Gap Winds

0. Objectives

After completing this module, the learner should be able to do the following things:

• Recall where in a gap the strongest wind speeds are typically observed
• Describe the different kinds of topographic gaps and their effect on gap flow
• List at least 3 natural hazards that may be associated with gap winds

• Describe how wind speed varies through the gap during a gap flow event
• Describe the temperature profile through a gap during a gap flow event
• Describe the pressure profile through a gap during a gap flow event

• Describe the conditions required for geostrophic flow
• Recall that gap winds are typically non-geostrophic
• Describe the origin of the pressure gradients that occur across gaps
• Recall that the thinning of low-level cool air at a gap exit can increase the pressure gradient across a gap
• Recall that adiabatic warming of downslope winds can increase the pressure gradient across a gap

• Qualitatively describe how varying the following factors affects wind speed through a gap:
  • Pressure gradient
  • Surface roughness
  • Gap length
  • Temperature
• Describe the horizontal resolution of a mesoscale model required to accurately forecast flow through a gap

1. Scenario

1.0.1 1500Z 23-Apr-02

I was working the day shift at the Morón AFB Base Weather Station in southern Spain. My name's Jack Slade and I'm a meteorologist. It was late afternoon and it had been a very slow day ops-wise. The weather was really cooperating, too, with clear skies, great visibility, and no change in sight for the next several days. Most of my shift duties had been completed and I was contemplating the arrival of my relief--just one hour to go.

Then the phone rang. "Morón Base Weather, Master Sergeant Slade."

"This is Colonel Atkins, we have an aircraft down in the Strait of Gibraltar. I'm sending a rescue chopper to the scene. How's the weather holding up? "

1.0.2 Current Conditions

"Well Sir, it's clear and a million out there and should stay like that for the next few days, but we do have some moderate to strong winds affecting the Strait at this time. Hang on just one moment sir and I'll check the current observations."

I looked at the latest satellite image and surface winds. There was a 30-knot wind report from Ceuta Point on the Moroccan Coast.

1.0.3 Prognosis

I checked the latest MM5 analysis from the 00Z cycle on the 23rd. The Strait of Gibraltar sits near the edge of the model domains for the European and African theatres. Not great. We can expect some boundary effects for the model. But, it's what we have and we need a forecast now. First I looked at the Africa domain. It's run with a 45-km grid spacing--not as high of a resolution as the Euro domain, but it shows the entire Gulf of Cadiz. The MM5 showed possible 15 to 20 knot winds near shore with 25 to 35 knot winds offshore just west of the Strait. This made sense since I knew gap winds are strongest in the exit region. I was happy to see reasonable agreement between the model run and the observations, despite the relatively coarse grid spacing in the model. Checking further through the MM5 progs, I found that the model predicted that the winds should start dropping off after 1200Z on the 24th.

"Sir, the surface winds are currently out of the east at 25 to 35 knots on the western end of the Strait and extending about two hundred miles into the Gulf. We're forecasting the winds to persist for about 30 hours and then slowly begin to diminish."

"Thanks Slade, keep a close watch on this area. We'll be wanting periodic updates."

1.0.4 Weather Briefing

While I was finishing up with Col. Atkins, my relief arrived and started briefing himself. Technical Sergeant John Seagall was a very competent weather forecaster with two years experience at Morón. The guys had started calling him "Bugsy," but he failed to live up to the name. Seagall was a quiet, intelligent, very professional NCO.

"I just got off the phone with the Director of Operations, Col. Atkins. An aircraft went down in the Strait and he's about to launch a rescue chopper to look for the pilot."

I showed him loops from the 12Z run of the 15-km MM5 over the European domain. The higher resolution gave us more detail, but cut off some of the Gulf of Cadiz.

"As you can see we're dominated by high pressure and the MM5 keeps us cloudless for at least the next 48 hours. The easterly winds on the south side of the high in the western Med are accelerating through the Strait and actually reach their max west of the strait, in the Gulf of Cadiz. We're currently getting exit region speeds of 30 to 35 knots. The latest model is forecasting no change until about the 30-hour point when the existing synoptic pattern begins to break down and the winds, in response, begin to drop off. Take a look at the next model run and keep the Command Post up to speed."

1.0.5 Next Morning

As I drove up to the weather station the next morning, I could see two rescue birds lifting off the tarmac. Obviously, they're still looking. As I walked into the building, Bugsy was busy briefing a crew. I started looking though the latest charts to get a jump on the in-brief. The 08Z observation from Ceuta Point was indicating 15 knots out of the east.

I was pleased to find good agreement between the observation and the most recent MM5 prognosis.

1.0.6 Rescue

Bugsy finally concluded the briefing and I watched the pilot walk away. Bugsy told me that they had found the debris but still hadn't located the pilot. For some reason his transponder wasn't working. They were using our wind forecast to develop a vector map of possible raft drift. Hopefully that would work. Just then the phone rang.

"Morón Base Weather, Master Sergeant Slade."

"Morning, Colonel Atkins here. Just wanted to let you know that we just found the pilot and he's OK. I also wanted to let you know that your forecasts were crucial to the success of this operation. Your wind forecast had the correct direction and enabled the Command Post to make a vector map. He'd drifted quite a bit further than we'd anticipated, though. But it's a big ocean and without that bearing, we'd probably still be looking. Nice job."

So the MM5 picked up the gap wind and got the direction about right, but underestimated the wind speed. I guess that's not too bad considering that we were operating at the edge of model domain and that we really needed better model resolution. It takes a pretty high-resolution model to depict all the details of wind whistling through a gap as narrow as Gibraltar. Next time we'll know to forecast stronger wind speeds.

2. Overview

2.0.1 What are gap winds?

Gap winds are low-level winds that are associated with gaps or low areas in terrain. Gap winds can range in width from hundreds of feet to over one hundred miles, and in unusual circumstances can be associated with strong winds exceeding 50 knots. These winds are normally quite shallow, extending hundreds of feet to a few thousand feet above the surface, with large changes of wind, or wind shear, at their upper and lateral boundaries. Gap winds are generally strongest when there is a large pressure drop across a gap, although there is one class of gap wind that does not depend on a pressure gradient across the gap.

2.0.2 Why are gap winds important for military forecasters?

Ship and aircraft operations require accurate information about low-level winds. Many coastal regions possess substantial terrain, with channels that can produce well-defined gap winds. These gap winds are not only important within the gaps, but also as much as 10 to 100 miles downstream as well. Gap winds can reach strengths of 20 to 60 knots, which can seriously affect military operations, with strong wind shear at the top and sides of the gap flow. Such large wind shears can be associated with moderate or even severe aircraft turbulence. Of course, gap winds can be very important in non-coastal regions as well, with strong winds in and downwind of channels or passes in terrain. Significant gap flows are found throughout the world, many of them in regions where the military has active or potential operations.

2.0.3 What will you learn from this module?

This module will provide a basic understanding of why gap winds occur, their typical structures, and how gap wind strength and extent are controlled by larger scale, or synoptic, conditions. With the availability of new remote sensing assets and high resolution numerical modeling, we now have a much better idea about the structures of gap flows and how they develop, and you will become acquainted with these new insights. It turns out that many of the old ideas, some still found in textbooks, are wrong or incomplete--such as the old funnel or venturi analogs. You will learn about a number of important gap flows in coastal regions around the world, with special attention given to comprehensively documented gap wind cases in the Strait of Juan de Fuca and the Columbia River Gorge. Basic techniques for evaluating and predicting gap flows will be presented. We will also review the capabilities and limitations of the current generation of mesoscale models in producing realistic gap winds. By the end of this module, you should have sufficient background to diagnose and forecast gap flows around the world, and to use this knowledge to understand their implications for operational decisions.

3. Basic Principles

3.1 Venturi Effect

3.1.1 The Funnel or Venturi Effect

In some introductory textbooks and in the "common knowledge" of many meteorologists, gap winds are explained by the "funnel" or "Venturi" effect, in which the strongest winds occur in constrictions. Although funnel effects can influence wind speed, the strongest winds are generally not in the narrowest portion of gaps but rather in gap exit regions. Furthermore, other mechanisms appear to be far more important in modulating wind speed in and near gaps.

3.1.2 The "Funnel" or "Venturi" Effect: Only a Part of the Story

Consider a simple sea-level gap through a mountain range, as shown in this figure. Assume that there is a rigid lid below the crest of the mountains, one that prevents any flow from passing over the mountain range. Thus, any air approaching the mountain barrier can only move across the barrier in one place: the gap. In the situation sketched in this figure, relatively weak easterly flow approaches the gap entrance. Because of conservation of mass, the air must speed up as it flows through the constriction, with the strongest winds at the narrowest point. In other words, the amount of fast-moving air moving through the narrow portion of the gap equals the amount of slower moving air approaching the gap from the east. As the air moves past the constriction into the gap exit region, the winds slow down as the cross section area of the gap increases. According to the Bernoulli principle, pressure should be lowest in the constriction, with the air accelerating from high to low pressure east of the constriction and decelerating downstream of the constriction as it traverses from low to high pressure.

Although physically plausible, the venturi mechanism often appears to be at odds with most real gap wind situations, where the strongest winds are generally over the exit region of the gap, not at the narrowest section or constriction.

There are a number of reasons for the failure of the funnel model. First, there is no rigid lid in the real world. As air approaches a gap, the depth of the approaching air (often relatively cold and dense) generally increases due to the blocking effects of the surrounding terrain. The increase in the depth of cold, dense air over and to the east of the barrier contributes to higher pressure near and upwind of the center of the gap. Subsequently, the wind tends to slow as it approaches the gap.

Second, in the gap exit region, the rapid widening of the gap causes the flow to spread out horizontally and rapidly thin. This thinning of the low-level air results in lowered pressure, which contributes to a pressure gradient over the exit region. As a consequence, winds accelerate over the exit region.

3.1.3 The Role of Synoptic-Scale Pressure Gradients

Another reason the flow tends to be strongest in the gap exit region is that there is often a synoptic scale pressure gradient across the gap. For example, a synoptic high-pressure area might be on one side and a low center might be approaching on the other--thus producing a strong pressure gradient across the barrier. Such pressure gradients result in an acceleration from high to low pressure that continues over the entire gap and thus tend to produce strongest flow over the gap exit.

Venturi effects are also lessened by the complex three-dimensional nature of the air flows in actual gaps. Often air in a gap does not simply move horizontally from the upstream source region to the exit. Rather, air can flow into the gaps from several directions and from several levels along the length of the gap. Thus, simple mass conservation arguments and the assumption that the gap is a closed system can be deceptive or wrong.

3.1.4 When the Venturi Effect Dominates

Although the Venturi effect is often not the dominant mechanism for mesoscale gaps 10 to 100 kilometers wide, it can be very important for small-scale gaps. In this case, the contributions of the larger scale pressure gradients or changes in the height of near-surface cold air are less important. These small-scale gaps are typically on the order of a few kilometers or less in length. Most of us have experienced the funnel effect while hiking though a pass in a mountainous region or felt the winds pick up in the gaps between buildings. A good example of a significant gap flow associated with a small-scale gap in terrain is the Nu'uanu Pali Pass that cuts across the steep, but narrow, Koolau Range of eastern Oahu, Hawaii. As northeasterly trade winds of 10 to 20 knots are funneled into the gap, winds frequently exceed 40 knots near the constriction. Guidebooks warn visitors to hold on to their hats when they visit this windy spot.

3.2 Pressure, Wind, and Terrain

3.2.1 The Relationships Between Pressure and Wind With and Without Terrain

Winds in gaps are typically highly non-geostrophic so that the flow accelerates down the pressure gradient from high to low pressure. In contrast, in geostrophic flow the winds are directed parallel to the isobars and thus move perpendicular to the pressure gradient. This section will review geostrophic and non-geostrophic winds and will explain why winds tend to be ageostrophic in gaps.

3.2.2 The Simplest Case: Geostrophic flow

It is useful to begin by reviewing the relationship of sea level pressure and surface winds when there is no topography and friction. As described in most basic textbooks, without terrain or friction, surface winds are expected to be geostrophic, with winds flowing parallel to the isobars and the strength of the winds proportional to the pressure gradient. When winds are geostrophic there is a balance between the pressure gradient force, directed towards lower pressure, and the Coriolis force, which is directed to the right of the motion in the Northern Hemisphere and left of the motion in the Southern Hemisphere. In reality, geostrophic flow also demands more than just the absence of friction: it also requires (1) that the pressure and wind fields are not changing rapidly in time and (2) that air trajectories are not highly curved. In rapidly evolving situations there is not enough time for the winds to adjust to the pressure fields, so geostrophic balance cannot be achieved. In highly curved flows, particularly when there are strong winds, the centrifugal force can also become important. However, since large scale flows are generally slowly evolving and fast highly curved flows are limited, much of the atmosphere above the surface boundary layer at latitudes higher than 10º is roughly geostrophic. Over the relatively smooth oceans during neutral or unstable conditions, geostrophic flow is often a good approximation.

3.2.3 The Addition of Surface Friction

Near the surface, another force--surface drag or friction--is usually important. Thus, a three-way wind balance is established in the lower atmosphere between the friction, Coriolis, and pressure gradient forces. As a result the wind speed is reduced below the geostrophic value, and the wind blows at an angle across the isobars towards lower pressure. The influence of surface drag depends on the roughness of the surface and the vertical stability of the air in the lower atmosphere. Rougher surfaces, such as hills, trees, and tall buildings, result in more drag, slower wind speeds, and larger cross-isobar angles. Over smoother surfaces, such as the ocean or bodies of water, the winds are more nearly geostrophic. Vertical stability can be important as well. For example, for unstable or less stable situations, such as when cold air passes over warm water, there is enough vertical mixing of fresh, unslowed air from above to allow the surface winds to be more geostrophic. Conversely, in stable situations the lack of mixing results in the drag effects being concentrated in a shallow layer near the surface, which causes far more slowing of the air near the surface.

3.2.4 The Influence of Terrain on Low-Level Winds

Large topographic barriers such as mountain ranges have a profound effect on low-level winds and greatly alter the relationship between pressure and wind. For geostrophic balance to occur, air parcels need sufficient time and space to adjust to the synoptic-scale pressure field. If there is a mountain range in the way and the flow becomes blocked, such an adjustment is impeded.

Consider the situation shown above in which a synoptic scale pressure gradient is oriented so that the isobars are perpendicular to the barrier. The atmosphere is stable so that air tends to move around rather than over mountains. If we assume a relatively ideal world and ignore the effects of surface friction and centrifugal force, then far away from the mountains the flow will be geostrophic, with the winds parallel to the isobars. Near the mountain barrier such westerly flow is not possible because of the blocking effects of the terrain. Instead, the flow tends to move parallel to the barrier, going from high to low pressure. How far away from the barrier does the terrain influence the winds? Typically, this distance, known as the Rossby radius of deformation, is approximately 100 to 200 kilometers.

3.2.5 Isobars Parallel to Mountains

It is also interesting to consider the situation when the isobars parallel a mountain barrier, as shown in the second schematic. In that case, geostrophic winds are possible even near the terrain, because mountain blocking effects don't come into play. Real-world events are a mixture between the two extreme orientations and the pressure gradient is oriented at an angle somewhere between perpendicular and parallel to the mountain barrier.

3.3 Gap Winds In Level Channels

3.3.1 Mechanisms Associated with Gap Winds in Nearly Level Channels

For many gaps in mountainous terrain, the gap is narrower than a Rossby radius, 100 to 200 kilometers. Thus, for such gaps the along-gap winds are not in geostrophic balance with the pressure gradient down the gap. In other words, for gaps less than 100 to 200 kilometers wide, the air moves down the gap towards lower pressure. Thus, if you know how pressure varies along the gap you know the direction of the gap winds and have an idea of the magnitude of the flow. In general, stronger pressure gradients along a gap produce stronger gap flows. In such non-geostrophic situations the Coriolis force, which takes hours to act, is not effective in balancing the pressure gradient force, and thus winds can accelerate greatly, with only friction to keep them in check.

The pressure gradients that drive gap winds have two major origins: (1) pressure gradients associated with synoptic or regional scale features, and (2) pressure gradients in or near the gap associated with rapid changes in the depth of cool air at low levels.

3.3.2 Synoptic-scale Pressure Gradients

Pressure gradients across a gap are often associated with synoptic-scale features. For example, if there is an anticyclone on one side of the gap and a low-pressure region or cyclone approaches or develops on the other side, a large pressure gradient can build across the gap. Generally lesser, but still significant pressure gradients along gaps can be associated with a low or high pressure region approaching or developing on one side of the gap. No matter what the origin, synoptic-scale pressure gradients across the gap tend to produce the strongest winds in the gap exit regions, on the low pressure side, since the winds can accelerate downgradient throughout the length of the gap.

3.3.3 Pressure Gradients Produced by Changes in the Depth of Cold Air: Hydraulic Effects

For many gaps, the barrier in which the gap is embedded separates a cool air mass on one side from a warmer air mass on the other. Surface pressure is higher on the cold air side because the denser cold air produces high pressure at low levels. In most cases the cold air is relatively shallow, 0.5 to 2 kilometers in depth, and overlain by warm air. As a result, there is an inversion or stable layer at the top of the surface-based cold air. In general the inversion is positioned below the crest level of the mountains, so that cold air can escape only through the gap. Otherwise the cold air would push over and across the barrier.

After the cold, dense air moves through the gap it spreads out in the gap exit region, where the gap widens rapidly. Because of conservation of mass, the spreading cold air must become more shallow. Since the surface pressure depends on the depth of the cold, dense air, the rapid thinning of the cold air layer at the gap exit results in a large drop in pressure. In turn, this pressure gradient contributes to an acceleration of the low-level gap flow over the exit region. Since this shallowing of cold, dense air is similar to the shallowing of a layer of water as it passes out of a reservoir, we term such acceleration a hydraulic effect.

Thus, both synoptic scale and hydraulic pressure gradients tend to produce the strongest winds over the exit of a gap, not the constrictions in its mid-section. In fact, the build up of cold dense air on the windward side of the barrier produces mesoscale ridging that tends to slow the air down entering the gap. As noted above, venturi or funnel effects, whereby winds accelerate in a constriction, are generally not the predominant mechanisms for mesoscale gaps with widths of kilometers to tens of kilometers. But, they can be important for smaller gaps where strong flow moves through a constriction. Some gaps are comprised of a series of constrictions and expansions. These variations are often reflected in both the depth of cold air and the wind field, with winds speeding up in and immediately downwind of the constrictions. Nevertheless, the strongest winds are nearly always in and immediately downwind of the final exit of the gap.

3.4 Gaps With Slopes or in Passes

3.4.1 Gaps With Slopes or in Passes

Until this point, the tutorial has only reviewed the nature of approximately level gaps in which elevation changes along the gap are minor, a few hundred feet at most. Such level gaps illustrate many of the key elements of gap flow and are representative of a large number of coastal gaps of interest to Navy operations. There are, of course, higher level gaps--or passes--in most mountain barriers, and many gaps are characterized by substantial height changes. Under the proper conditions, winds can accelerate down the lee side of a mountain barrier with the strength and location of these downslope winds greatly influenced by gaps or channels in the barrier. Thus, a forecaster must be aware of some of the complex interplay of gap and mountain effects that often come into play.

3.4.2 Downslope Wind Acceleration and Gaps

Under the proper conditions, air can accelerate as it passes over and down a mountain barrier. The most intense leeside acceleration events, known as downslope windstorms, can produce winds exceeding 100 knots and can cause major damage, loss of life, and dangerous conditions for aviation. Downslope windstorms are found throughout the world, with well-known examples including the damaging winds along the eastern slopes of the Colorado Front Range, the bora that descends down the terrain north of the Yugoslavian Adriatic Coast, and the Enumclaw winds of Washington State.

3.4.3 Mountain Waves

Most forecasters are familiar with mountain waves that can produce a train of wave clouds downstream of orographic barriers. Such features are atmospheric gravity waves that form when a relatively strong, stable flow approaches an orographic ridge under conditions in which the wave energy is trapped, so that it is confined to the lower atmosphere.

Under the proper conditions, mountain lee waves can amplify so that the flow descends and accelerates down the lee slopes of the barrier. In general, such downslope acceleration is most evident when strong flow approaches a barrier and either a stable layer (for example, an inversion or isothermal level) or a critical level (where the wind reverses direction) is near or just above the crest of the barrier.

3.4.4 Bora Wind

A closely related feature is the bora, in which cold, dense air deepens sufficiently on the windward side of a barrier so that it passes over the crest and then accelerates down the lee slopes. The detailed physics of downslope windstorms are complex and will be reviewed in more detail in another mesoscale module.

3.4.5 Stampede Gap

There is often a profound interaction between gaps and downslope windstorms, with features of both being evident at the same time. For example, a gap in the barrier can preferentially allow cold air to push across certain sections of a ridge, and then to accelerate down the slopes towards the lowlands. Such a phenomenon is clearly evident over the Cascades in Washington, where a mesoscale gap area, known as Stampede Gap, facilitates the movement of air across the Cascades.

3.4.6 Enumclaw Windstorm

When high pressure builds to the east of the Cascades a substantial east-west pressure gradient across the barrier can develop. In addition, the flow can have a substantial westward component near crest level. Cold, low-level air pushes over the mountains in the mesoscale Stampede Gap. As the flow then accelerates down the western slopes of the Cascades it strikes the town of Enumclaw and adjacent communities with winds that can exceed 100 knots during the worst events. Easterly winds of 50 to 70 knots occur annually in this area. The winds are generally isolated in a swath about 15 miles wide that extends towards Puget Sound.

Flow through Stampede Gap shares many common characteristics with that at many other places. The sharp boundaries of the downslope/gap flow, and its ability to maintain integrity dozens of miles downstream of the gap, occur at locations around the world.

An interesting phenomenon can occur when air both passes over a barrier and moves through a gap in its midst. The sinking (subsiding) air on the lee side of the barrier warms adiabatically as it is compressed by higher pressure, thus causing sea level pressure to fall on the lee side. On the windward side of the barrier cold air can be "dammed up" and deepen as it is blocked by the terrain, producing a mesoscale windward pressure ridge. Thus, both lee troughing and windward ridging can increase the across-barrier pressure gradient and consequently the strengthen the gap wind though the terrain.

3.5 Other Weather

3.5.1 Gap Winds and Other Weather Elements

So far, this tutorial has stressed the importance of gaps in producing strong winds, particularly in the gap exit region. Although winds are probably the key gap wind feature in most situations, other weather parameters can be significantly affected. For example, for gaps in coastal terrain the gap wind outflow is often associated with cold, dry air from the interior that can starkly contrast with more temperate marine air over the coastal and offshore waters. Near the exit regions of such gaps low-level air temperatures can be 10 to 20ºF (or more) cooler than in adjacent regions, resulting in gap wind-associated variations in precipitation type during the cool season. Depending on the temperature structure aloft, the cooler temperatures in and just outside of the gap can produce freezing rain or snow, while outside of the gap (on the downwind side) rain would be the rule.

3.5.2 Columbia River Gorge

An excellent example of such a gap wind effect on precipitation type is the Columbia Gorge at the border of Washington and Oregon. The Cascade Mountains act as a barrier that impedes the westward movement of cold, continental air from the interior, with the Gorge acting as a near sea level gap across the mountains. During the winter, cold air from eastern Washington and Oregon moves westward though the Gorge across Portland and the northern Willamette Valley. When a Pacific weather system with attendant clouds and precipitation approaches the coast, the gap flow increases as the pressure difference across the Gorge builds. These Pacific systems are often relatively warm with high freezing levels (4000 to 8000 ft) that bring rain to most of western Oregon and Washington.

3.5.3 Effects of Cold Air

However, in the Gorge the gap flow is often associated with subfreezing air that can extend to a few thousand feet above sea level. As the rain falls into the subfreezing air it is supercooled and freezes on contact with the ground. Such freezing rain, known as the "Silver Thaw," can bring treacherous driving conditions and can cause substantial aircraft icing for planes inbound or outbound from Portland. Similar freezing rain events have been observed in and near coastal gaps at locations around the world.

3.5.4 Effects of Dry Air

Outflow through coastal mountains is often relatively dry, since the source region is the continental interior. Such drying is accentuated if the gap flow descends from a high interior plateau region. Low-humidity gap flow air can play an important role in initiating or maintaining snow in marginal situations, since evaporation of precipitation falling from above can be an important cooling mechanism. During the summer and fall months, such dry gap flow can be associated with critical fire hazards since both strong winds and low humidities are present. A well-known example are the strong winds (such as the Santa Anas) that descend down the western slopes of the California coastal mountains. Such descending winds are generally most intense in the gaps, where they can fan wildfires and can make driving large trucks difficult.

3.5.5 Stability and Clouds

The temperature effects associated with gap flow tend to decrease rapidly away from the gap, particularly when the cold, gap flow passes over warm water. As evidenced by growing cumulus elements, the movement of cold air over warm water produces considerable vertical instability that substantially warms the lower atmosphere within approximately 100 miles of the gap.

4. Diagnosis/Prediction

4.0.1 Diagnosing and Predicting Gap Wind Flow

As noted previously, winds in nearly level gaps are closely related to the surface pressure gradient (or pressure difference) across the gap. Wind direction is relatively easy--the winds will tend to blow parallel to the gap axis and flow from high to low pressure, with the wind speed roughly proportional to the pressure difference and the strongest winds in the gap exit region. Although these ideas are useful, one must go further to quantitatively estimate or predict gap wind speeds. This section will describe approaches of varying sophistication for estimating gap wind speeds.

4.1 Simple Diagnostic Models

4.1.1 Simple Diagnostic Relationships

The simplest relationship for gap wind flow is a form of the Bernoulli equation, which is derived assuming frictionless, steady-state flow in a constant elevation channel or gap. This equation relates the acceleration down a gap to the pressure difference down the gap. Using this equation, one can calculate wind speed differences across a gap for various pressure differences.

For example, let's assume a wind speed at the gap entrance of 5 meters per second, or about 10 knots. Then let's assume an entrance pressure of 1010mb and an exit pressure of 1000mb. Last we assign a value of 1.2 kg per cubic meter to the density of air.

The pressure drop through the gap then computes to 10 mb.

For the equation to work we need consistent units, so we need to translate this into Pascals. 10mb is the same as 10hPa. To convert from hPa to Pascals, we multiply by 100, so 10 millibars is equivalent to 1000 Pascals.

Then we just do the math:









When we solve the equation, we see that the wind will accelerate to 41 meters per second at the gap exit. That's about 80 knots!

4.1.2 Windspeed at Any Point Along a Gap

An alternative form of the same equation gives the wind at any point in the gap if you know the starting wind speed along the gap and pressure difference from the gap entrance to the point of interest.

4.1.3 Windspeed with Friction

The simple Bernoulli equation usually overestimates the wind acceleration in gaps. A major source of the discrepancy between the observed and calculated wind speeds arises from friction. There are, in fact, two types of drag important in gap flows: (1) surface friction due to the roughness of surface features and (2) drag due to mixing at the upper boundary of the gap wind flow. A better estimate of wind speed can be calculated by adding a term that accounts for friction or drag.

An illustration of the usefulness of such simple Bernoulli equation approaches is found in this graph. The red line shows the wind speeds down the Strait of Juan de Fuca measured by the NOAA P3 research aircraft by Overland and Walter on 21 February 1980. The blue line presents the wind speed calculated using the observed along-gap pressure gradient and the simple Bernoulli equation without friction. Clearly, without friction the gap winds are overestimated. In contrast, using equation 2 with friction (purple line) produces a wind speed estimate that is close to reality.

4.1.4 Gap Wind Calculator

We have built a simple calculator that computes wind speed at different points down the length of a gap. This calculator computes the wind speed with and without friction. It also lets you vary other initial conditions. Try plotting the gap winds with the default values, then type in new values and see how the wind speed profile changes. After trying this, answer the following questions.

1. Using a linear SLP profile, does doubling the pressure gradient double the resulting wind speed?
Discussion:
No, doubling the pressure gradient will not double the wind speed. For the wind speed without friction, we use the Bernoulli equation discussed earlier. In that equation, the wind speed term was squared, but the pressure difference term was not. Therefore, in order to double the wind speed, you would need to increase the pressure difference by 2 squared, or by a factor of 4, in order to double the wind speed. Go ahead and try it on the calculator.
For the wind speed with friction, there are other variables to consider as well, including the surface roughness and the length of the gap. In this case, it's hard to make general rules, but doubling the pressure difference by itself should never double the wind speed.

2. All else being equal, will you see higher wind speeds for a gap over hilly forested terrain or for a gap over the ocean? Why?
Discussion:
You should observe higher wind speeds over the ocean because it exerts much less drag than forested terrain.

3. How does changing the length of the gap change the resulting wind at the gap exit?
Discussion:
The simple Bernoulli equation that we use to calculate wind speed without friction depends only on the pressure difference, not the length of the gap. This is because it assumes no friction, so the wind is free to accelerate. As a result, the wind speed stays the same, no matter what the length of the gap.
When friction is considered, however, the length of the gap plays a very important role. In that calculation the friction essentially acts to decrease the pressure gradient force. Not only that, but the friction effect grows exponentially with distance. That is why the wind velocity appears to reach maximum for longer gaps.

In practice, such Bernoulli equations can be used as operational diagnostic or forecasting tools by using observed pressure gradients or by applying pressure gradients produced by synoptic scale numerical models such as NOGAPS.

4.2 High-Res Models

4.2.1 High-Resolution Numerical Models

As we will see, high-resolution mesoscale models, such as COAMPS or MM5, are capable of realistically diagnosing and predicting gap wind flows--IF they have sufficient horizontal and vertical resolution AND the large scale conditions are being well handled. Thus, to evaluate the usefulness of mesoscale model forecasts in a gap flow situation a forecaster must ask at least two key questions:

1. Is the synoptic scale flow being well forecast?

2. Does the model have sufficient horizontal and vertical resolution?

4.2.2 Is the Synoptic Scale Flow Being Well Forecast?

Whenever you contemplate using a high-resolution mesoscale model forecast, be it for gap wind flow or anything else, you must evaluate the realism of the model's synoptic scale forecast. A high-resolution model is like a high-powered rifle--when aimed at the right direction it can be very accurate, but when aimed at the wrong direction it is nearly useless, except for its educational value. In fact, poor mesoscale forecasts are worse than useless if impressive detail instills overconfidence in an incorrect solution. An accurate synoptic-scale forecast provides the model with the correct aim, and with sufficient resolution and correct model physics, it may well be able to correctly fill in the local details. In short, as a first step to using a high-resolution forecast of gap flow, one must verify the accuracy of the larger scale forecast, both in terms of structure and timing. If there is only a timing error, you can attempt to make the necessary time displacements so that the model's mesoscale detail aids your forecast tasks.

4.2.3 Does the Model Have Sufficient Horizontal and Vertical Resolution?

Relatively narrow gaps require high resolution to have a chance of reasonably simulating gap flow. For example, a mesoscale model with 10-km grid spacing is completely inadequate for forecasting the flow in a 10-km gap. A rough rule of thumb is that it takes at least four grid points to even grossly describe a wave-like disturbance. Thus, to simulate a 10-km-wide gap a forecast model would should have at least 2.5-km grid spacing, with higher resolution being advisable. Vertical resolution is also important, particularly if one wants to resolve the often sharp upper boundary of the gap flow and the mixing that can occur across this interface. Previous research simulations suggest that 35 to 40 levels are typically required for adequate gap flow forecasts, with approximately fifteen of those levels below 850mb.

4.3 The Columbia River Gorge

4.3.1 The Columbia River Gorge MM5 Simulation

To gain some appreciation for the resolution needs in forecasting gap wind flows, let's examine gap flow in the Columbia River Gorge. The Gorge is approximately 10 km wide. Consider the results of an MM5 simulation of easterly Gorge flow at 36, 12, 4, 1.33, and 0.44-km grid spacing using the domains shown below. Only the region around the Gorge is displayed to highlight the effects of resolution. The comparison will be limited to hour 21 of the forecast, a time at which the winds at Troutdale, in the exit region of the Gorge, were sustained at approximately 20 knots with gusts of 30 to 35 knots. The winds over the water in the central Gorge were probably stronger. The MM5 wind should be compared to the sustained winds, because short-term gustiness is not modeled at the resolutions used in these simulations.

4.3.2 36-km MM5

Using 36-km grid spacing, no real gap exists in the model terrain, but rather a saddle-like feature is apparent. Only a minor acceleration to 10 knots is apparent and there is no evidence of cold air in the Gorge.

4.3.3 12-km MM5

At 12-km resolution, the terrain in the Gorge is lower, but is still better described as a saddle rather than a gap. Isolated pockets of cold air in eastern Washington are now apparent and winds in the Gorge have increased to 15 knots.

4.3.4 4-km MM5

Using 4-km grid spacing produces obvious benefits: cold air east of the mountains is far more extensive and pockets of lower temperatures are apparent even within the Gorge. A gap-like channel through the Cascades is visible, although the sharpness of the Gorge walls is absent.

4.3.5 1.33-km MM5

A dramatic improvement is seen at 1.33 kilometer grid spacing. The Gorge channel is clearly evident, including the steeper slopes on the southern side. Within the Gorge, a continuous river of cold air flowing from east of the Cascades to Portland is apparent. Winds now reach 20 knots in the exit region of the Gorge.

4.3.6 0.44-km MM5

Finally, a smaller domain, run with 0.44-km horizontal grid spacing, realistically defines the Gorge terrain and the extent of the cold air. Winds have increased to a maximum of approximately 30 knots in the Gorge exit region.

So we see that mesoscale models can simulate gap flows, given an appropriate resolution. In this case the model required a grid spacing of 1 to 2 kilometers to accurately capture the details of flow within the 10-kilometer wide Columbia River Gorge. At a typical operational resolution of 12 kilometers, the model shows a modest wind acceleration though the gap. Thus, forecasting the gap flow requires recognition of the conditions that lead to strong winds combined with a local climatology.

4.4 Stampede Gap winds

4.4.1 Stampede Gap winds MM5 simulation

High-resolution models also appear to be quite capable of handling downslope windstorms and their interactions with gaps in mountain barriers. As an example, consider a mesoscale forecast using the MM5 of a gap/downslope windstorm that stuck Enumclaw and nearby areas on the western side of the Washington Cascades. This figure shows two domains (9- and 3-km grid spacing), which were nested within a far larger 27-km domain. Note that there are two passes in the Cascades, with Stampede Gap (the southern one) being the most significant.

This figure shows the 3-km wind and sea-level pressure fields for a 24-hour forecast. Two swaths of strong winds are noticeable, each downwind of a major mesoscale pass in the barrier. The strongest winds are not within the passes at high elevations, but downwind along the lower slopes as a result of the downslope acceleration of the subsiding flow. Both the distribution of the winds and their magnitude compare well with surface observations.

A vertical cross section through the Stampede Pass area shows potential temperature and wind structures across the barrier. Shading indicates the strongest wind regions. Note that strongest winds occur along the final lee slopes as the air descends rapidly towards sea level.

In summary, high-resolution numerical weather prediction promises to be a crucial tool for forecasting gap wind flow. However, realistic local forecasts only occur when the synoptic scale flow is well predicted. The conclusions about the effects of resolution shown above should be applicable to other mesoscale models (such as COAMPS) and to other regions around the world.

5. Island Wakes

5.0.1 Gap Winds Without Pressure Gradient Acceleration: Wakes vs. Open Regions

A major aspect of the gap flow discussion and examples provided above is the acceleration of wind within gaps, be it a level gap or one with some slope. Gap flow acceleration, usually associated with a pressure gradient along the axis of the gap, can produce very strong winds of importance to both maritime and aviation operations. However, any complete discussion of gap wind flows should include a different type of gap flow, one in which gap accelerations are minor. Such gap winds are often associated with a chain of mountainous islands separated by substantial stretches of open ocean. Downstream of the islands the winds are usually weak for tens to several hundred miles in what is known as a wake. Essentially, the blocking effects of the mountains and greater surface drag over terrain (compared to water) results in weak winds downstream of the islands. In contrast, winds remain nearly unchanged and strong in the gap regions.

An excellent example of island-produced gap winds and wakes is found downwind of Japan. This figure shows surface winds downwind of Japan derived from the TRMM microwave imager (TMI), along with the topography. Geographically, Japan is a chain of volcanoes, which includes high mountains such as Mount Fuji, separated by substantial intervening gaps. Strong winds are found downstream of these gaps, with lesser winds (wakes) downstream of major mountainous areas.

An even more dramatic and instructive example is found in this figure, which shows northwest winds approaching the Alaska Peninsula. Much like Japan, the Alaska Peninsula is composed of a chain of volcanoes separated by low-lying gaps. To the north of the peninsula, on the windward side, the winds are relatively uniform and strong across a wide region. On the lee, or southern, side the wind varies greatly with very light wind in the wakes downwind of high terrain and strong winds downwind of gap areas. The wakes extend from tens of miles to as much as one hundred miles downstream of the terrain.

6. Additional Examples

6.0.1 Additional Examples of Gap Winds Around the World

Gap winds occur at localities all around the world. In this section we shall review some well-documented examples. To continue, click a locality on the map, or choose from the menu on the left.

6.1 Strait of Juan de Fuca

6.1.1 A Detailed Case Study: The Strait of Juan de Fuca of Washington State

One of the most intensively studied sea-level gaps is the Strait of Juan de Fuca, which provides a sea level passage from the Pacific Ocean to Puget Sound. A number of important U.S. Navy facilities are found on the inland side of the Strait, including Whidbey Island Naval Air Station, the Bangor Submarine Base, the Everett Homeport, and the Bremerton docks. Approximately 100 kilometers long and 20 kilometers wide, the Strait of Juan de Fuca is a low-level gap between the 3000-4000 ft mountains of Vancouver Island and the even higher Olympic Mountains to the south. The Strait exit region is well known for strong easterly winds: for example, a 1931 study by Reed noted over 200 occurrences of easterly winds exceeding 36 knots during a 5-year period at Tatoosh Island at the western terminus of the Strait. With a large pressure gradient associated with inland high pressure and a trough over the offshore waters, easterly winds in the western Strait can easily reach 50 to 70 knots.

6.1.2 Surface Observations

On 9 December 1995 a special field experiment (COAST) explored the flow in the Strait by flying the NOAA P3 aircraft down the Strait during a moderate easterly wind event. An analysis using the P3 data and other observational assets is shown in this figure. In addition a sea level pressure analysis is shown. One notes there is a pressure gradient across the gap with higher pressure to the east. The wind appears to accelerate in the gap to around 35 knots in the exit region.

The NOAA P3 aircraft possesses a dual-Doppler radar that enabled the scientists on board to "paint out" the winds within the Strait. The results, shown at a number of levels, are presented in this figure. As indicated by the color shading, the strongest winds, exceeding 21 meters per second, or 42 knots, were found in the exit region at 100 meters. The gap winds are quite shallow--by 1100 meters the easterly flow had greatly weakened.

This vertical cross section of the winds down the Strait is also quite informative. Note how the easterly flow deepens down the Strait, but then collapses over the western exit. The strongest winds are found near the surface where the easterly flow had collapsed the most, from approximately the 50-km mark westward.

To evaluate how well a high-resolution mesoscale model forecast can duplicate the observed wind field in the Strait, the Penn. State/NCAR mesoscale model versions 5 (MM5) was run down to 1.33-km grid spacing. An example of the wind field from such a model run, valid at 1700 UTC 9 December 1995, is found below. As in the observations, the model output at 100 and 300 meters show the strongest winds just outside of the western exit region of the Strait, with the easterlies weakening greatly by 1100 meters.

6.2 Central America: The Tehuantepecer

6.2.1 Geography of Chivela Pass

Chivela Pass, a gap that has both important atmospheric and oceanographic effects, cuts through the Sierra Madre of Mexico. The gap is approximately 220 kilometers long, 40 kilometers wide, and has a maximum elevation of only 250 meters. It provides a path for air from the Bay of Campeche, in the southern Gulf of Mexico, to the Pacific Ocean's Gulf of Tehuantepec.

During the winter, when cold, high pressure systems move southward along the eastern slopes of the Rockies and the Sierra Madre Mountains, a large pressure gradient can build across the gap. This results in strong northerly winds, known as Tehuantepecers, immediately downstream of the Pass. Tehuantepecers can reach 20 to 40 knots, with gusts exceeding 100 knots in extreme cases. This figure illustrates such an evolution, one associated with the "Storm of the Century" cyclone of 13 March 1993. High pressure moved southward along the east of the southern Rockies and Sierra Madres, producing a large pressure gradient between the Gulf of Mexico and the eastern Pacific.

High-resolution mesoscale models appear to be capable of realistically simulating the development of the Tehuantepecer. 3- and 15-hr forecasts of the 13 March 1993 Tehuantepecer were made using the MM5 with a horizontal grid spacing of 6.67 kilometers. As shown in this figure, the northerly gap flow was forecast to reach 50 knots after pushing through Chivela Pass into the Pacific Ocean. Note how the strongest model winds were in the lee (south) of the gap over the ocean with the coldest air and highest pressure coincident with the strongest winds. The higher pressure in the center of the gap exit flow caused the strong winds to spread out in a fan-like pattern.

The Tehuantepecer and other larger scale gap flows are often apparent in satellite-based scatterometer wind measurements. Such measurements are produced by relating the wind speed and direction to the microwave radiation scattered off the sea surface.

Tehuantepecers and other strong, persistent atmospheric gap flows can have a significant influence on the surface waters of the coastal ocean. The strong winds that blow through the mountain gaps of Central America, such as Mexico's Chivela Pass, can produce sustained winds of 30 to 40 knots for 5 to 7 days. Such strong winds result in substantial upper ocean mixing that can bring cooler sub-surface water to the surface, causing sea surface cooling of 4 to 8ºC. This figure shows satellite-based scatterometer winds and sea surface temperatures during the Tehuantepecer event of 18 February 1997. Note the coincidence of lower SSTs and gap flows. Also, the strong, persistent winds from Tehuantepecers can create waves that can propagate as swell as far south as the Galapagos Island, nearly 1000 miles away.

A detailed tutorial on Tehuantepecers, including the application of TRMM and SSM/I satellite imagery for observing these strong gap winds, is available from the Naval Research Laboratory.

NRL Tutorial.

6.3 The Strait of Gibraltar

6.3.1 The Strait of Gibraltar

The Strait of Gibraltar, located at the western entrance to the Mediterranean, is frequently associated with strong gap winds that can produce dangerous seas, especially when they blow against tide and current. As shown here, the Strait represents a narrow sea-level passage about 15 kilometers wide and 55 kilometers long that is surrounded by terrain reaching several thousand feet.

The most pronounced gap wind though the Strait is from the east and is known as the Levanter, which can produce winds of 20-40 knots in and to the west of the gap. The typical synoptic "set-up" is shown in the sea-level pressure analysis for 1200 UTC 25-Aug-81. High pressure is found over the western Mediterranean, with lower pressure to the west of Gibraltar. The sinking motion accompanying such anticyclonic conditions often results in the formation of an inversion a few thousand feet above the surface. Such an inversion provides a vertical stable layer or cap that contains the low-level air and results in greater topographic blocking and stronger gap flow. A large horizontal pressure gradient exists over the Strait, and winds accelerate downgradient from high to low pressure within the gap. Under such circumstances, the winds can go from near calm in the western Mediterranean to gale force strength on the Atlantic side of the Strait. It is important to stress that the strongest winds are not observed mid-Strait, as might be expected if the funnel mechanism was dominant; rather, the strongest winds are over the western Strait and immediately downwind to the west. Levanters are most frequent during the warm season from May through October.

An excellent illustration of the distribution of Gibraltar gap winds under easterly conditions is provided by this visible polar orbiting satellite image. This image was taken by the NOAA-6 satellite at 0858 UTC 25-Aug at a time with sun glint over the Gibraltar region. Sun glint is sunlight reflected off a smooth water surface. Notice the wedge shaped region of darkness over and to the west of the Strait. This darkening is caused by the strong Gibraltar gap winds roughening the surface, greatly lessening the amount of light reflected back to the satellite. The darkest colors, and thus strongest winds, are found to the west of the Strait, and extend approximately one hundred kilometers to the west. This image also highlights why Gibraltar possesses such exceptional gap flows: the terrain surrounding the western Mediterranean forms a topographic bowl with only one low-level exit--the Strait of Gibraltar.

6.4 Hinlopenstretet

6.4.1 The Hinlopenstretet Strait near Spitsbergen

A high-latitude example of a gap flow is found in the strait between Spitsbergen and Nordaustlandet, named Hinlopenstretet. Sandvik and Furevik studied gap flows through this strait under southeasterly conditions. In their study they used both synthetic aperture radar imagery, which provides surface winds, and high-resolution MM5 simulations.

Using a 6-km grid spacing in the model, the trajectories over an 18-hour forecast (starting 1200 UTC 14-Aug-1996), show the flow being deflected around the substantial terrain on Spitsbergen and Norauslandet. A large portion of that flow passes through the Hinlopenstretet.

Another figure shows the wind vectors and wind speeds for the same time. Upstream of the Strait the winds are quite weak, ranging up to 4 meters per second, or 8 knots. Wind speeds accelerate through the gap and reach their maximum in the gap exit region. Strong winds extend downstream for tens of kilometers in a relatively narrow swath.

Some insight into the nature of the acceleration is provided by a vertical cross section of wind speed along the gap. Note how the depth of the approaching flow, indicated by the stronger winds, collapses and strengthens down the gap. The strongest winds occur at a height of approximately 150 meters in the gap exit region.

7. Summary

Some key ideas that you should remember:

1. In mesoscale gaps in terrain, the winds generally flow from high to low pressure and thus are highly non-geostrophic.

2. The most important mechanism in producing strong gap flows is the acceleration of air as it moves from high to low pressure. Thus, the strength of gap flows generally are proportional to the pressure gradient (or difference) across the gap. Venturi or funnel effects are generally not the dominant mechanism in mesoscale gaps, but can be important locally.

3. The pressure differences across a gap have two main origins: (1) large-scale or synoptic pressure gradients such as when an anticyclone is on one side and low-pressure center approaches the other, and (2) changes in the depth of low-level cold air across the gap.

4. For a narrow chain of mountains, or mountainous islands, a different type of gap flow is possible. Downstream of mountains or islands, we find weak winds in the wake of terrain. Downstream of the gaps, we find stronger winds, similar to those on the upstream side of the island or mountain chain.

5. The strongest gap winds are typically in the gap exit region.

6. Simple dynamical relationships, such as the balance between the pressure gradient force, drag, and acceleration, are often quite good in relating the gap wind speeds to pressure gradients.

7. High-resolution numerical models are valuable tools for forecasting gap wind flow, if the synoptic scale forecasts are realistic.

Links:

NASA Winds Homepage: http://winds.jpl.nasa.gov/

Ocean Surface Winds: http://manati.orbit.nesdis.noaa.gov/doc/oceanwinds1.html

Near real-time QuikSCAT wind products (Ocean Surface Winds Derived from the SeaWinds Scatterometer): http://manati.orbit.nesdis.noaa.gov/quikscat/

Near real-time SSM/I wind products (Ocean Surface Winds Derived from the SSM/I Radiometer(s)): http://manati.orbit.nesdis.noaa.gov/doc/ssmiwinds.html

NRL Regional Windspeed Composites: http://www.nrlmry.navy.mil/sat-bin/composite2.cgi

TRMM Homepage: http://trmm.gsfc.nasa.gov/

NRL Tutorials:

Mediterranean Mistral
Tehuantepecer
Vietnam Gap Wind

References

Bendall, A. A., 1982: Low-level flow through the Strait of Gibraltar. The Meteorological Magazine, Vol. 111, pp. 149-153

Colle, B. A., C. F. Mass, 2000: High-Resolution Observations and Numerical Simulations of Easterly Gap Flow through the Strait of Juan de Fuca on 9-10 December 1995. Monthly Weather Review: Vol. 128, pp. 2398-2422.

Colle, B. A., C. F. Mass, 1998: Windstorms along the Western Side of the Washington Cascade Mountains. Part I: A High-Resolution Observational and Modeling Study of the 12 February 1995 Event. Monthly Weather Review: Vol. 126, pp. 28-52.

Colle, B. A., C. F. Mass, 1998: Windstorms along the Western Side of the Washington Cascade Mountains. Part II: Characteristics of Past Events and Three-Dimensional Idealized Simulations. Monthly Weather Review: Vol. 126, pp. 53-71.

Dorman, C. E., R. C. Beardsley, and R. Limeburner, 1995: Winds in the Strait of Gibraltar. Quarterly Journal of the Royal Meteorological Society: Vol. 121, pp. 1903-1921

Overland, J. E., and B. A. Walter, 1981: Gap winds in the Strait of Juan de Fuca, Monthly Weather Review: Vol. 109, pp. 2221-2233.

Sandvik, A. D. and B. G. Furevik, 2002: Case study of a coastal jet at Spitsbergen-comparision of SAR- and model-estimated wind. Monthly Weather Review: Vol. 130, pp. 1040-1051

Schultz, D.M., W.E. Bracken, L.F. Bosart, G.J. Hakim, M.A. Bedrick, M.J. Dickinson, and K.R. Tyle, 1997: The 1993 Superstorm cold surge: Frontal structure, gap flow, and tropical impact. Monthly Weather Review: Vol. 125, pp. 5-39.

Scorer, R.S., 1952: Mountain-gap winds; a study of the surface wind in Gibraltar. Quarterly Journal of the Royal Meteorological Society: Vol. 78, pp. 53-59

Sharp, J., 2002: The Mesoscale Meteorology of the Columbia River Gorge. Master's Thesis, University of Washington, 248 pp.

Sharp, J. and C. F. Mass, 2002: Columbia Gorge Gap Flow: Insights from Observational Analysis and Ultra-High Resolution Simulation. Bulletin of the American Meteorological Society: Vol. 83, pp. 1757-1762.

Steenburgh, W. J., D. M. Schultz, B. A. Colle, 1998: The Structure and Evolution of Gap Outflow over the Gulf of Tehuantepec, Mexico. Monthly Weather Review: Vol. 126, pp. 2673-2691.