In a coastally trapped wind reversal, it is the wind reversal the actually defines the event. The clouds are not always present, or may have a much more complex structure than the disturbance itself. Therefore, we refer to these phenomena as coastally trapped wind reversals or coastally trapped disturbances, rather than stratus surges, because the stratus is a secondary feature.

These disturbances propagate northward faster than the wind, which suggests that they are not simply an advective process, but rather an internal, wave-like disturbance trapped against the coast (in other words, a Kelvin wave). The disturbance occurs within the inversion layer, above the surface.

High pressure in the cloudy region is due to an increased depth of the boundary layer or, more precisely, the inversion thickness.

Coastally trapped wind reversals are important in the coastal region primarily due to the rapid change from clear to cloudy conditions and the rapid change in wind direction.

These plots are a composite of several events, taken at the start of the event, centered on Buoy 13, near San Francisco. They illustrate the key aspects of the synoptic-scale structure that result in coastal wind reversals.

At sea level, we see the subtropical high pushed north of its climatological location into the Pacific Northwest, north of California. This is very characteristic of these types of events. Also note the warm thermal trough extending out toward the coast in an east-west direction. Normally this thermal trough is associated with the warm temperatures in the interior of California. The thermal trough tends to weaken the along-coast pressure gradient, or maybe even reverse it, so that we get higher pressures to the south and lower pressures to the north.

At 850 mb we see a structure very similar to what we see at the surface, with ridging into the pacific northwest, and an axis of lower pressure extending out to the coast in southern California.

Associated with this intrusion of the ridge, we would expect to see northeasterly winds, and indeed that is the case, as shown here. So we have offshore flow just above the surface over northern and central California. This offshore flow is critical for the set up of coastally trapped wind reversals.

First, this offshore flow leads to clear conditions along the coast, which is atypical in and of itself.

Then, as this offshore wind comes across the mountains, we see adiabatic warming due to downslope flow in the lee of the coastal mountains.

At 850 mb, we also see an extension of warm temperatures extending from the desert southwest out across the coast and over the ocean.

At 500 mb, we see the ridge axis just off the coast. So we have the surface high under the upper-level ridge axis. The extension of high pressure into the northwest is associated with increased subsidence downstream from the 500mb ridge axis. This sets up the low level offshore flow at 850 mb, which in turns sets up the coastally trapped wind reversal.

The evolution of the conditions at the surface for our case, prior to the wind reversal and after, follow a pattern similar to that seen in many cases.

At 12Z on the 21^st, we have a thermal trough up through central California, extending out toward the coast, with the subtropical high off to the southwest and weak ridging into the Pacific Northwest. This pattern is typical of the climatological summertime conditions.

By 12Z on the 22^nd, just prior to when the disturbance initiates, the trough extends offshore, the subtropical high has moved north, and the ridge is intruding into the Pacific Northwest.

At this point, even on the Eta model analysis, we can see a reversal of the pressure grad along the coast, with a low to the north near Monterey and a high to south in the California Bight. This gradient will favor southerly flow south of Monterey and northerly flow to the north of Monterey.

12 hours later, that reversal in the pressure gradient becomes more pronounced, with a well-defined nose of low pressure extending offshore from central California. Resulting ageostrophic winds will have a southerly direction from southern California to central California.

Offshore, we continue to see the subtropical high very far to the north with prominent ridging into the Pacific Northwest.

Continuing into the disturbance, by this time (12Z on 23 May) the stratus has propagated north to about Cape Mendocino, though it doesn't show up in the Eta model. The trough axis extends toward the coast up into southern Oregon, and the subtropical high has pushed into southern Canada.

As the disturbance continues up the coast this day, the same pattern holds with ridging into the Pacific Northwest and a thermal trough extending offshore in northern California or southern Oregon. However, these features are starting to decay, and pressure gradients are starting to weaken.

At 1200Z on the 24^th, we see the synoptic structure decaying. The thermal low has weakened and moved very far north. The ridging is also weakening as the subtropical high is starting to rebuild to the southwest.

At the end of the sequence, we see the thermal low breaking down and the subtropical high retreating to the southwest. This sequence of the subtropical high building northward and into the Pacific Northwest followed by subsequent decay and retreat to the southwest is found in sea-level pressure evolution in composites of many cases.

At 850 mb we see an evolution where initially coast parallel winds turn to become directed offshore to initiate a wind reversal.

At 1200Z on the 21st, prior to initiation of the disturbance, we see the subtropical high offshore to the south with the ridge axis extending into the Pacific Northwest. Winds are northerly across northern California, directed just slightly offshore and starting to clear out the stratus to set the stage for the coastally trapped disturbance.

At 1200Z on the 22^nd, the winds swing around more northeasterly as the subtropical high at 850 mb as well as the surface builds northward into the Pacific Northwest. This gives us stronger offshore flow, particularly across the northern and central California coast, with relatively warm temperatures across this region at 850 mb, suggesting warming with subsidence across the coastal mountain ranges.

We also see troughing at 850 mb from Cape Mendocino southeast across interior California, which is coincident with the warmest temperatures.

As the disturbance progresses through the day on 22 May, we see the thermal low strengthen and extend offshore. The 850mb ridge continues to build up into Canada with strong offshore flow all the way north into Washington, and the disturbance progresses up the coast during this time.

By 1200Z on the 23^rd the subtropical high is in retreat with the approach of a storm to the northwest. Offshore flow weakens during this time as the ridge in the Pacific Northwest weakens and the thermal low continues to move offshore.

By 1200Z on the 24^th, winds have turned to the northwest along the coast as the subtropical high retreats to the southwest and there is a return to typical climatological conditions with even more pronounced with northwesterly winds all along the coast by later on the 24^th.

This evolution is similar to synoptic composites of other wind reversals and shows that the wind reversal is directly associated with the period of pronounced offshore flow at 850 mb. The wind reversal initiates as this offshore flow develops to the north and then it stops and decays as the offshore flow returns to typical northwesterly flow.

Now let's look at the evolution of the 500mb heights and vorticity starting on 21 May. The 500mb ridge axis extends northward along the coast, but is located offshore. This is consistent with the composite evolution of many wind reversals 24-36 hours prior to the event. In this particular case, we also see a well defined closed low aloft over southern California. This results in northerly flow aloft over California at this time.

These conditions lead to subsidence and associated drying at low levels, which tends to clear out the coastal status in this region.

24 hours later, at 1200Z on May 22, the 500mb ridge moves into the Pacific Northwest. The disturbance at low levels has started by this time, which started as the closed low migrated slightly to the northeast and northeasterly flow aloft developed.

At 1200Z on the 23^rd, the 500mb ridge axis migrates down the coast, weakens, and becomes ill-defined, while the closed low migrates eastward. The low-level disturbance weakens just beyond this time period.

By 1200Z on the 24^th we're starting to see the approach of a deeper trough offshore. By now the disturbance is completely over at the surface. Satellite imagery shows the stratus has made it about as far north as Cape Mendocino.

Overall, we see a progression aloft that starts with the 500mb ridge initially offshore. Then the ridge builds and moves into the Pacific Northwest as the disturbance is initiated in the low levels. Subsequently, the ridge weakens and moves east as the disturbance proceeds in the low levels. This upper-level evolution is very consistent with the synoptic evolution determined from composites of many cases.

Reviewing the synoptic scale surface feature, there are a couple of features that are important to note:

The disturbance essentially represents down-gradient flow from high pressure in southern California to low pressure to the north.

This ageostrophic flow northward along the coast is due to terrain blocking of surface winds by coastal mountains, in which the onshore flow is blocked by topography, and consequently turns toward lower pressure, which is located to the north.

Note that this ageostrophic response tells us why the wind reverses, but not why the disturbance propagates.

The wind reversal propagates northward much faster than the thermal trough, suggesting that there is a forced response associated with the disturbance. This forced response perturbs the marine boundary layer, lifting the top of the capping inversion and propagating northward as a Kelvin wave.

The synoptic scale is forcing a coastal low at a particular point, and even though that coastal low may drift northward, it tends to remain relatively fixed over the time period in which the disturbance propagates rapidly northward.

Here we see a satellite image from the last case taken at 1800Z on May 22^nd. The image is overlain with sea level pressure and 850mb winds. At this time the disturbance was in full swing and the stratus had propagated to a point just north of San Francisco.

The reversed pressure gradient extends from the California bight region up the extension of the trough over the coast. This coincides with northern extent of the stratus, which marks the leading edge of the disturbance at the surface.

We see the offshore flow at 850 mb over northern California, which leads to adiabatic warming over the coast, tending to lower sea level pressure out ahead of the disturbance.

There are cases where this synoptic-scale setup persists, just like this, and we see no northward movement of the stratus. However, in this case the stratus continues to propagate northward along the coast, much faster than the synoptic features.

If we look the mesoscale sea level pressure pattern associated with one of these disturbances, we can infer relationship between the synoptic scale evolution and the structure of the perturbed marine boundary layer thickness that produces the coastal high pressure associated with the disturbance.

Prior to the disturbance, we see low pressure in the California Bight region and increasing pressure north along the coast. This is consistent with the boundary layer inversion sloping down toward the coast, which is the climatological structure.

By 6 hours later, we see an axis of lower pressure extending offshore just north of Point Conception. This region of low pressure occurs in the region of strongest offshore flow above the boundary layer and marks the initiation of the disturbance.

As time progresses, we see the axis of low pressure along the coast has moved north to cross the coast near Monterey Bay. However, the offshore low remains relatively fixed at a location just north of the California Bight.

An axis of high pressure lies over Point Conception. This axis of high pressure mirrors what we see in satellite imagery and marks the disturbance and the stratus surge. The high pressure represents the deeper marine boundary layer associated with the disturbance.

Another 6 hours later, we see the nose of high pressure right along the coast and a closed low offshore. The region of high pressure is propagating up the coast and is associated with southerly winds and stratus.

The low pressure is in about the same position as it was initially forced, just north of Point Conception; it hasn't really moved. This lack of movement reflects the tendency for the synoptic scale to evolve slowly. The strongest offshore flow is occurring in this region, forcing the low to form. The disturbance propagates northward under this offshore flow, leaving behind the coastal low that initiated the disturbance.

The formation of the mesoscale coastal low that initiates the disturbance is due to localized warming along the coast. This can be seen in this comparison of soundings from Monterey. We see relatively cool temperatures in the initial sounding, to the left, and warming in the lower atmosphere 24 hours later, mostly below 850 mb.

But what causes the warming? If we look at winds, at the initial time we see relatively strong, low-level northeasterly winds. These winds cause the temperatures to rise in the low levels, which forms the low that initiates the disturbance. The region of strongest offshore flow marks the region of greatest warming and largest pressure falls along the coast. By the time the disturbance is under way, shown on the right, the forcing has weakened and we no longer see the northeasterly flow.

In this particular case, the low level warming is due almost entirely to advection of warm air offshore, though in some cases, there is a mix of advection and adiabatic warming.

Horizontal cross sections at a height of 300 m above sea level. The coastal mountains lie along the right side of each cross section. Contour lines show potential temperature (contour interval = 2 K). Maximum wind vectors are approximately 14 m/s. The position of the imposed offshore flow is indicated on the far right.

The tendency for offshore flow to produce localized warming and a coastal low are characteristic synoptic-scale aspects of a disturbance. What we want to try to understand is how those synoptic features actually force this low pressure to occur along the coast, which then initiates a propagating wind reversal.

Shown here is a simplified model, which starts with strictly along-coast flow at inception. We then perturb this flow by blowing winds across the coast, as shown to the far right, and let the atmosphere evolve.

At one day into the simulation (far left) we start to develop a region of low pressure right at the axis of the strongest offshore flow. Winds are deflected away from the coast upstream of the low, and then around and toward the coast downstream of the low.

Coastal mountains block the winds flowing toward the coast. The blocked winds slow down, then respond ageostrophically, turning north toward the low pressure. We see this response about 1.5 days into the simulation with the narrow zone of southerly flow right along the mountains. We also see that the low pressure is still located right at the point of strongest offshore flow at this time.

At day 2, as the disturbance continues to evolve, and we see the axis of low pressure building and extending offshore and the wind reversal propagating north right along the coast, underneath the region of strongest offshore flow.

The pressure distribution looks very much like what we see in the observations with a narrow zone of high pressure right along the coast and a pressure gradient falling off to the north and west. This pattern reflects our forced Kelvin wave, with this feature now propagating up the coast.

This graphic summarizes the model result and shows the evolution of a coastally trapped wind reversal.

At our start time, t₀, we see the climatologic norm with high pressure offshore, low pressure inland, and strong along-coast flow out of the north. The corresponding cross section shows a marine layer inversion sloping down into the coastal mountains.

At the next time step, t₁, we initiate the zone of offshore flow. That offshore flow sets up the coastal low. The northerly surface flow gets perturbed around the low and turned in toward the coast around the south end of the low.

At the last time step, t₂, the coastal mountains block the onshore flow, so it turns north toward the low, setting up the disturbance that propagates northward. The coastal low stays under the region of strongest offshore flow while the wind reversal and the coastal high propagate northward. The bottom cross section shows the inversion sloping down to the coastal low, then sloping back up to the coastal mountains, which reflects our coastally trapped disturbance.

Starting with synoptically-forced, offshore flow at 850 mb, we develop a region of low pressure right at the axis of the strongest offshore flow. Winds are deflected away from the coast upstream of the low, and then around and toward the coast downstream of the low.

Coastal mountains block the winds flowing toward the coast. The blocked winds slow down, then respond ageostrophically, turning north toward the low pressure. The low pressure is still located right at the point of strongest offshore flow.

As the disturbance continues to evolve, we see the axis of low pressure building and extending offshore and the wind reversal propagating north right along the coast, underneath the region of strongest offshore flow. The pressure distribution has a narrow zone of high pressure right along the coast and a pressure gradient falling off to the north and west. This pattern reflects our forced Kelvin wave, with this feature now propagating up the coast.

To forecast a coastally trapped wind reversal, there are a few things to note and watch for:

1. For the synoptic set-up, watch for the intrusion of the subtropical high into the Pacific Northwest (for U.S. West Coast events), the extension of the thermal low over the coast, and the development of offshore directed winds at about the 850mb level of the atmosphere.
2. Watch for where strongest offshore flow is. That will initiate the coastal low. The synoptic scale models should give reasonably good guidance on this location.
3. Once the position of the coastal low is known, we should see the stratus surge develop to the south. Monitor buoy pressures to see if the pressure gradient reverses. That stratus surge may then propagate north, past the coastal low.

How well do mesoscale models predict coastally trapped wind reversals?

The synoptic scale forcing is handled quite well by mesoscale models. These models typically show us the inland intrusion of the subtropical high, the offshore extension of the thermal low, and the development of offshore winds.

In addition, mesoscale models can sometimes depict coastal wind reversals and changes in the depth of the marine boundary layer. However, mesoscale models tend to overemphasize sea breezes and diurnal forcing, which can mask some of the features of coastally trapped wind reversals and make those features hard to see.

However, do not expect mesoscale models to depict the exact timing and location of this type of disturbance. The details of the wind reversal are not handled well and the clouds themselves are handled more poorly still.

Here is an example of a mesoscale model, the MM5, run for the 22 May case we looked at earlier.

This figure shows a 3-hour forecast of surface winds overlain on the satellite image. We can see that the disturbance is in full swing with stratus reaching nearly to San Francisco. The model predicted southwesterly flow in the clouds, at a time when observations indicated southeasterly flow, so the model is missing the disturbance at this point.

Three hours later, the disturbance has moved up the coast. The mesoscale model has been able to turn the winds into a more coast parallel direction, and therefore, is starting to see the disturbance. However, the model shows the wind reversal occurring somewhere south of Monterey, while the leading edge of the disturbance is well north of San Francisco.

Another 3 hours later, at 2100Z, the clouds show the disturbance continuing to propagate north. The model predicts southerly winds up into Monterey Bay and shows the disturbance propagating up the coast. However, while the model recognizes the coastally trapped wind reversal, it is still places the disturbance too far to the south. The model is missing on the timing; it started late and cannot seem to catch up.

Looking now at a loop of sea level pressure and wind from the MM5.

In the analysis, we see a weak coastal low starting to perturb the along-coast flow.

3 hours into the model run, the low pressure offshore strengthens. As a result, the model starts to turn the winds in toward the coast.

Six hours into the forecast period, the offshore low continues to build and we start to see coastal southerlies. At this time, the model progs the leading edge of the dusturbance along the northern side of Monterey Bay, near Santa Cruz. In reality, the stratus lies well north of San Francisco at this time

As the model proceeds, it propagates the disturbance up the coast. The model now locates the disturbance about half way between Monterey Bay and San Francisco.

Here at 12 hours, the model-defined disturbance extends up to about San Francisco, with low pressure offshore and high pressure in the disturbance.

The 15-hour forecast shows the disturbance continuing to propagate northward. However, it also shows a surge into interior California, which was not observed.

Overall, with mesoscale models we can see that some of the details are wrong and the timing is off. And though it isn't shown here, if we were to look at cross sections, we would find that the marine boundary layer structure is oftentimes not particularly well handled, though it can capture bulging of the MBL along the coast.

Sometimes numerical models can provide some useful guidance. They can simulate some of the features of a disturbance, but they are very sensitive to the timing of the synoptic scale forcing and to the details of the marine boundary layer structure that they start with. In some weaker cases the model may completely miss forecasting any of the salient features that typically occur.

As the synoptic conditions return to more zonal flow aloft and northwesterly flow at 850 mb, the disturbance tends to end because there is no low-level offshore flow to maintain the coastal low. Oftentimes the clouds are left in place along with a deeper marine boundary layer.

We then have climatological conditions and a deep marine boundary layer along the coast until the next trough comes through to clear things out.