A Local-Area Model Comparison of the October 1997
Rocky Mountain Blizzard and Severe Wind Event

John S. Snook
NOAA Forecast Systems Laboratory, Boulder, CO 80303

Michael P. Meyers
NOAA, National Weather Service, Grand Junction, CO

Gregory S. Poulos
Colorado Research Associates, Boulder, CO

Douglas A. Wesley
UCAR, Cooperative Program for Operational Meteorology, Education and Training, Boulder, CO

1. Introduction

A wide variety of significant weather covering a large region was observed during the Rocky Mountain winter storm of 23-25 October 1997. Two particularly hazardous areas included, 1) blizzard conditions and heavy snowfall from Wyoming to southern New Mexico along the Front Range of the Rocky Mountains, and 2) extreme winds exceeding hurricane force at several locations west of the mountains. Observations showed total snowfall amounts exceeding 100 cm in many Colorado mountain locations east of the Continental Divide. In addition to the heavy snow, a high mesoscale variability in snowfall amounts was observed with up to a 100-cm snow total variation over a 15-km distance. Meanwhile, on the west side of the Divide, snowfall amounts were light, but high winds caused numerous weather related problems. Wind gusts at 3800 m (12,500 feet) exceeded 45 m s-1 (100 mph) for 5 hours at the Arapahoe Basin ski area just west of the Continental Divide, and a peak gust of 51 m s-1 (114 mph) from the east was recorded. Wind gusts were possibly even higher northeast of Steamboat Springs where a region of severe wind destroyed over 20,000 acres of forest. The destruction had the appearance of a Colorado Front Range downslope windstorm, but it was unusual in that the strong winds were easterly and the damage was located to the west of the mountain barrier. Further observational details of this storm are described in a companion paper by Wesley et al. (1998).

The National Oceanic and Atmospheric Administration Forecast Systems Laboratory (FSL) has been demonstrating a data assimilation system called LAPS (Local Analysis and Prediction System). LAPS is designed to function in a local forecast office using affordable computer hardware. All available data sources, including Doppler radar, satellite, wind profiler, pilot reports, automated surface observations, and other conventional data, are collected and used to generate local area analyses of the atmosphere (Albers et al. 1996, McGinley 1995). The analyses are then available as initialization to local area forecast models (Snook et al. 1998, Snook and Pielke 1995, Snook et al. 1995). The National Centers for Environmental Prediction (NCEP) suite of numerical prediction models provided excellent regional scale guidance for the area and timing of heavy snow (Poulos et al. 1998). These national domain models, however, are not configured to provide detailed forecasts of mesoscale phenomena such as the snowfall variability and the local areas of extreme winds observed in the October 1997 Rocky Mountain storm. LAPS analyses and forecast products are designed to provide additional mesoscale prediction guidance and understanding of local weather events to the local forecast office. This paper focuses on the utilization of local area numerical predictions and their application to the October 1997 storm to demonstrate these local area concepts.

2. Local Area Model Forecasts

This study is designed to investigate the value added and limitations of using high-resolution, local-area models to predict highly variable mesoscale weather events. The local-area model forecasts completed thus far focus on the high wind event observed west of the Colorado Continental Divide. Peak wind gusts were observed during the morning of 25 October between 1100 and 1700 UTC (Wesley et al. 1998). The predictive component of LAPS is designed to utilize any available mesoscale forecast model. Typically, LAPS analyses are used to initialize the forecast model, and a larger domain model (usually NCEP's Eta model) is used as forecast lateral boundary conditions. The archived LAPS analyses are not available at this time for this study. As a substitute, model initializations were accomplished using the 60-km Rapid Update Cycle (RUC) analyses (Benjamin et al. 1991). Plans are to retrieve the LAPS analyses for the October 1997 case, re-run the local area forecasts, and present these results at the conference.

All model forecasts were initialized with 25 October 0600 UTC RUC analyses, and the models generated predictions out to 18 h using the NCEP operational Eta model as forecast lateral boundary conditions. Results from other model forecasts investigating the high-resolution details of the heavy precipitation and wind are described in companion papers by Poulos et al. (1998) and Meyers et al. (1998). Three forecasts have been completed thus far using 1) the Regional Atmospheric Modeling System (RAMS) developed at Colorado State University (Pielke et al. 1992), 2) the NCEP operational hydrostatic Eta model (Black 1994), and 3) the NCEP nonhydrostatic Eta (nh-Eta) model.

RAMS was the first mesoscale prediction model to be used in conjunction with LAPS (Snook et al. 1995). A nonhydrostatic version of the model is implemented with a full microphysics option (Walko et al. 1995). RAMS was configured with a two-way interactive double grid domain, in which the outside domain used a 61 x 61 15-km horizontal grid covering nearly all of Colorado and Wyoming and portions of surrounding states (Fig. 1). The inner nest domain used a 71 x 71 5-km grid that covered the regions of observed high winds (Fig. 1). Table 1 summarizes the three model configurations.

A version of NCEP's operational hydrostatic Eta forecast model was implemented at FSL in January 1998 to function on high-resolution, local-area domains. The model has been running in real time, twice daily, within the LAPS framework. A hydrostatic Eta case study forecast was completed using a single 10-km horizontal grid (the Eta model does not have an interactive grid nest capability) that covered approximately the RAMS 15-km outer nest domain (Fig. 1). Extra eta levels were used to maintain reasonable vertical resolution in the boundary layer of the mountainous terrain (Table 1). FSL recently acquired a test version of NCEP's nh-Eta forecast model. The third local area model forecast used nh-Eta applied on the same 10-km grid used by the hydrostatic Eta prediction. Both Eta models use identical numerics and physics, hence any significant Eta model output differences can be attributed to nonhydrostatic effects.

3. Forecast Results

Topographies used by the RAMS and Eta models are depicted in Figure 2. Figure 2a shows the entire 5-km inner grid domain used by RAMS and Figure 2b is a subset of the Eta 10-km grid. Note that the topography used in RAMS has been smoothed with a 4-dx filter while no smoother has been applied to the Eta topography. The large area of forest destruction occurred to the north-northeast of Steamboat Springs in the Routt National Forest located just west of the Continental Divide (Wesley et al. 1998). The Arapahoe Basin ski area, which reported a peak wind gust of 51 m s-1, is located approximately 80 km west of Denver (Fig. 2).

Six-hour wind forecasts from the three models are illustrated in Figure 3a, Figure 3b, and Figure 3c. Both RAMS and nh-Eta indicate three areas of high winds exceeding 17.5 m s-1: 1) north of Steamboat Springs and west of the Continental Divide, 2) along the Continental Divide in the vicinity of the Arapahoe Basin ski area, and 3) west of the Continental Divide in Grand County (approximately halfway between Steamboat Springs and Denver). Hydrostatic Eta also shows the three areas of higher winds, but the magnitudes are about 8-10 m s-1 lower. RAMS and nh-Eta both show a north-south elongated region of high winds that corresponds well with the area of forest destruction; however, the magnitude is significantly underforecast. The underforecast is partially due to the models predicting a sustained wind, and it is unrealistic to expect the models to capture the strength of peak gusts using these model grid resolutions.

RAMS and nh-Eta also appear to have successfully predicted the area of high winds along the Continental Divide west of Denver as observed at the Arapahoe Basin ski area, but again the magnitude has been underforecast. Although no high winds were observed west of the Continental Divide in Grand County, it is reasonable to expect that high winds did occur in this region but they were not observed in this sparsely populated area. These latter two regions of high winds are much more localized in nh-Eta than in RAMS. This may be due to the unsmoothed topography used by the Eta model. Plans are to use smoothed topography in Eta to investigate this effect.

Predictions of high winds downwind of the mountain barrier suggest that mountain wave activity was an important component and also, given the lower wind speed forecasts in hydrostatic Eta, suggest the importance of nonhydrostatic effects. West to east vertical cross sections through the region of strongest winds north of Steamboat Springs are depicted in Figure 4a, Figure 4b, and Figure 4c for the 6 h forecasts of the three models. All forecasts show a very stable lower troposphere and a relatively unstable middle and upper troposphere. The strong decrease in stability with height corresponds to a rapid decrease with height in the Scorer parameter, which suggests a trapped lee wave scenario (e.g. Pielke 1984). Trapped lee waves are typically confined to the low stable layer and decay rapidly with height. The RAMS forecast has a well-developed mountain wave within the stable layer, winds exceeding 25 m s-1 on the lee of the Continental Divide, and a relative wind speed minima of less than 10 m s-1 in the upper portion of the wave.

The nh-Eta forecast shows a somewhat less developed mountain wave and a smaller region of maximum wind speeds also exceeding 25 m s-1. Two factors may contribute to the smaller sized maximum -- the topography is unsmoothed and the middle troposphere is somewhat more stable than the corresponding RAMS prediction. The latter factor may be related to the large amounts of precipitation and latent heat release upstream of the mountains. The importance of these upstream features is a topic of current investigation. The hydrostatic Eta forecast clearly shows no indication of a mountain wave and only a small increase of wind speed in the vicinity of the mountain barrier. These results suggest the importance of nonhydrostatic effects in the prediction of downslope wind storms.

4. Conclusions

LAPS is a local area data assimilation and prediction system designed to provide additional mesoscale guidance to the operational forecast office. The 23-25 October 1997 Rocky Mountain blizzard and severe wind event exhibited a large mesoscale variability in snowfall amounts and in extreme winds. LAPS concepts applied to this case study illustrate the additional mesoscale guidance that local area models can provide.

Three local area model forecasts using RAMS, Eta, and nh-Eta were completed to investigate the high wind event of 25 October 1997 that destroyed over 20,000 acres of forest near Steamboat Springs, Colorado. RAMS and nh-Eta forecasts successfully predicted the regions of highest winds. These forecasts were unable to predict the strength of the observed peak wind gusts, but substantial mesoscale value was added when compared to the current coarser resolution, hydrostatic operational models. Forecast vertical cross sections from RAMS and nh-Eta indicate that the high wind event resulted from a trapped lee wave located in a very stable lower tropospheric layer. The event was unusual in that the winds flowing over the mountain barrier were easterly and that the high winds were observed west of the barrier.

The hydrostatic Eta forecasts were unable to capture the details of the mountain wave and high winds suggesting the importance of using a nonhydrostatic model for local area predictions. The model results indicated that smoothed topography and upstream precipitation may affect the successful prediction of a mountain wave. These factors are currently being investigated in more detail. The case study results are indicative of the improved understanding and predictability of mesoscale phenomena that locally generated, local area prediction models can provide to the operational weather forecast office.

5. Acknowledgements

The authors wish to thank Dr. George Kallos of the University of Athens for providing Eta preprocessing code. Thanks to Tom Black and Mike Baldwin for their help in providing Eta model code. Alan Henceroth provided observations from the Arapahoe Basin ski area. Drs. Roger Pielke and William Cotton of Colorado State University and Dr. Craig Tremback of Mission Research Corporation are acknowledged for their continued permission to use RAMS for this project. Ligia Bernardet reviewed the article and Nita Fullerton provided technical editing support. RAMS was developed under the support of the National Science Foundation and the Army Research Office.

6. References

Albers, S. C., J. A. McGinley, D. L. Birkenheuer, and J. R. Smart, 1996: The Local Analysis and Prediction System (LAPS): Analysis of Clouds, Precipitation, and Temperature. Wea. Forecasting, 11, 273-287.

Benjamin, S. G., K. A. Brewster, R. Brummer, B. F. Jewett, T. W. Schlatter, T. L. Smith, and P. A. Stamus, 1991: An isentropic three-hourly data assimilation system using ACARS aircraft data. Mon. Wea. Rev., 119, 888-906.

Black, T. L., 1994: The new NMC mesoscale Eta model: Description and forecast examples. Wea. Forecasting, 9, 265-278.

McGinley, J. A., 1995: Opportunities for high resolution data analysis, prediction, and product dissemination within the local weather office. Preprints, 14th Conference on Weather Analysis and Forecasting, 15-20 January 1995, Dallas, TX, Amer. Meteor. Soc., 478-485.

Meyers, M. P., J. S. Snook, D. A. Wesley, and G. S. Poulos, 1998: Numerical investigation of the 25 October 1997 blowdown event over Colorado Park Range using RAMS. Preprints, Eighth Conference on Mountain Meteorology, 3-7 August 1998, Flagstaff, AZ, Amer. Meteor. Soc.

Pielke, R. A., 1984: Mesoscale Meteorological Modeling. Academic Press, 612 pp.

Pielke, R. A., W. R. Cotton, R. L. Walko, C. J. Tremback, W. A. Lyons, L. D. Grasso, M. E. Nicholls, M. D. Moran, D. A. Wesley, T. J. Lee, and J. H. Copeland, 1992: A comprehensive meteorological modeling system - RAMS. Meteor. Atmos. Phys., 49, 69-91.

Poulos, G. S., D. A. Wesley, J. S. Snook, and M. P. Meyers, 1998: Modeling small-scale spatial heterogeneity of snowfall in complex terrain: The October 1997 Front Range blizzard. Preprints, Eighth Conference on Mountain Meteorology, 3-7 August 1998, Flagstaff, AZ, Amer. Meteor. Soc.

Snook, J. S., P. A. Stamus, J. Edwards, Z. Christidis, and J. A. McGinley, 1998: Local-domain mesoscale analysis and forecast model support for the 1996 Centennial Olympic Games. Wea. Forecasting, 13, 138-150.

Snook, J. S., and R. A. Pielke, 1995: Diagnosing a Colorado heavy snow event with a nonhydrostatic mesoscale numerical model structured for operational use. Wea. Forecasting, 10, 261-285.

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, 864-875.

Wesley, D. A., G. S. Poulos, M. P. Meyers, J. S. Snook, and A. Judson, 1998: Observations and forcing mechanisms during the October 1997 Front Range blizzard and forest destruction. Preprints, Eighth Conference on Mountain Meteorology, 3-7 August 1998, Flagstaff, AZ, Amer. Meteor. Soc.

Walko, R. L., W. R. Cotton, J. L. Harrington, and M. P. Meyers, 1995: New RAMS cloud microphysics parameterization. Part I: The single-moment scheme. Atmos. Research, 38, 29-62.