May 22 2008 Tornado Outbreak

On May 22-23, 2008, a favorable pattern for severe thunderstorms developed for the high plains, front range and eastern slopes of the Rockies. Slow moving meridional troughs have historically provided some of the more notable severe weather outbreaks for this region. Examples include  April 23 1960 and June 14-17  1965.

Several tornadic storms occurred on May 22, and a few of these produced large, long-lived tornadoes. The method of counting tornadoes has varied over the decades. Also, many more tornadoes are spotted since the advent of storms chasing. These days, some eager storm chasers are reporting every little dust-whirl that remotely looks like a tornado. Also, NWS verification has also led to an increase in tornado reports since the early 1990s. Storm chasing and NWS verification have undoutbedly led to an explosion in the number of tornado reports since the early 1990s. Many of these are the weaker tornadoes(or in some cases dust whirls) since the much larger tornadoes were often too obvious to be overlooked. To me, the longevity and quality of individual tornadoes have more meaning than the actual number of tornado touchdowns. For example, one F4, mile-wide tornado that stays on the ground for 50 miles obviously is much more significant than 50 tiny, weak tornadoes that stay on the ground for 30 seconds and do little damage. In the 1950s, individual paths of tornadoes tended to be much shorter because detailed storm surveys were usually not done. In reality, these continuous tracks were oftentimes a family of tornadoes, with breaks in between the individual tornadoes. Also, one tornado on the ground continuously for 50 miles obviously does more damage on average than a family of 10 tornadoes along the exact same path. The tornado on March 18, 1925 is officially listed an one continuous tornado for 234 miles that killed 695 people. Obviously this was a catastrophic tornado day in our history.

This severe weather episode was well forecasted by the ECMWF model. This model indicated that a deep, slow moving, meridional trough would approach the high plains on May 22-23. It also predicted a deep trough over New England, with an associated surface front behind this system through the lower midwest  into northeast Kansas, southwest Nebraska and eastern Wyoming. The ECMWF did an excellent job with these features well in advance. The 144 hr ECMWF showed a stationary front through northern Kansas and southern Nebraska, with strong upslope flow across western Nebraska and eastern Wyoming. On May 17, I expressed my thoughts about the severe weather pattern.

Towering cumulus clouds developed southeast of Denver International Airport around 16 UTC on May 22. By 1635 UTC, 45 dbz echoes were noted at 23,000 ft about 3 miles northeast of  the airport. By 1657 UTC,  a 61 dbz echo was located 14 miles north of Denver at 21,000ft. The first 50 dbz on the lowest slice occurred at 1648 UTC. The first 60 dbz echo on the lowest slice was noted 5 miles west-southwest of Hudson at 1652 UTC. By 1701 UTC there was a 62 dbz echo up to 26,000ft and 47 dbz echo up to 32,000ft. The storm was severe at this point. At 1705 UTC, the storm had 63-66 dbz echoes northwest and west of Hudson at the lowest slice.

Radar loops for the Colorado part of this tornado outbreak will be coming soon.

It is well known that thunderstorms tend to develop along the front range earlier than on the low plains. This is because capping tends to be weaker due to elevated heating, and lower moisture content of the air allows for a rapid warmup to the convective temperature by local noon. Also, the Laramie Ridge is above the traditional capping level. But convection began before local noon on this day. The first tornado report was east of Platteville at 1726 UTC. The tornado was already doing damage by 1726 UTC and was reported to be 1/2 mile wide at 1727 UTC. This tornado became very large and damaging and continued for 34 miles to west of Wellington, CO through 1812 UTC. Very large hail up to baseball size occurred along and west of the tornado path. The tornado moved generally to the north-northwest at 35 to 40 mph. The elapsed time between the first 50 dbz echo on the .5 deg slice from the Denver radar and initial tornado damage was 37 minutes!!                 

The storm weakened a little as it moved northwest of Wellington, CO, but then strengthened and accelerated as it passed east of Virginia Dale. This storm produced a 2nd tornado from 1857 UTC to 1935 UTC. There may have been a break in the damage northwest of Overlook Rd after 1922 UTC. This tornado was accompanied by quarter to golfball sized hail. It is unclear whether the tornado that hit Laramie was a 3rd tornado from the same supercell or whether the 2nd tornado continued into Laramie. The tornado in the picture looks very small. It is more probable that the tornado shrunk in size just  before entering the eastern edge of Laramie. The tornadic storm continued well northwest of Laramie. Radar indicated another possible tornado 7 miles north of Harper, WY at 2017 UTC. 

Since I have a strong interest in high elevation severe weather, and since this was a particularly rare and exceptional tornado event on the high terrain, I decided to independently document the Wyoming part of the tornado outbreak. This was accomplished using the internet white pages along with live maps.  But when I first started this task, I had no names to work with. So I ordered a phone book for Cheyenne, WY called the "Country Cowboy". I started with the H's, and after about 5 minutes I found the name Paul Hanselmann on Ramshorn Road. Ramshorn Road was close to the tornado path. So I called Paul Hanselmann. This immediately paid off as his house was unfortunately hit by the tornado. He gave me 2 other names and then those people referred me to others. So information piled up quickly. I want to thank all those who took the time to share information over the phone.

The 1st and strongest Albany county tornado touched down about 0.4 miles east of the intersection of Albany, Larimer and Laramie counties, or about 3/4 mile west of Harriman road along the state line(elevation 7500 ft) at 1857 UTC (1257 pm MDT). I used radar to determine the exact times as this is usually the most accurate method. Prior to the tornado, dense fog shrouded the eastern slopes of the Laramie mountains, with visibilities almost zero at the Walno residence near the tri-county border. Wylie Walno arrived home just  before the storm hit the area. He said that the visibility suddenly jumped from near zero to unlimited as the storm passed to his north. He could see low-hanging clouds pass by. Golfball sized hail also occurred at the Walno residence. The first signs of tornado damage occurred at the residence of Richard Miller. Two trees on his property were downed and his garage door was bent. Half dollar sized hail also occurred there. Immediately to the northwest, 20 ponderosa pine trees were downed on the Claire Hoover farm as the tornado passed between the house and a barn. Then the tornado toppled 4 more trees on Belinda Scott's property. A few trees were downed on Wylie Walno's property.

Fairly extensive tree damage occurred just northwest of the initial touchdown location as the tornado widened. Peter Hansen reported to me that the tornado downed trees for several miles on his property.  Tim Warfield told me that there was extensive tree damage on his land. By 1904 UTC the tornado was over 1/4 mile wide and was climbing up the Laramie Ridge to 8000ft. Very old pine trees 3 to 4 ft. in diameter were mowed down by the tornado. Tom Nowak, Jim Price and another person were putting fish into Imson Pond in the dense fog with visibilities near 100ft. Quarter to ping-pong ball sized hail chased them to their trucks. This is a good thing since they were then hit by the tornado around 1908 UTC. They described a frightening experience. The tornado buffeted their vehicles. One truck contained a 1000 lb fish tank. This truck rocked back and forth by the tornado and most of the windows were smashed out. Another truck was actually lifted off the ground and set back down, with windows knocked out as well. A camper shell was broken off one of the trucks and flung 1/2 mile to the south. Debris was flying everywhere during the tornado including picnic tables.  Large trees were downed on both sides of the road near the pond.

After leaving the pond, the tornado hit on Ramshorn Rd. Ted Lewis measured 153 mph winds on his Davis Monitor 2. His house faired fairly well even though huge pines were blown down. A 12 ft. aluminum boat was blown 500 yds. The tornado then hit the Paul Hanselmann house also on Ramshorn Rd. The front half of his roof was blown off, with pieces of it found over 2 miles away. The back part of the roof was heavily damaged. The Hanselmann house was well constructed with concrete-filed styrafoam and was reinforced with steel rebar anchored to the foundation.

After leaving the Hanselmann house, the tornado moved over very rural territory for several miles. But there was a continuous damage path all the way northwest to Overlook Rd, with trees and fences downed all along the .6 to .9 mile wide path according to Bob Adams who documented the tornado path. Along this path the tornado climbed in elevation to 8500ft. Major damage occurred on West Vedauwoo road. The Gayle Wilson house was destroyed by the tornado. The roof was taken off  and the walls collapsed. 2X4's from her roof were embedded in the ground several feet. She told me this was quite an accomplishment since the ground is so hard(gravel-like) that it is difficult to even dig a shovel into it. She also reported that nails from her roof were embedded the wrong way into fence posts 500 yds away at another residence. Ping-pong to golfball sized hail occurred on Overlook Rd and Howe Lane. A grove of pines was downed by the tornado on Overlook Rd. The Harriman-Laramie tornado path was continuous for 18 miles from west of Harriman to north of Overlook Rd, and possibly for 31 miles to north of Laramie. There is a mountainous area between Overlook Rd and Laramie where no people live. Bob Adams told me that the continuous path seems to have ended beyond Overlook Rd. I am not sure if it will ever be determined whether the tornado continued all the way to Laramie continuously.

A tornado moved across I-80 southeast of Laramie around 1928 UTC, and then across the far eastern and northeastern part of Laramie between 1930 and 1935 UTC. F1 damage was done to many structures. I did not independently document the Laramie segment of the tornado. A quick internet search revealed the following damage:

Numerous homes damaged
Church damaged
Dance hall damaged
College of Agriculture greenhouse facilities damaged (5 of 18 greenhouses damaged and the hoophouse greenhouse destroyed)
Storage shed and wooden pole barn were destroyed near the greenhouses. Huge pieces of the barn were carried 250 yds.
10 large spruce trees were either uprooted of snapped in half at the greenhouse facilities
Many trees were snapped in half or uprooted at the Jacoby golf course

The tornadic storm continued to the north-northwest through central and northern Albany county. No tornado damage occurred, but this area is very rural. Given the fog, it is possible that tornadoes went unreported.
                       
The Harriman-Laramie tornado moved to the north-northwest at an average speed of 47 to 50 mph. The heading of the tornado was 320 degrees at the beginning of the path and 330 degrees toward the end. This was an exceptionally fast moving tornado by Wyoming standards. Typically the mid level (600-300mb) flow over southeast Wyoming is fairly weak in tornadic situations, hence strong right-movement and slow storm motion (10 to 30 mph). The fast storm motion on May 22 is more typical of the southeastern United States in winter or early spring. The storm accelerated as it climbed closer to the strong mid level flow. Basically, the storm was closer to the strong mid level winds after it climbed to 7500ft. As previously mentioned, the tornadoes on this day moved to the northwest and north-northwest. This is unusual, but certainly unprecedented. The April 23, 1960 tornado moved to the north-northwest.                      

                       

A small tornado apparently touched down 3 miles south of I-80 on Harriman Road a little later in the afternoon from another storm. This tornado moved to the north-northwest and downed trees in several locations. Several trees were downed 3 miles south of I80 on Harriman Rd on the William Prince property with this storm. They estimated winds up to 80 mph. An old log cabin was extensively damaged on Crystal Lake Road, with the roof blown across the road. This tornado continued north-northwest. A few trees were downed and shingles were torn off of a house. A heavy camper was turned upside down. This damage occurred about 4 miles east-northeast of Buford. Residents described this event as a mini-tornado. The tornado most likely started between 2115 and 2125 UTC (325 pm MDT) and ended before 2140 UTC. These times were obtained by matching the radar imagery to the locations that received damage. However, the radar signatures were not nearly as clear cut with this storm since the tornado was so small. This storm was not as strong as the Harriman-Laramie storm, but still containted dime sized hail that was blowing horizontally at the William Prince residence. The southern end of the storm was centered just east of Wellington, CO at 2038 UTC, 6 miles east of Harriman at 2108 UTC, 3 miles west of Granite at 2129 UTC and 4 miles west-northwest of Granite at 2137 UTC. The area southeast of  Granite to east of Harriman is completely devoid of people. It is possible that tornadoes occurred earlier with this storm.

The path of this tornado was very close to the path of  the April 23, 1960 tornado. It appears that the 1960 tornado path was about 1 to 2 miles west of this tornado, and about 4 to 7 miles east of the Harriman-Laramie tornado. I drew the path of the 1960 tornado in 2000 with the help of Walter Ferguson whose family has resided in the local area for several generations.

According to Wylie Walno, another tornado apparently destroyed a barn 4 miles west of the tri-county border. Hail accumulated to a foot deep in this area and took 3 days to melt. I am still trying to confirm this tornado, but the area is very sparsely populated. The same storm produced large amounts of hail west of Virginia Dale. This center of the storm was located 5 miles southeast of Virginia Dale at 2012 UTC, from Virginia Dale to 4 miles northwest of Virginia Dale at 2025 UTC, 4.5 miles northwest of Virginia Dale at 2029 UTC (6 to 8 miles west of Harriman) and 1.5 miles east of Tie Siding at 2038 UTC. 

A terrain map with the Windsor and Harriman-Laramie tornado paths can be found here. A more zoomed in version can be found here.  A terrain map with both Wyoming tornadoes plotted is here. The possible tornado that occurred several miles west of Harriman is not plotted here. Wylie Walno reported to me that there may have been another tornado that destroyed a barn about 4 miles west of his house(or about 5 miles west of Harriman). Interestingly, a severe storm did move along US-287 from west of Virginia Dale to 5-8 miles west of Harriman to just east of Tie Siding. But there is just not enough information thus far about this event to plot a damage path.                           

A deep upper trough was digging into the intermountain west at 00 UTC May 22, 2008. 500mb winds of 100 kts on the back side of this trough were indicative of a deepening system. The 500mb height in the center of the upper low was 550 dm over central UT. The surface chart at 00 UTC showed a surface front stretching from central LA into north TX and then into northeast NM and eastern Colorado. Only marginal moisture was in place across western Kansas with surface dewpoints in the 50-55F range. However, rich moisture in the Red River Valley of southern Oklahoma and north Texas was poised to make a fast return.                

The 03, 06, 09 and 12 UTC surface charts show a strong surge of moisture through western Oklahoma, western Kansas and eventually eastern Colorado. By 03 UTC, 60-65F dewpoints were surging through northwest Oklahoma and into the eastern Oklahoma panhandle. In fact by 06 UTC, 55-60F surface dewpoints were already surging into eastern Colorado. By 09 UTC, the dewpoint at Limon, CO was up to 58F, with 53-55F dewpoints along the front range of  northern Colorado. Limon reported overcast skies at 1800ft, which indicates the low level moisture was at least 1800 ft deep.

By 12 UTC the moist axis was located from southwest Kansas into eastern and northern Colorado and had shifted a little to the northeast since 09 UTC. The 12 UTC 500mb chart indicated strong cooling since 00 UTC. The 500mb temperature was down to -14C at Denver. The pacific cold front had already progressed through Albuquerque as seen on the 700mb chart. The 700mb temperature was down to -1C at Albuquerque. Mid level cooling had obviously occurred even ahead of the front across the plains and at Denver.

By 15UTC, the warm front had progressed into central CO and western Kansas. Rich moisture was in place across the front range of northern CO with 54 and 55F dewpoints at Greeley and Akron respectively. The surface theta-e axis extended from central Kansas into northwest Kanas to Woodrow and Greeley, CO. Strong upslope flow was occurring and rich moisture was being transported into the Laramie Ridge and up the Laramie mountains. A surface dryline was beginning to take shape from the western Panhandles to extreme eastern Colorado.


Assuming saturation, what T/TD would be required at sea level to achieve the same theta-e as Harriman, WY?  Since the T/TD were 47F/47F at Harriman, a T/TD of 66F/66F would be required at 1000mb to yield the same theta-e. Why is this? To understand this, let's look at the potential temperature and mixing ratio's for both locations. For Harriman, the potential temperature and mixing raio was 89F and 9.2 g/kg. At 1000mb, a location with T/TD of 66F/66F would have a potential temperature of 66F and mixing ratio of  13.8 g/kg. Thus, the mixing ratio would be 50% lower at Harriman than at the sea level location. However, the potential temperature would be 23F higher at Harriman. Thus the notion that it was too cool on the Laramie Ridge on May 22, 2008 for severe storms is obviously misguided. In fact, it was warm enough so that the level of free convection was near the ground. This is despite dense fog and actual temperatures from 44 to 48F. Thus before drawing conclusions about the severe weather environment, one should modify soundings using actual surface observations. Sometimes this requires the use of mesonet data since surface observations are sparse.

Let's compare(Table 1) the theta-e values on the elevated terrain and lower terrain by displaying temperature, dewpoint, mixing ratio, potential temperature and equivalent potential temperatures at 17 UTC. Note that only temperature data were available for Harriman, Lynch, Virginia Dale 7 ENE and Emkay. Dewpoint data actually were available at the remainder of the stations including Estes Park, Crystal Lakes and Red Feather Lakes. However, since dense fog was present at these 4 Wyoming stations through 19 UTC, we will assume that the dewpoints were equal to the temperatures.

Notice that Lynch, WY actually has the same theta-e as Salina, KS even though the temperature/dewpoint are 21F/13F higher at Salina. The mixing ratio is 30% higher at Salina, so the potential temperature must compensate to yield similar theta-e values. Indeed, the potential temperature was 86.7F at Lynch and only 73.6F at Salina.

Surface theta-e continued to increase from 17 to 18 to 19 UTC. Tables 2 and 3 show temperature, dewpoint, mixing ratio, potential temperature and equivalent potential temperature values for various sites over the plains. Again, this is done to demonstrate that T/Td values cannot be used without elevation to assess how "juiced up" the surface layer is.  In Table 5 the theta-e values between 330K and 335K are highlighted in red. The theta-e values from 330 to 335K are shown in a partially analyzed surface chart for 19 UTC.

Let's compare(Table 2) the theta-e values on the elevated terrain and lower terrain by displaying temperature, dewpoint, mixing ratio, potential temperature and equivalent potential temperatures at 18 UTC.

Notice that Greeley, CO actually has a higher theta-e than Emporia, KS even though the temperature/dewpoint are 6F/8F higher at Emporia. The mixing ratio is 16% higher at Emporia, so the potential temperature must have compensated to yield a higher theta-e at Greeley. Indeed, the potential temperature was 98.4F at Greeley and only 82.3F at Emporia.

Notice that Red Feather Lakes, CO actually has a higher theta-e than Concordia, KS even though the temperature/dewpoint are 23F/15F higher at Concordia. The mixing ratio is 36% higher at Concordia, so the potential temperature must compensate to yield a higher theta-e at Red Feather Lakes. Indeed, the potential temperature was 90.7F at Red Feather Lakes and only 75.1F at Concordia. Severe storms passed just east of Red Feather Lakes in the early afternoon.

Notice that Harriman, WY actually has almost the same theta-e as Scandia, KS, even though the temperature/dewpoint are 22F/14F higher at Scandia. The mixing ratio is 31% higher at Scandia, so the potential temperature must compensate to yield similar theta-e values. Indeed, the potential temperature was 89.7F at Harriman and only 77F at Scandia.

At 19 UTC, two mesonet observations and 1 cooperative observer location recorded hourly temperatures. The temperature was 47F at Harriman (756mb), 48.5F at the cooperative observer site 7 miles east-northeast of Virginia Dale (767mb) and 48F at Lynch (762mb). These 3 observations lie along the same moist adiabat, as one would expect in moist upslope flow. So I have fairly high confidence in the accuracy of these measurements. Veta Mitchell, the cooperative observer 7 miles east-northeast of Virginia Dale provided me with the hourly temperature measurements for her location. The tornado actually first touched down about 2 miles north-northwest of her house. So the hourly measurements that she collected are very useful in determining surface based CAPE.
                       
An important thing to note is that a 47F dewpoint at Harriman actually has about the same mixing ratio as a 54.5F dewpoint at 1000mb. Of course this assumes that the sea level pressure at Virginia Dale and 1000mb are similar. If the sea level pressure is higher at the lower elevation then the difference would be greater. Also, even though 47F seems chilly, this temperature at 7500ft actually lies along the same dry adiabat as 90F at 1000mb. 

Since visibilities were near zero before the storm, I am assuming that dewpoints were equal to the temperatures. I constructed approximate soundings for these locations using the 18 UTC RUC initialization and 18 UTC Denver sounding. Of course, the boundary layer had to be modified based on the surface observations. I modified using the 19 UTC observations since these are just prior to the tornadic storm. The RUC soundings were more representative than the NAM/WRF soundings. The nam soundings were superadiabatic near the surface and dry adiabatic above the surface layer. This is not reasonable. The RUC soundings were closer to moist adiabatic from the surface to above 700mb. The modified 18 UTC RUC soundings yielded similar CAPE values to the modified 18 UTC Denver sounding. I have determined that the surface based CAPE was 1000-1300 j/kg near the beginning of the Harriman-Laramie tornado path where surface measurements were available. The theta-e values at the three locations were almost identical.

Table 4 and Table 5 show the surface based CAPE values for Harriman, Virginia Dale and Lynch. Table 1 uses the 18 UTC Denver sounding while Table 2 uses the 18 UTC RUC initialization. These are modified using the temperature readings from the 3 stations and assumes saturation (there was dense fog).                     

The vertical wind shear profile featured strong shear. The surface wind backed to the northeast by midday at the mesonet locations. However, windspeed is not represented the same way at the mesonet locations. For some hours the wind speed was the same as the wind gust while for other hours they were vastly different. Also, some of the winds measurements were influenced by inflow into the storm. The pre-storm winds were probably about 25 kts from the east-northeast or northeast(060 deg at 25kts). Wylie Walno reported that the winds was about 20-30 kts. For the winds above the surface we used the Medicine Bow and Platteville profilers as well as the 18 UTC RUC. The wind just above the moist layer (600mb) was about 140 deg at 55kts. So there was tremendous shear between the surface and 600mb (1.5 to 1.8 km agl). The 500mb wind was from 150 deg at 55 kts while the 400mb wind was from 165 deg at 75 kts. 400mb is 4.4 to 4.7 km above the surface, so the shear from the surface to 4.5 km was about 80 kts.

It is very difficult to achieve low dewpoint depressions, relatively high theta-e values at low-levels and excellent vertical wind shear at 7500-8500ft on the Laramie Ridge/Mountains. As previously mentioned, the dewpoints were equal to the temperatures along and east of the summit of the Laramie Ridge. But surface based CAPE values still exceeded 1000 j/kg. In typical low plains severe storm situations, 1000 j/kg CAPE would be considered very marginal. When CAPE is marginal, tornadic storms can still occur, especially when LCL and LFC heights are low and considerable surface based CAPE exists at low levels. This was indeed the case on May 22 on the Laramie Ridge. In fact, after initially weakening upon moving into cooler air on higher terrain northwest of Wellington, CO, the storm quickly reintensified after encountering dense fog east of Virginia Dale. 

One would think that it would be easier to get sufficient CAPE and shear on the high terrain(7000+ ft) in June or July than in April or May. But this is not necessarily the case. Strong synoptic scale systems in spring can have very strong upslope flow associated with them, whereas systems in late spring and summer tend to be weaker, with weaker upslope flow. That said, the upslope flow tends to be cooler in April and May and oftentimes more stable. Upslope flow tends to be located on the cool side of a surface front or north of a developing surface low. Again, these airmasses tend to be too cool in April and much of May. This is why significant tornadoes are so rare in the immediate lee of the Laramie Mountains. The upslope flow in the May 22 case was "cooler", but 1000+ j/kg surface based CAPE values were still achieved. Elevations from 5000-6000 ft do have more severe weather in June and July compared to April and May. But further west on the very high terrain, strong upslope flow is generally required to obtain adequate theta-e values, low dewpoint depressions and high shear that typically accompany tornadic storms. By mid-June, strong upper systems become less common so that very strong upslope flow is rare. In the May 22 case, gulf moisture raced northwestward from north Texas and southern Oklahoma into northern Colorado and southeast Wyoming in 9 to 12 hours from 00 to 09 UTC. Then during the day of May 22, a very strong upslope flow developed north of a surface front. Very strong upslope flow is required to keep adequate mixing ratios at the 7000-8000 ft elevations on the eastern slopes of the Laramie mountains. It is also important to understand that it is the potential temperature that is important in achieving higher theta-e values and not the temperature.Even in the presence of only moderately cool 500-300mb temperatures of -13C, -26C and -40C, surface temperatures from 7 to 9C were still sufficient to yield moderately high theta-e values, and hence moderately high CAPE values. When potential temperatures are high, mixing ratios do not have to be very high to achieve sufficiently high theta-e values. By later in June and July, surface temperatures become so warm that cloud bases are typically too high for tornadic storms given the meager mixing ratios typically found at these elevations.

Additional surface charts will be coming soon. Since terrain is so crucial to this meteorological discussion, I like to plot the surface observations on top of the terrain. However, this is a very labor-intensive process, especially since I need to include the mesonet observations.               
   
Hodographs will be constructed soon.