1. Introduction

The NWS (National Weather Service) and GHCC (Global Hydrology and Climate Center) Collaborative Workshop on Mesoscale Modeling and Short-term Convective Forecasting was held at the GHCC in Huntsville Alabama on 30 June and 1 July 1999. The workshop was organized by Tom Bradshaw (WFO BHM [Weather Forecast Office Birmingham AL]) and Bill Lapenta (GHCC/NASA). The purpose of the meeting was to foster interactions between the research and operational communities in the southeastern United States. NWS personnel were introduced to the types of atmospheric research performed at the GHCC. In turn, the operational community provided feedback to GHCC scientists regarding possible applications of their research in the field.

A total of thirty-two invited attendees were present. The following NWS Offices were represented: Birmingham, Huntsville, and Mobile in Alabama; Peachtree City and the Southeast River Forecast Center (SERFC) in Georgia; Tallahassee in Florida; Paducah in Kentucky; and Morristown and Nashville in Tennessee. In addition, scientific organizations and universities represented included NASA, NCAR (National center for Atmospheric Research), CIRA (Cooperative Institute for Research in the Atmosphere), ENSCO Inc., FSU (Florida State University), and UAH (University of Alabama in Huntsville).

2. Summary of Presentations

Tom Bradshaw (WFO BHM), Bill Lapenta (GHCC/NASA) and Don Perkey (GHCC) welcomed the workshop participants. Opening remarks discussed how the collaboration between BHM and the GHCC developed during the past year in support of a project under the USWRP (US Weather Research Program) funded by Dr. Rahmesh Kakar of NASA Headquarters. It was also noted that the encouragement and support of Dan Smith (NWS Southern Region Headquarters) fostered the organization of the workshop. The meeting proceeded with three sessions.

Session 1: Operational Issues Related to Short-term Forecasting.

• Dan Smith (NWS/SRH/SSD) began the session by providing insight into the trend of forecast offices to use output from mesoscale models other than those provided by NCEP.

• Judi Bradberry (SERFC) then discussed the use of the quantitative precipitation forecast (QPF) in river stage forecasting. She provided an overview of the SERFC operations and an explanation of who creates the QPF and how it is used in the Hydrologic Model.

• Tom Bradshaw (WFO BHM) discussed SCRAPE (Summer Convective Rainfall in Alabama Prediction Experiment, which is being conducted at WFO Birmingham this summer. The goal of SCRAPE is to determine the feasibility of producing short-term, high-resolution QPFs for the Birmingham forecast area, and to evaluate the relative utility of various observational and forecast model tools in this effort. A fundamental source of data for this exercise is MM5 (Mesoscale Model 5) mesoscale model output, provided on a real-time basis by Dr. Bill Lapenta of GHCC/NASA. Preliminary results so far have indicated that the introduction of MM5 data into the forecast process leads to an improved QPF product which hopefully can be of benefit to the general public.

Session 2: Ground and Space-based Observational Platforms Related to Nowcasting.

• Cindy Mueller (NCAR) provided an overview on the Auto-Nowcasting system currently being deployed at Redstone Arsenal in northern Alabama. [Additional details will be provided in the near future].

• Kevin Knupp (GHCC/UAH) [Summary will be provided in the near future]

• Brian Motta (CIRA) discussed the importance of integrating the remote sensor data in an AWIPS environment for analyzing and diagnosing conditions worthy of warnings. An example from the Fort Collins Flood of 1998 showed how flooding rainfall could occur with warm-topped convection and minimal lightning activity. Such events would have high reflectivity on radar but meteorologists also need to look at the cloud phase information available from satellite and lightning data. The importance of using GOES Rapid Scan Operations data was also discussed since this capability in NWS offices is relatively new with the installation of AWIPS.

• Ron Suggs (GHCC/NASA) then provided an overview of the GHCC GOES satellite data activities. The remote sensing group at GHCC has put together two satellite data systems that provide geostationary image data in various formats and also GOES derived products. The first data system acquires and makes available data measured by GOES, METEOSAT and GMS satellites. This data is acquired from various data providers having access to ground stations for the above satellites. The GHCC also has a ground station for GOES-8 from which data is provided near real time for the eastern U. S. The geostationary satellite data acquisition system provides the collected image data in various binary data formats including HDF and McIDAS area files which can be obtained by the research and operational communities as well as the general public through anonymous FTP at our web site.

• Steve Goodman (GHCC/NASA) [Summary to be provided in the near future]

• Todd Lericos (FSU) discussed analysis of ten years of data from the National Lightning Detection Network (NLDN) for selected flow regimes over the Florida peninsula. Data for the years 1989-98 are organized by map types that primarily represent placement of the sub-tropical ridge and other commonly found synoptic patterns. Only warm season data between 1 May and 30 September are used. The placement of the sub-tropical ridge with respect to the Florida peninsula is found to play an important role in the spatial and temporal distribution of lighting. In addition, complexities in the coastline of Florida are responsible for producing areas of higher flash densities in four major geographic areas. Nocturnal analysis of lighting data reveals a relation to ocean currents such as the Gulf Stream, as well as coastline orientations. The study provided insight into the spatial and temporal evaluation of convection associated with each synoptic pattern and helps identify areas favored for convection. Furthermore, the study results identify areas of convective activity that are not easily understood, serving to give guidance for future study.

• John Manobianco (ENSCO/AMU) started the session with a presentation entitled "Mesoscale modeling over the Florida peninsula." The presentation highlighted the Applied Meteorology Unit's (AMU) evaluation of the Mesoscale Atmospheric Simulation System (MASS), the NCEP Meso-Eta model, and the Regional Atmospheric Modeling System (RAMS) currently running in the Eastern Range Dispersion Assessment System (ERDAS). In addition, results were shown from Dr. Anil Rao's dissertation research using the Advanced Regional Prediction System (ARPS) at 100-m resolution to simulate boundary interaction and Kelvin Helmholtz waves over the Kennedy Space Center/Cape Canaveral Air Station. Although the AMU evaluations included both comprehensive objective and subjective verification, the presentation focused on model capabilities to forecast warm season convection and convective initiation.

• Bill Lapenta (GHCC/NASA) then showed results from a GHCC project where GOES-derived and surface products are assimilated into the PSU/NCAR MM5 over the Southeast. The motivation is to achieve a realistic surface energy budget in the presence of differential land surface characteristics. Preliminary results showed this type of assimilation can be performed in near real-time and improve the simulations of near surface temperature and relative humidity.

• Dick McNider (GHCC/UAH) discussed pollution/ozone transport modeling in the SE US. [Additional details will be provided in the near future]

• Qi Mao (GHCC/UAH) discussed an optimal model output calibration (MOC) technique that has been tested recently with the PSU/NCAR MM5 model for maximum temperature forecasts over a summer period of 1-28 July 1997. This technique, which differs from existing methodologies and the traditional Model Output Statistics (MOS), uses information from observations and model forecasts of the most recent seven days to objectively estimate and adjust current model forecast errors and make refined predictions. Since the MOC does not rely on long-term statistics of a specific model version and it varies from day to day and from site to site, it can be used whenever an evolving model is applied or when an insufficient amount of experience is available for a particular model configuration.

• Matt Duke (FSU) concluded the presentations and showed how he incorporated high-resolution satellite-derived SST data into the MM5 initialization in place of a SST field derived from the ETA 2meter temperature field. His results showed that in certain wind flow and temperature regimes, there were large differences in the SST initialization and in the resulting thermal circulation of the sea breeze.

3. Summary of Discussions

There were two primary areas of discussion during the workshop. The first was related to the generation of short-term forecasts (nowcasts) by the NWS and the second focused on the use of high resolution model data within WFO’s. Below are some of the highlights of those discussions.

Nowcasts

The discussion at the conclusion of Session 2 centered on the future direction of nowcasts issued by the NWS. Currently, nowcasts consist of short (usually less than eight line) text products which describe expected weather conditions for multi-county areas during the next six hours. Tom Bradshaw of WFO Birmingham mentioned that nowcasts today often tend to be either: 1) too detailed to remain useful for a sufficient time after issuance, or: 2) worded so generally that they are of limited use to specific users. Several forecasters in attendance noted that the production of detailed, yet timely nowcasts represents a significant workload. This is an important human resource issue, given the declining staffing levels at NWS offices.

Two NWS attendees were uncertain of the usage of their offices’ nowcasts. Dan Smith of NWS Southern Region Headquarters replied that one of the tasks of each office’s Warning Coordination Meteorologist (WCM) should be to identify present nowcast users, and to promote this product to potential users. One forecaster stated that some of his users preferred visual radar displays and text-based radar summaries (no longer issued by the NWS) over nowcasts. He questioned the utility of the nowcast, given the availability of real-time radar displays on television and the Internet. Dan Smith and Tom Bradshaw both responded that the nowcast is supposed to be a six-hour forecast, not simply an hourly radar-based extrapolation of precipitation. Thunderstorms and other small-scale weather phenomena often evolve significantly over short time scales. In theory, nowcasts, incorporating a synthesis of data sources (e.g., radar, satellite, lightning, surface observation networks, mesoscale model output), should represent a more accurate depiction of expected weather conditions in the 0 to 6 hour time frame than that achieved by radar imagery alone.

An important element of the discussion focused on potential new avenues of dissemination for nowcasts and other types of NWS forecasts. A meteorologist from the Peachtree, Georgia office noted that text nowcasts were seldom displayed on the Weather Channel, and that relatively few members of the public had access to NOAA Weather Radio (NWR). Tom Bradshaw and others mentioned the advantages of creating graphical forecasts, given the increasing role that visual-based mediums like television and the Internet play in disseminating information in our society. It was suggested that short-term QPF products like those being created experimentally at WFO Birmingham could be made available to the public via NWS Internet web sites. Feedback regarding the usefulness of these products could be solicited from users on these same sites via e-mail. Others, however, stressed the continued importance of text-based products, which are critical input to the new automated NWR Console Replacement System (CRS). The consensus of the operational and research meteorologists in attendance was that in order for short-term forecasts to enjoy maximum use, they would need to be disseminated in a variety of textual and visual formats. It was noted that nowcast creation and dissemination will continue to challenge the NWS during the coming years. This area will likely represent a fertile topic for further collaboration with non-NWS agencies (similar to the link developed between WFO Birmingham and the GHCC).

Use of supplemental high resolution model data within a forecast office generated on site or by an external institution.

The discussion at the conclusion of the modeling session focused on the use of "supplemental" mesoscale model data within the NWSFO’s. It should be referred to as supplemental since it is typically used along with numerical guidance provided by NCEP. There are currently a number of forecast offices around the country that make use of supplemental output from mesoscale models (Seattle, Tallahassee, Melbourne and others). A forecaster [TJ Turnage] from the Tallahassee office described how model data from MM5 are incorporated into their forecast procedure on a daily basis. These data are also used by forecasters to construct climatologies of the simulated sea breeze.

A question that arose from the discussion was "How can an WFO determine if supplemental mesoscale model output can be of value to forecasters?" It was suggested that the office would need to develop a detailed plan that considered the following issues:

1. What can model data having a higher spatial and temporal resolution provide the office that the current suite of NCEP operational models can not? Each forecast office typically has a unique forecasting problem. Sea breeze circulation’s along the Gulf Coast, elevated convection, precipitation type, cold-air damming along the Appalachian Mountain chain, etc. Therefore, it must be determined if supplemental model data can provide additional information on these phenomenon above and beyond that provided by the Eta and RUC systems.

2. Can the model be run at the office? There are a number of mesoscale models now available that can run efficiently on moderately priced workstations. However, the user needs some level of expertise in numerical modeling in order to decide how the model is to be configured (i.e., domain resolution/size, model initialization procedure, physics options). A model that is improperly configured may not provide useful information to the forecast office.

How to visualize output/products? Models can produce an enormous amount of output. Each office must choose the variables to be extracted from the model and the temporal frequency at which to dump them. In addition, there are many desired diagnostic quantities that are not explicitly output by the models that must be calculated by a post-processing code. Office personnel must also decide how best to visualize the model data. The most efficient way would be to display the data within the AWIPS framework.

The MetEd (Meteorological Education and Training) program funded by NOAA has put together detailed instructional materials on numerical weather prediction as part of the Professional Development Series (PDS) modules. This type of training is highly recommended for those offices interested in using supplemental model data. It is also noted that interactions between WFO’s and local research institutions having mesoscale modeling activities can be beneficial for both the operational and research communities.

4. Concluding Remarks

The workshop organizers would like to thank all those who attended the meeting. Despite being only 10 hours in length, the general consensus among participants was that the workshop was very productive. The goal of fostering dialog between GHCC researchers and the operational community in the Southern Region of the NWS has begun. We can only hope that over the next several years additional collaborative efforts are initiated between the NWS and GHCC as well as other research-based institutions.

Summary

Presentations

Attendees

Back to home page

Weather Hazards | Graphical Forecast | Present Weather
Past Weather | Tropical Weather | Contact Us