WESTERN REGION TECHNICAL ATTACHMENT
NO. 97-19
JUNE 10, 1997
A LOCALIZED DOWNSLOPE WINDSTORM IN
NORTHERN ARIZONA
Brian E. Tesar and Steve Keighton - NWSO Flagstaff, AZ
Introduction
Strong winds are a common occurrence across northern Arizona, especially during the
cool season when the jet stream and associated baroclinic zone frequents the lower
latitudes of the Southwest. High winds create a significant hazard for the general public,
particularly for the many motorists along Interstate 40, which runs east to west through
Flagstaff and Winslow, as well as the many tourists who visit this Region. Due to the
varied and unique terrain across the Region, many of the strong wind events are localized,
making them quite difficult to forecast. One of the primary challenges facing forecasters
at the National Weather Service Office in Flagstaff, Arizona, is to accurately assess the
potential and specific location for high winds.
On 2 January 1996, strong and damaging winds developed overnight in the Flagstaff
vicinity, peaking in a localized area on the east side of town. This area is only a mile or
so directly downstream from a small, relatively isolated mountain (Mt. Elden), which rises
about 2500 feet above the elevation of Flagstaff (Fig. 1). Wind speeds were estimated
near 80 mph, mainly from damage to pine trees and power lines. However, less than 10
miles downstream, the Pulliam Airport ASOS reported winds under 30 mph. Figure 2
shows a wider view of the terrain over much of northern Arizona. The location of Mt.
Elden, as well as the higher San Francisco Peaks (Humphrey's Peak being the highest
peak in Arizona) and their elevations, are shown in relation to Flagstaff. Also shown in
Fig. 2 are the locations of the NWS upper-air site and Pulliam Airport (ASOS) in relation
to Flagstaff and the above-mentioned terrain features.
This Technical Attachment will examine this windstorm event and the utility of model
gridded data in predicting it. Theoretical concepts generated from studies of downslope
windstorms in other parts of the country (and the world) are first reviewed. The authors
then examine how these theories might be applied to this particular case given the
observations and gridded model forecasts, and how this case may differ somewhat from
the more classic cases most prevalent in the literature.
Basic Theory And Forecast Methods
Previous studies of localized wind events in other mountainous regions have led to a
number of theories concerning the mechanisms responsible. The main objective of this
study was to determine if the windstorm on 2 January 1996 in Flagstaff could be explained
by any of these mechanisms, or if several might have contributed. One complication is
that the somewhat isolated character of the mountain peaks in this particular case is a
potentially significant difference from the more two-dimensional mountain ridges that the
common theories are based on. However, to our knowledge, very little work has been
done to determine how the mechanisms for generating windstorms in the vicinity of
isolated peaks may differ compared to mountain ridges, although some studies have
simulated the general airflow patterns influenced by isolated mountains (see Durran 1990
for a summary of a few of these). Therefore, we had to begin with the assumption that
current theories applied to two-dimensional ridges would have at least some application
to the isolated peak. [In reality, the mountain in question near Flagstaff is not completely
isolated, but neither are the mountains studied in other regions perfect two-dimensional
ridges.]
Many studies and discussions on downslope windstorms and mountain waves often begin
with the "hydraulic jump" concept. However, a simple hydraulic model by itself is often not
considered a viable solution due to its limited view of the atmosphere where energy cannot
be transported vertically through the upper boundary. In the real world, waves can indeed
propagate energy in the vertical (Durran 1986).
One of the earliest and often-referred to studies on low-level amplification of vertically
propagating mountain waves is that of Klemp and Lilly (1975). In this study, the authors
focused on the importance of the stability structure of the troposphere in the low-level
amplification of vertically propagating gravity waves. The structure favorable for this low-level amplification consists of a shallow layer of stable air near mountaintop level followed
by a relatively deep layer of unstable air through the mid-troposphere (see Fig. 3a).
Clark and Peltier (1977) and Durran (1986) were among the first to discuss the role of a
"critical level" in contributing to low-level mountain wave amplification (and ultimately the
generation of a downslope windstorm). The critical level, defined as the level at which the
wind component perpendicular to the ridge goes to zero, reflects energy back down toward
the wave perturbation, which leads to an increase in the low-level winds. While mean
state critical levels are not always observed upstream from the ridge during wave-induced
windstorms, studies of Boulder windstorms have shown that a critical level can be induced
by the mountain wave itself. Colman and Dierking (1992) suggested that an upstream
critical level may be more important for shallower ridges compared to taller mountain
ranges.
In the absence of a mean-state critical level, the vertical wind shear may also be important
in determining the potential for strong downslope winds. The existence of weak vertical
shear, or ideally reverse shear, is more favorable for low-level amplification compared to
forward shear (Brown 1986). Strong forward shear may tend to prevent a wave-induced
critical level from developing, and is thought to trap vertically-propagating waves, thus
preventing amplification and wave breaking.
Work by Holmes (1994) focuses on the importance of synoptic scale forcing when
determining the likelihood of strong downslope winds. The most significant contribution
from synoptic scale downward forcing is believed to be its effect on the generation of the
vertical stability structure in the lower troposphere, thought necessary for amplified
mountain waves. Gridded output of NCEP numerical model data allows for three-dimensional diagnosis of this structure as well as the synoptic forcing mechanisms
responsible for it.
Summarizing the above-mentioned studies and theories, a basic set of conditions thought
to favor the onset of strong downslope winds can be considered to help forecast these
events:
1. The wind flow is across a terrain barrier (generally within 30 degrees of
perpendicular).
2. The atmospheric layer immediately above mountaintop level usually exhibits strong
stability with a layer of weaker stability above.
3. A level may be observed that exhibits a wind direction reversal or where the cross-barrier flow simply goes to zero (the mean state critical level).
4. The existence of weak, vertical wind shear or reverse shear is more favorable than
forward shear.
5. Synoptic scale downward forcing helps to generate and reinforce the vertical static
stability structure conducive to strong mountain waves.
In addition, strong surface pressure gradients (that are not a direct reflection of the wave)
may enhance downslope windstorms, but in many cases are thought to be incidental to the
overall synoptic pattern generating the event. While surface pressure gradients play a
primary role in gap flow windstorms, some studies in other regions suggested in can be
difficult to separate gap flow and mountain wave effects in generating lee slope windstorms
(Mass and Albright, 1985; Colman and Dierking, 1992).
With these basic principles in mind, the authors set out to examine the Flagstaff windstorm
of 2 January 1996, and the performance of gridded data sets in forecasting this event over
the variable and unique terrain of northern Arizona.
Case Study - 2 January 1996
The prelude to this event began on the previous morning (1 January 1996), when a wind
advisory was issued for much of northern Arizona. Winds of 30 to 40 kts were noted just
above the surface on the VAD wind profile from the KFSX WSR-88D (not shown).
However, the advisory was canceled at midafternoon as strong winds failed to develop at
the surface. Early the next morning (January 2) damaging winds raked through
northeastern portions of Flagstaff, downing trees and power lines with estimated wind
gusts of 80 mph around 0900 UTC. Fifty mph gusts continued until around 1400 UTC.
Strong winds (25 mph with gusts to 40 mph) continued through the day in most areas, but
never again reached the level of the localized winds in the early morning hours.
Numerical models were forecasting strong synoptic scale downward motion in association
with a jet streak over northeastern Arizona during the early morning hours of January 2.
This strong subsidence likely contributed to the development of a stable layer near
mountain top level (shown later in Fig. 7). As noted earlier, favorable conditions for the
low-level amplification of mountain waves include a stable layer just above the mountain
top level.
[Note: The mountain in question in this case is Mt. Elden (about 9,300 feet above sea
level), which is the relatively isolated mountain shown just upstream from the wind-damaged area of Flagstaff in Figs. 1 and 2. The taller San Francisco Peaks (the highest
of which is 12,633 foot Humphrey's Peak) are a few miles to the northwest of Mt. Elden,
and also relatively isolated. It is unknown whether any damaging winds occurred
downstream or in the immediate vicinity of these peaks since this area is sparsely
populated.]
Figure 4 shows the 29 km Meso Eta model 9-hour forecast of 250 mb winds valid at 1200
UTC. A 155 knot jet streak is entering the northeast quarter of Arizona placing the right
exit region over northern Arizona. Further evidence of downward motion associated with
this feature can be ascertained from Fig. 5, which shows the Meso Eta model 9-hour
forecast of 700-500mb layer Qn vectors (the component parallel to the temperature
gradient vector, and dominant for straight jets) and thicknesses valid at 1200 UTC. This
shows a large area of strong Qn vector divergence (in association with the jet streak) at
the mid levels, indicating strong ageostrophic flow and significant downward motion over
northern Arizona. [Note: while gravity waves can often dominate QG diagnostics of
vertical motion fields in high resolution numerical models, in this case there is a
pronounced larger scale signature in association with the jet streak circulation in the 29km
Eta model.]
The Meso Eta model time-height section forecast of equivalent potential temperature from
the 0300 UTC run (Fig. 6) clearly depicts the changing vertical structure of the lower
troposphere as isentropes begin to pack together. This developing stable layer lowers to
around mountaintop level (roughly 700 mb) by 0900 UTC. Notice also the higher wind
speeds gradually dropping to lower levels throughout the early morning hours. The
observed FGZ sounding (see Fig. 2 for upper-air location at NWS Office) at 1200 UTC 2
January did indeed indicate a stable layer near mountaintop level (Fig. 7).
In Fig. 6, the forecast winds near mountaintop level can also be seen veering somewhat,
beginning around 1200 UTC, which would clearly mean a change to the cross-ridge
component of the wind speed (a decrease if you consider the orientation of the very short
ridge line shown in Fig. 1). Since the damaging surface winds also began to diminish at
this time, it does not seem unreasonable to conclude that this particular ridge orientation
may have influenced the low-level amplification of a mountain wave. If the area affected
by the damaging winds shifted westward as the winds aloft veered, that might suggest the
ridge line was less important compared to the isolated nature of the mountain. On the
other hand, without conducting a very high resolution numerical simulation (one that
captures all the details of the terrain), we cannot be sure exactly how the shape of this
mountain influenced the observed windstorm.
Both the observed sounding and model forecast wind profiles show no wind reversal or
enough change in direction with height to suggest a cross-barrier component going to zero
(in fact, all wind directions are cross-barrier if the mountain is assumed to be truly
isolated), thus, there is no evidence for a mean-state critical level in this case. In addition,
the forecast and observed wind profiles indicate especially strong forward shear through
a deep layer. While the absence of a mean-state critical level is not necessarily a
detriment to the production of a downslope windstorm (as mentioned earlier), the character
of the vertical shear observed in this case would generally be considered a negative
influence, since strong forward shear may act to prevent a wave-induced critical level.
Perhaps the forward shear between 0900 and 1200 UTC in the layer where the mountain
wave formed was not quite strong enough to prevent a wave-induced critical level from
developing.
By 0000 UTC 3 January (late afternoon), the stable layer was forecast to drop below
mountaintop level (Fig. 6), while wind speeds near mountaintop level were also forecast
to decrease throughout the day on 2 January. During the afternoon and evening, a
significant decrease in surface winds was observed across the forecast area.
Conclusion
The windstorm of 2 January 1996 in northeast Flagstaff appears to have been primarily
the result of two factors. First, the exceptionally strong jet streak diving from the north
across the area resulted in significant cross-barrier flow from north to south over Mt. Elden
and the San Francisco Peaks. Secondly, the strong synoptic scale subsidence forecast
to occur with this jet streak helped create a stability profile favorable for the low-level
amplification of mountain waves. On the other hand, the absence of a mean-state critical
level coupled with apparently strong forward vertical wind shear suggests conditions a little
less favorable for an amplified mountain wave.
There could have been a few other factors which influenced the development of the
damaging lee-side winds. A moderate northeast-southwest pressure gradient existed over
the Region (not shown), but with sparse surface observations it is very difficult to
determine the local strength of the gradient. Since some of the damage occurred more on
the edge of Mt. Elden instead of strictly downstream, it is possible that surface winds
accelerated around the side of the isolated mountain in a manner similar to gap flow
accelerations.
The shape of the slope and the time of occurrence may have also played roles by reducing
the frictional drag. A steep downwind slope (which exists in this case) may limit the length
where surface friction can occur, and may also help amplify the effects of the wave (Lilly
and Klemp, 1979). Since the event occurred primarily at night, a decoupled boundary
layer may have allowed winds near mountaintop to increase with the reduced frictional
drag, resulting in stronger surface winds when the vertically propagating mountain wave
reached the surface. Other similar downslope windstorm events have been known to
occur more frequently at night, such as the "Sundowner" winds in southern California
(Ryan, 1996).
Gridded model data did well in recognizing the downward motion and strong flow with this
case as well as the existence of a stable layer. However, the resolution of the operational
models would not normally allow a forecaster to pinpoint the location of high winds during
an event like this. It is hoped that simulations of this event using the MM5 model with an
inner grid resolution of 3 km will reveal further evidence of the mechanisms behind this
event.
The problem we will have with future events is related to the somewhat isolated nature of
the mountains. A variety of wind directions could result in topography-related windstorms,
but each at different locations relative to the mountain. Furthermore, most of the research
efforts on mountain waves so far have dealt with elongated ridges or a range of mountains
rather than isolated peaks typical of northern Arizona; much is still unknown about the flow
characteristics around and over this type of isolated terrain. For example, how can gap
flow arguments be applied around the edges of a relatively isolated peak?
Unfortunately, the relatively sparse observation and spotter networks across northern
Arizona make it difficult to verify many of these localized wind events. Yet with more
experience, higher resolution models, and further investigation, forecasters at NWSO
Flagstaff will attempt to link certain synoptic patterns and atmospheric structures to local
terrain-induced events and provide accurate forecasts of these windstorms to the public.
Acknowledgments
The authors wish to thank the following individuals for their thoughtful and constructive
suggestions for improvement: Mr. Eric Thaler, Science and Operations Officer (SOO),
Denver NWS; Dr. Brad Colman, SOO, Seattle NWS; and Dr. Louisa Nance, University of
Washington
References
Brown, J.M., 1986: A Decision Tree for Forecasting Downslope winds in Colorado.
Preprints, 11th Conference on Weather Forecasting and Analysis, Kansas City, MO, AMS
(Boston), 83-88.
Clark, T.L., and W. R. Peltier, 1977: On the evolution and stability of finite amplitude
Mountain waves. J. Atmos. Sci. , 34, 1715-1730.
Colman, B.R., and C.F. Dierking, 1992: The Taku Wind of Southeast Alaska: Its
Identification and Prediction. Wea. and Forecasting, 7, 49-64.
Durran, D.R., 1986: Another look at downslope windstorms. Part I: The development of
analogs to supercritical flow in an infinitely deep, continuously stratified fluid. J. Atmos.
Sci., 43, 2527-2543.
Durran, D.R., 1990: Mountain waves and downslope winds. AMS Monograph Atmospheric
Processes over Complex Terrain, Chap. 4.
Holmes, R. T., 1994: Using new technology to operationally forecast severe downslope
windstorms along the front range. Central Region Technical Attachment No. 94-19,
National Weather Service, Central Region.
Klemp, J.B. and D.K. Lilly, 1975: The dynamics of wave induced downslope winds. J.
Atmos. Sci., 32, 320-339.
Lilly, D.K., and J.B. Klemp, 1979: The effects of terrain shape on non-linear hydrostatic
mountain waves. J. Fluid Mech., 95, 241-261.
Mass, C. F., and M. D. Albright, 1985: A severe windstorm in the lee of the Cascade
Mountains of Washington state. Mon. Wea. Rev., 113, 1261-1281.
Meir, K., 1994: A Southern California Wind Event: An Alternative Explanation. Western
Region Technical Attachment, No. 94-11, National Weather Service, Western Region, Salt
Lake City.
Ryan, G., 1996: Downslope Winds of Santa Barbara, California. Western Region Technical
Memorandum, NWS WR-240, National Weather Service, Western Region, 44 pp.
Western Region Technical Attachment, No. 90-31, Downslope windstorms and mountain
waves.