Diablo Winds: California's Critical Fire Weather Pattern

As we enter the fall in northern California, one thing is one everyone’s mind. Are we going to have Diablo winds like we did in 2017? Next week marks the one-year anniversary of the deadly Wine Country Fires. Three of the Wine Country Fires (Tubbs, Redwood Valley, and Atlas) as well as the Tunnel Fire (Oakland Hills) of 1991, rank in the top twenty of California’s deadliest wildfires (CalFire, 2017). All four of those fires, with the addition of the Nuns Fire (2017) are also ranked in the top twenty most destructive fires (CalFire, 2018) in California's history.  The Tubbs Fire alone burned over 36,000 acres and caused 22 fatalities. These fires were driven by extreme winds known as Diablo Winds.

What is a Diablo Wind?

Diablo winds are offshore wind events that flow northeasterly over Northern California’s Coast Ranges, often creating extreme fire danger for the San Francisco Bay Area. Diablo winds are driven by a surface pressure gradient that forms in response to an inverted pressure trough that develops over California.

 How frequent are Diablo Winds?

A 17-year climatology of regional surface stations was used to develop a definition of Diablo wind events as well as an analysis of their spatial distribution and event frequency.  A synoptic composite of the identified events illustrate that Diablo wind events are associated with an inverted pressure trough that develops over California creating a pressure gradient from higher pressure over the interior northern Great Basin to lower pressure near the California coast. Results indicate Diablo winds affect regions throughout the San Francisco Bay Area with greater frequencies concentrated in the Coast Ranges nearest the Sacramento Valley. During the 17-year study period, the region experienced a mean annual frequency of 2.5 events with the highest frequency of Diablo wind events occurring in October when the live fuel moisture is also at a seasonal minimum leading to the most severe fire danger conditions for the San Francisco Bay Area.

The critical issue for the Bay Area is that Diablo Winds are most frequent during the fall and in particular October, when the fuel moisture content is lowest (See Figure 1 below).

Figure 1: Monthly Frequency of Diablo Winds and average Live fuel moisture content (Dashed line).

Numerical Simulations

To better understand the dynamics of Diablo Winds, Graduate student Carrie Bowers, who is wrapping up her MS thesis investigating Diablo Winds and the cause of the extreme fire spread observed during the Tubbs Fire, conducted high-resolution computer simulations of the Tubbs Fire event. Below is a cross-section of her high-resolution model simulations using WRF. You can see that the surface winds were extremely high during the event and that a unique hydraulic-jump feature formed in the lee of Mt. Helena east of Santa Rosa. This unique aspect of the wind flow classifies Diablo Winds as a downslope windstorm.

High-resolution WRF simulations of the winds during the Tubbs Fire, 8 October 2017. The times listed are Pacific Local Time and the figure represents a west-to-east cross-section starting about 5 km west of Coffey Park and ending 30 km to the east. Red colors indicate winds from the east (from the right) and blues indicate winds from the west (from the left).

Did Tropical Moisture Contribute to the Valley Fire Rapid Rate of Spread?

The wind whipped Valley Fire devastated the communities of Middletown and Hidden Valley Lake on Saturday September 12th. Now that much of the smoke has cleared, officials have had a time to tally the loses: some 1000+ homes have been destroyed, and sadly, three lives were lost to the inferno making this one of California's most destructive wildfires.  Contributing to the massive losses was the fire's rapid rate of spread, which was was quite remarkable, consuming 40,000 acres in the first 12 hours. The rapid advance begs the questions what conditions lead to this tragedy. Specifically, what caused the the strong winds that drove the fire and contributed to long distance spotting that hampered containment efforts? The answer is, in part, somewhat surprising: Tropical Moisture. 

In this post I'll make the case that tropical moisture, streaming north from the remnants of Hurricane Linda, may have played a significant and unexpected role in generating the strong winds that propelled the Valley fire through Middletown and Hidden Lake Valley. To set the stage, lets start by looking at the mid-level (500 hPa) relative humidity and geopotential height fields leading up to the day of the fire (Fig. 1).

Fig. 1. 

Two days before the fire, on 11 Sept., Hurricane Linda is apparent as a series of closed contours west of Baja California and as a extensive plume of high relative humidity air overspreading southern California (gray shading). Over the next two days Linda's circulation decayed, but the moisture plume advanced north along the California coast, overspreading the Bay Area during the day on the 12th of September. 

This surge of tropical air was mostly confined to the mid-levels of the atmosphere, overriding much drier air near the surface. For example, the radiosonde launched from Oakland International airport at 00 UTC on 12 September (Soild lines, Fig. 2) shows the high humidity air at 500 hPa, but an extremely dry layer of air near 850 hPa.

Fig. 2. Comparison of Temperature and Humidity Profiles on the morning (solid lines) and afternoon (dashed/thin lines) of 12 September 2015. The morning (evening) wind profile is shown in orange (black).

What is notable, however, is that by Saturday afternoon  (dashed lines, Fig. 2) the dry layer had significantly moistened and cooled. So too, the winds in this layer, which ultimately affected the fire, increased dramatically, reaching sustained speeds of 45 kts (~52 mph) out of the northwest. 

The cooling at 700 hPa, and the commensurate increase in winds, was caused by rain falling from clouds associated with Linda in the mid-levels and evaporating into the dry air near the surface. In fact this process was captured in compelling detail by the roof top LiDAR and webcams at SJSU (Fig. 3). The LiDAR data shows the arrival of mid-level clouds, the bases of which progressively lowered through the afternoon. Then around 20:00 UTC virga is observed, and also captured on by the webcams (inset photo). The evaporative cooling in this band of precipitation aloft causes a strong downdraft, which descended to the top of the marine layer and caused a series of gravity waves (undulations in the shading near the surface). These waves showed up in local wx observations as some peculiar fluctuations. 

Fig. 3

But the story doesn't end there. The evaporative cooling was far more extensive in a layer offshore, west of the Bay Area. Figure 4, for example, shows RUC model analyses from the morning and afternoon, including precipitation (shaded colors), 850 hPa geopotential heights (black contours), and 850 hPa winds (wind barbs). From these figures it is clear that a "meso-high" forms just offshore due to the evaporative cooling in the precipitating region (marked as an H). Since cold air is denser than warm air, the evaporative cooling generates higher pressure, a phenomenon that is well known with severe thunderstorms, but not often seen here in Central California. 

Fig. 4

The net result is a significant increase in the pressure gradient over the North Bay, and since winds are driven by pressure gradients, we see a corresponding increase in the NW flow. This burst of NW winds is particularly evident in the radar wind profiler data at Bodega Bay (Lower Panel Fig. 4), where at 00 UTC 45 Kt winds are observed just above the surface. 

These same winds were manifest over the area of the Valley Fire. Figure 5, for example, shows the time series of wind and temperature from the Kelseyville Remote Automated Weather Station (RAWS). There is a significant increase in NW winds in the afternoon of the 12th, with peak sustained winds of 25 mph and gusts to 35 mph. 35 mph winds are no laughing matter, especially when it comes to a fire moving through late season cured fine fuels. 

fig. 5. KelseyVille RAWS time series

But perhaps the biggest effect of the wind was to cause a tilted plume structure, which contributes to long distance spotting (Fig. 6). The strong winds cause what would ordinarily be an upright convective column to lean in the downwind direction, carrying burning embers far in advance of the flaming front of the fire. The resulting spot fires then drive a nonlinear fire propagation that can be extremely difficult to contain, and extremely dangerous. The fire is no longer simply moving from point a to point b, but jumping in leaps and bounds across the landscape, with individual fires merging from all directions.

Fig 6

In summary, while many factors likely contributed to the Valley Fire's rapid advance there is clear evidence that tropical moisture contributed to enhanced NW winds that propelled the fire. Specifically:

(1) Topical moisture moved north, overlaying much drier air near the surface.
(2) Evaporative cooling due to rain falling from aloft into the dry air caused a meso-scale region of high pressure.
(3) The "meso-high" increased the pressure gradient over Lake County, generating strong NW winds
(4) These winds contributed to the rapid rate of spread and long range spotting.

This case would be an interesting modeling exercise. For example, to further isolate the role of tropical moisture, we might run two simulations: one with and one without precipitation effects. Do the strong winds still develop if we turn the precipitation off? How much slower would the fire spread be in the absence of the enhanced NW winds? 

-Neil Lareau

Wildfire in Corsica

While spending time in Corsica it is easy to recognize the similarities between its wildfire hazards and those in California. It's no wonder our climate in the Golden State is Mediterranean.  The Maquis of Corsica is very similar to California's chaparral shrub ecosystem and presents a dangerous fuel type in the right conditions. 

THE BEAUTIFUL SCENERY ABOVE THE VILLAGE OF FRANCARDU, CORSICA. STEEP SLOPES COVERED IN MAQUIS SHRUB ACCENTED BY LIMESTONE CLIFFS. ©fireweatherlab.

Corsica is a beautiful place where alpine peaks meet the Mediterranean Sea in a very short distance. The landscape is characterized by very steep canyons where eruptive fire behavior can have devastating consequences.   

Bastia, Corsica where the mountains meet the sea. ©fireweatherlab

Maquis shrub meets the sea, Corsica @fireweatherlab

Extreme Fire Behavior

Extreme fire behavior is often associated with explosive nature and rapid fire spread in steep canyons. The Palasca Fire occurred in a steep canyon near Corsica's coast. The fire exhibited very extreme and explosive fire behavior and overran firefighters with a devastating outcome-two firefighters lost their lives battling the fire. 

Monument to firefighters who lost thier lives battling the Palasca Fire, 17 Sept. 2000. The monument is placed at the head of the canyon just above the beach. ©fireweatherlab

A view down the canyon where the palasca fire occurred on 17 Sept. 2000, Corsica. 

The University of Corsica has one of Europe's leading wildfire research teams. The university is located in the village of Corte where the Tavignano and Restonica rivers meet.  It is a spectacular setting and wonderful place to visit.  I was fortunate to be able to spend some time here and work with the wildfire research team. 

View of village center from the university of Corsica. @fireweatherlab

Wildfire ignition studies at University of Edinburgh

Prof. Clements had an opportunity to visit Prof. Albert Simeoni and his team at the University of Edinburgh this past month. It was great to see the fire research at the BRE Centre for Fire Safety Engineering! Edinburgh is an amazing city. 

Here is a photo of the combustion studies instrumentation in action, operated by PhD student, Jan Thomas. 

Prof. Clements (left) and Prof. Simeoni (right)  in front of the lab building at the university of Edinburgh, kings buildings. May 2015

Basic 32 Field Day

On May 22nd, researchers Neil Lareau and Matthew Lloyd attended field training hosted by the Tahoe National Forest to complete their requirements for fireline certification. They were accompanied by our newest graduate research assistant, Carrie Bowers, who served as a liaison between the “fire folks” and the researchers.

Upon arrival at the field site, all attendees were briefed on the mission for the day as well as the hazards they would be exposed to: ticks, poison oak, mountain lions, hazard trees, and each other.  After learning how to properly identify poison oak, the group set out into a field full of the hideous plant and got to work.

The first task of the day was to construct a progressive hoselay along the edge of an imaginary fire. Everyone took turns running out a length of hose and spraying water on the fire. Communication was essential and the group began to get more comfortable with each other.

After a quick lunch, the group returned to the imaginary fire where they learned they had run out of hose and would have to construct fireline with hand tools. Everyone lined out and began working through the brush and vegetation to clear a fireline down to mineral soil.

Throughout the day, small groups broke out into stations to learn how to put together a drip torch, light a fusee and sling weather.  Neil, being the resident wise guy, when asked if his group had an opportunity to sling weather, responded “I’m a meteorologist”. Our lab may not be welcome back on the Tahoe.

As the day wore on, the mock fire got out of control and spot fires forced the group to run out their escape route back to the trucks where they deployed practice fire shelters. With everyone safe, the group emerged from their shelters and participated in an After Action Review.

The Fire Weather Lab is very grateful to the Tahoe National Forest for providing training and certification for our members, managing our fireline qualifications in the Incident Qualifications and Certification System (IQCS) as well as maintaining our status in the Resource Ordering and Status System (ROSS).  

Mobile Doppler Lidar In Action

The fire lab recently tested our Halo Photonics Doppler Lidar in a new mobile configuration which allows us to record vertical velocity and aerosol backscatter while driving! The test deployments were conducted as part of a research project, CALGEM, in the southern portion of California's Central Valley. Our role in the project is to monitor the temporal and spatial evolution of the boundary-layer, which has important implications for the mixing of green house gases. While not exactly fire weather, the structure of the boundary layer plays a key role in fire behavior, and developing improved measurements will help us this summer when we deploy to big wildfires.

The Doppler lidar is mounted to the back of a Ford F-250 pickup truck (Fig. 1). In the past we've been able to deploy to a site, park the truck and then conduct scans and profiles of atmosphere. Now, with the addition of a solid state drive to the Lidar we can conduct vertical stare scans while the truck is in motion, opening up a range of new observational strategies! I'll show some examples below.  

Figure 1. California State University Mobile Atmospheric Profiling System (CSU-MAPS). F250 with Halo Photonics Doppler Lidar and Microwave Profiler.

But first, prior to the mobile deployment, we operated the Lidar in the convectional fixed location setup to monitor the growth of the convective boundary layer (CBL) on 17 April. Figure 1 shows an overview of this data. 

Figure 2. (a) Logarithmic backscatter intensity showing the variations in aerosol concentration and layering. (b) Vertical velocity (red=updrafts, blue=downdrafts). The black line with triangle markers shows the approximate height of the convective boundary layer (CBL) while the dashed line indicates the top of the aerosol layer. 

The data reveal that the convective boundary layer (CBL) depth increases throughout the day, starting at ~800 m MSL and reaching ~1500 m MSL by mid-afternoon. The CBL top is apparent as the height to which surface based thermals (bright reds and blues in the bottom panel) penetrate.  An unexpected and interesting feature on this day are gravity waves in the layer above the CBL. These waves are particularly apparent around 18:30 UTC (11:30 AM local), residing in the layer between the top of the CBL and the top of the aerosol layer (Fig. 2). It is likely that the waves result from the surface based thermals rising into the stable air aloft. As the rising plumes of heated air push up on the stable layer a wave motion is initiated. 

Figure 3.  Panels as in Fig. 2, but showing a detail of convective plumes and gravity waves. Note the up-down couplets of vertical velocity at a lower frequency and intensity in the layer above the CBL. These wave correspond to the undulations in the aerosol backscatter in the top panel. 

An added factor in the generation of the gravity waves is the presence of a shear layer at 1000-m MSL (Fig. 3 inset). Below this level (e.g., within the CBL) the wind is from the Northwest and stability is about neutral, whereas aloft it is from the Southeast and the atmosphere is stratified. The transition zone has near zero velocity and ~180 degrees of direction shear, a condition that is sometimes associated with wave trapping, reflection, or ducting. 

Now for the good stuff: After completing our primary objective of monitoring the CBL growth we conducted our first mobile deployment by driving an "up and over" transect of the Elk Hills, a small patch of complex terrain on the SW side of the valley. Figure 5 shows the driving route and truck speed, while Figure 6 shows the data that we collected.

Figure 5. Top panel shows the driving route across the Elk Hills near Bakersfield, CA. The bottom panel shows the truck speed (m/s). The speed drops to zero at stop signs. 

Figure 6. Vertical velocity as a function of time and distance across the Elk Hills. Note that the CBL is now much deeper than earlier in the day, with the deepest plumes reaching to ~1800 m MSL. The ground elevation is shown in in grey, note that the first lidar data are about 100 m AGL. 

The transect reveals that the CBL had grown appreciably in the late afternoon, likely due to destabilization of the airmass by an approaching cut-off low. It is also noteworthy that the CBL appears to be somewhat terrain following, with convective plumes above the hill crest reaching as high as 1800 m MSL whereas the CBL depth over the adjacent plain is ~1600 m MSL. The hill itself is about 200 m tall. In this single transect there is also a tendency for the most vigorous thermals to occur over the hill itself, suggesting that the elevated heating generated by the hill contributes to convergence in that region.  The resulting enhanced plumes may mix pollution to greater depths than over the flat agricultural lands. Additional transects would be required to confirm this hypothesis. 

Following Thursday's success we conducted a second mobile deployment on the afternoon of Friday 18 April. This time we did a complete west to east transect of the central valley (Fig. 7). The deployment was much longer, taking nearly an hour to complete. The data reveal the combined spatial and temporal variations in the convective updrafts and downdrafts as well as the variation in aerosol loading. For example, substantially higher aerosol concentrations were observed on the eastern third of the valley (Fig. 8).

Figure 7. Driving route on 18 April.

Figure 8. West-East transect of central valley. (a) vertical velocity, (b) Aerosol backscatter

We're extremely excited about the success of these first experimental mobile deployments. This summer we'll be extending these tests into the Sierra Nevada to examine the structure of the boundary layer over complex terrain during high fire danger days.  

Fire Front Passage from Above, FireFlux II

Follow Us on Twitter!

The lab is now on Twitter!

• 
  SJSU FireWeatherLab

  Our Mt Umunhum Weather Station in the Santa Cruz Mountains recorded a gust of 97 mph at 11:05am! What a day. #CAwx… https://t.co/1wgSacxgOu

  Mar 14, 2023, 7:54 PM