Midterm Colleen Heck

Storm Overview

On August 14, 2020, a line of thunderstorms swept across Minnesota. The conditions to produce severe thunderstorms were very favorable this day. Figure 1 below is a sounding taken from Minneapolis in the early morning. Temperature and dewpoint lines indicate the air was very moist, feeding the conditions for potentially severe weather. CAPE values were already over 500 J/kg, meaning there was instability in the atmosphere. High CIN values could later get mixed out during the course of the day and fuel the already brewing atmosphere. 

Figure 2 is an upper-level map showing pressure contours, divergence as well as wind speeds. This storm's setup was perfectly located, sitting downstream of an upper-level trough where it would be expected to see divergence aloft. These upper-level divergences contours sitting over the state's eastern edge would force convergence at the surface and the air would be forced to rise. 

In Figure 3, the surface analysis also coincides with favorable thunderstorm producing conditions. Just over Minnesota, there was a warm front and cold front meeting. This merging of the two different air masses and the other parameters talked about earlier made it highly likely that this area will see intense convective storms, and it did. 

Figure 1. Early morning sounding from Minneapolis, MN at 12 Z on 8/14/2020. This sounding is from about 150 miles south of Duluth, MN. 

Figure 2. 300 mb map of height (black contours), divergence (pink contours) and wind speeds (fill)

Figure 3. Surface Analysis from 19:30 Z on 8/14/2020. The map labels high and low pressure areas and pressure contours as well as fronts. The red lines are warm fronts, the blue lines are cold fronts and the dashed orange lines mark trough and ridge axes. 

When the storm finally came through in the late afternoon/early evening of August 14th, many warnings were issued with the event. Storm spotters reported lots of high winds and a few areas with large hail just south of Duluth. However, the most significant threat that this day was the possibility of flooding and tornadoes. The first storm reports came in around 17:33 CDT and didn't stop until well after the storm had passed. Figure 4 below shows local storm reports from 17:33 CDT on the 14th through 11:45 CDT the following morning.

Figure 4. Local storm reports before, during and after the storm

 Overall, my partner and I made issued 11 warnings and the NWS also issued 11. We issued 5 tornado warnings and the NWS issued 4. We issued 4 severe thunderstorm warnings and the NWS issued 5. We both issued 2 flash flood warnings. 

Warning 1

Figure 5a. A flash flood warning we issued at 23:42 UTC on August 14th until 1:50 UTC on the 15th. This show a picture of the warned area plotted over the radar reflectivity from that exact moment in time

Figure 5b. The NWS issued flash flood warning at 11:47 UTC August 14th until 1:47 UTC on August 15th.

My partner and I based our warning on a storm report from a local sheriff who had reported flooding in the area. At 23:39 UTC, the report came in and said, "STATE PATROL REPORTS THAT 2 FEET OF WATER WAS COVERING THE ROAD ON 169 NEAR GLENWOOD DRIVE. ROAD WILL BE SHUT DOWN"

Upon receiving the warning, my partner and I immediately looked at the reflectivity and saw very high levels of reflectivity in that area that had been moving with the storm. Based on this area's report and geographic knowledge, meaning this being home to many lakes and rivers, we immediately took action and issued a flash flood warning. 

We are happy that the NWS shortly reported the same warning, for the same duration of time, for a similar area just south of the Chippewa National Forest. I don't know if we gave enough lead time since when the report came in, there were already 2 feet of water on the road, but we got our warning out fast soon after. The 5 minutes between our warning and the NWS's warning was very impactful because a lot can happen over 5 minutes.  

If we could redo the warning, I would redraw it a little more eastward with the storm's direction, similar to how the NWS did. 

Warning 2

Figure 6a. A tornado warning we issued at 00 UTC on August 15, 2020 until 1:50 UTC. This picture shows the warning area mapped over the radial velocity from the radar at that time. 

Figure 6b. The NWS issued tornado warning at 11:59 UTC on August 14, through 00:30 UTC on August 15. 

Figure 6c. Radial velocity of a possible velocity couplet at 23:54 UTC August 14.

My partner and I based our warning on a local storm report. The report that came in was a delayed report at 11:58 UTC and said, "DELAYED REPORT. DRONE VIDEO SHOWED TORNADO DAMAGE TO A STAND OF TREES. SEVERAL TREES WERE BROKEN APPROXIMATELY 25 TO 30 FT AGL WITH SEVERAL MORE TREES UPROOTED. DAMAGE IS CONSISTENT WITH LOW-END EF-1 TORNADO WITH MAX WIND GUSTS OF 85 TO 90 MPH. TIME ESTIMATED FROM RADAR."

Since there was actual video drone proof of tornado damage and another report came in moments later, we decided to issue a warning. We were not far off the NWS's report time, only a minute after. However, the area of our warning was a little off compared to the NWS's. Our warning was more north-west than it should have been, and you can see this by comparing the Grayling Marsh WMA location in relation to the warning sites in both Figures 6a and 6b. 

We also looked at radial velocity to try and confirm the location of a possible tornado. We were looking for a velocity couplet, but we could not find one near our warning. Even with a lack of radar evidence, the reports that came in were enough evidence to make us want to issue the warning. Looking back, there is a possible velocity couplet we failed to act on. We did see this in the radial velocity and talked about it but could not tell if it was a couplet since the shape is not extremely well defined like other examples we had seen. We decided to ignore it since ZDR and CC didn't show any hail evidence or a debris ball from a tornado. Figure 6c is the possible velocity couplet at 12:54 UTC on August 14th in the exact area the tornadoes were later reported. Based on this evidence, our warning was out 6 minutes late and north of where it should have been. I don't think we gave the best warning for the residents based on the timing or location. I also don't know if the NWS gave the best warning either. They might have also chosen to ignore the possible couplet or wanted to wait for more evidence since their warning was issued 5 minutes after the radar velocity image came in. 

Warning 3

Figure 7a. A severe thunderstorm warning we issued at 24:06 UTC on August 15. This warning area show the radar reflectivity from that time 

Figure 7b. The NWS severe thunderstorm warning for 24:10 UTC on August 15, 2020.

 

We issued our warning based on storm report evidence. There was not just one report, but many reports of trees down and even a couple of docks uprooted and flipped over. These reports focused our attention on the radial velocity and the possible wind speeds that the area might be seeing. With the help of radar and storm reporters, we issued a severe thunderstorm warning. 

I think the warning could have been a little sooner, but it was not a moment too late. Over the next 20 minutes after our warning went out, reports of wind damage continued to come in. 

Compared to the NWS warning, we were early and had a very similar location to their warning. You can see this when comparing the Douglas County forecast location in figures 7a and 7b. Considering the number of damage reports and storm reports that came in after our warning was first issued, residents still did have a reasonable amount of time to seek shelter indoors. 

Forecasting Difficulties

The lack of county lines on the radar made it challenging to talk with my partner and issue the correct warnings. One of us shared our screen to try and make sure we were looking at the same thing but having to describe pinpoint locations was a challenge. County lines or more city names, when zoomed out, would have been beneficial too in drawing the polygons for the warning areas.

There are challenges or difficulties with the waring processes. First, I am sure that it is so easy to want to issue a warning for every storm report that comes in in the real world. However, warning coordinators and forecasters have to resist that urge. Issuing too many warnings can be problematic for the NWS and prevent the public from trusting them later down the road. That is why the NWS uses the reports to help guide their decision, and it is important that there is some radar evidence when making the final call. I can only imagine the stress behind the job because you want the public to be safe, but you also want to keep their trust. Missing a warning or over issuing is a massive problem in the modern system. 

One challenge I faced in this exercise was That my internet dropped, and my partner was forced to make the decisions by herself for a few minutes while I got my internet back up. If I remember correctly, my partner was either in the middle of issuing a warning and had to give one while I was gone. Having reassurance from another forecaster is definitely necessary when issuing warnings. That extra person or two acts as a way to check your work or help give you any confirmation if you have any doubts you may have.