995). There are two major exceptions. First, the model is integrated out to 18 h, instead of the 12-h runs described by Snook et al. Second, the daily FSL model runs used a primitive moisture representation (condensation and evaporation only; no precipitation) because the original RAMS microphysics algorithm slowed the model unacceptably. For the workstation's version of the model, an efficient microphysics algorithm (Schultz 1995) was implemented which computes cloud and precipitation processes with about 5\% overhead in compute time, compared to 30-40\% typical of most similar models, including RAMS.
__________________________________________________
Table 2. Mesoscale model grid and methods.
dimensions 61 x 61 x 25
horizontal increment 10 km
vertical coordinate terrain-following sigma-z
vertical increment 300 m near sfc, stretched to 750 m aloft
time step 30 s
depth of domain 15.3 km
top boundary Rayleigh friction
vertical momentum nonhydrostatic
moisture physics cloud liquid and ice, rain, snow, graupel
cumulus param none
lateral boundaries Davies relaxation to NGM forecast
surface 11-layer soil model
turbulence Smagorinsky deformation K
radiation shortwave and longwave modified by vapor and cloud
__________________________________________________
The model was initialized with LAPS-analyzed fields from 0600 UTC, or midnight Mountain Daylight Time. The lateral boundaries were provided by the NGM grids, specifically, the 0000 UTC model run's 12-, 18-, and 24-h forecasts. With a domain this small and a forecast this long, the solution is obviously quite dependent on the quality of the NGM solution. For example, a parcel traveling at 25 ms-1 would traverse the domain in less than 7 h.
In the future, the model will use boundary conditions from the Eta model and/or RUC model, as provided by NMC.
On the workstation, the same fields produced by the analysis component are also produced by the mesoscale model (Table 1), to the extent possible.
The intent behind the August portion of the exercise operations was to identify system malfunctions and design flaws. Datasets were not archived, and forecasts were not scored. The WFO-Advanced evaluation team (Roberts et al.) carefully recorded and compiled extensive logs of comments and opinions about the workstation and each of its components, including LAPS. Generally,
· The model was considered useful for forecasting the location and timing of the day's first convection, but the details of propagation and location of later development were not usually correct.
· The model was quite robust: it did not "crash" during the exercise.
· The workstation computer system was not always able to deliver the model output reliably or early enough to allow maximum utility of its services. However, the hourly analyses were quite reliable, although sometimes late. Under ideal conditions, analyses using data on the hour (e.g., SAOs) are available about 30 min after the hour, but they were sometimes as much as an hour late. This had some negative impact on the perceived utility of the analyses.
· The workstation made it very convenient to compare model forecasts to observed fields, which allows the forecaster to monitor the quality of the prognoses. For example, it is very simple to overlay surface wind forecasts with subsequent analyses. Also, the precipitation fields produced by the model are easy to compare to radar reflectivity fields.
The October/November exercise was designed to be much more quantitative in nature. Model performance statistics on forecasts of precipitation, winds, and temperatures will be presented at the conference.
Albers, S. A., 1995: The LAPS wind analysis. Wea. Forecasting, 10, 342-352.
Albers S. A., J. A. McGinley, D. L. Birkenheuer, and J. R. Smart 1995: The Local Analysis and Prediction System (LAPS): Analyses of clouds, precipitation, and temperature. Wea. Forecasting, in review.
Birkenheuer, D., 1991: An algorithm for operational water vapor analyses integrating GOES and dual-channel microwave radiometer data on the local scale. J. Appl. Meteor., 30, 834-843.
McGinley, J. A., S. C. Albers, and P. A. Stamus, 1991: Validation of a composite convective index as defined by a real-time local analysis system. Wea. Forecasting, 6, 337-356.
Schultz, P. J., 1995: An explicit cloud physics parameterization for operational numerical weather prediction. Mon.Wea.Rev., 123, in press (November).
Snook, J. S., J. M. Cram, and J. M. Schmidt, 1995: LAPS/RAMS. A nonhydrostatic mesoscale numerical modeling system configured for operational use. Tellus, 47A, in press (October).
(All below, Preprints, Twelfth International Conference on Interactive Information and Processing Systems for Meteorology, Oceanography, and Hydrology, Atlanta, Amer. Meteor. Soc.)
Davis, D. L., and S. M. Williams: The development of the WFO-Advanced hydrometeorological workstation.
Edwards, G. J.: Implementation of FSL's data acquisition component of the WFO-Advanced.
Grote, U. H., and R. J. Kahn: Using RISC-based workstation architecture and distributed network design in meteorological workstations.
MacDonald, A. E., and J. S. Wakefield: WFO-Advanced: An AWIPS-like prototype forecaster workstation.
Mathewson, M. A.: Using the AWIPS Forecast Preparation System (AFPS).
Roberts, W. R., P. Kucera, C. M. Lusk, D. C. Walker, and L. E. Johnson: 1995 real-time forecast exercise for WFO-Advanced.
Wier, S. K., and J. S. Wakefield: Using numerical model output to provide initial forecasts of surface weather for the AFPS.
This document is maintained by Joe Wakefield.
Last updated 15 Feb 96