Thunderstorm Initiation

The thunderstorms that smashed into the Chicago-Mackinac fleet around midnight on Sunday began forming several hours earlier in two distinct areas separated by over 150 miles. These two storm systems developed independently, later merging near upper Lake Michigan. The first complex of storms formed in Douglas County, WI just south of Duluth, MN, while the second formed near Marquette, MI, in the western portion of the upper peninsula (map). If you are unfamiliar with the ingredients for thunderstorm development, or need a refresher, please review this instructional page.

Wisconsin Storms
At 5:17 pm (2117Z), the broad perspective of base reflectivity from the National Weather Service (NWS) Doppler radar at Duluth, MN (KDLH) (figure 1) showed a large number of storms to the west and northwest, and a compact cluster of thunderstorms just a little to the north. With the exception of a couple of minor showers, the area to the south and southeast of KDLH was storm-free. However, by 5:53 pm (2153Z) (figure 2) several small storms were underway in this area.

Figure 1: KDLH base reflectivity 5:17 pm (EST) (2117Z). (larger image)
Figure 2: KDLH base reflectivity 5:53 pm (EST) (2153Z). (larger image)

The initiation of thunderstorms appears to have been prompted by a complex interaction of a warm front, lake breeze boundaries, and terrain. As previously discussed, a curved warm front was draped across the northwest corner of Wisconsin. In addition, the combination of Lake Superior surface water temperatures in the low 60s F and extremely warm temperatures along the shore led to the widespread development of lake breezes. The tell-tale signs of the lake breeze fronts were clearly evident on both radar and visible satellite imagery. The presence of the lake breeze is also confirmed by the hourly observations at Superior, WI which showed both a wind shift to the northeast and a dramatic decrease in temperature beginning at approximately 3:00 pm (1900Z).

Surface boundaries, such as a warm fronts and lake breeze fronts, are characterized by converging winds and the resulting upward motion of air, processes which support the development of thunderstorms. It is also likely that the topography of the region acted to enhance thunderstorm development. The area to the southeast of Duluth, MN is characterized by a broad plateau that is 300 to 400 feet higher than the region to the west. In addition to providing orographic lift, higher elevations typically warm faster than the surrounding area which may result in a localized area of relatively lower surface barometric pressure. These processes may have increased convergence near the surface and upward motion in the air column.

The thunderstorms in northwest Wisconsin underwent significant development between 5:58 pm (2158Z) (figure 3) and 6:55 pm (2255Z) (figure 4). During this hour, individual thunderstorms blossomed and subsequently merged into an Mesoscale Convective System (MCS) over north-central Wisconsin (radar loop). A Mesoscale Convective System is a collection of individual thunderstorms organized in either a linear or oval shape whose overall size is greater than the individual storms comprising the system.

Figure 3: KDLH base reflectivity 5:58 pm (EST) (2158Z). (larger image)
Figure 4: KDLH base reflectivity 6:55 pm (EST) (2255Z). (larger image)

Over the next two hours, the main body of the system continued to expand, but the most interesting development was the creation of a narrow, linear extension from the sytem's western edge (figures 5 and 6) (radar loop). In a process called back-building, a downdraft from a mature thunderstorm promotes the development of a new storm, which in turn promotes yet another new storm as it matures. The new storms are typically initiated on the upwind edge of the system (usually to the west or southwest) and act to increase the system's overall size. One storm during this period evolved from a fledging storm to a towering giant reaching 60,000 feet above the surface in just 37 minutes. Early in this period (7:11 pm (2311Z)), the development of strong updrafts across northern Wisconsin prompted the SPC to issue Mesoscale Discussion (MD) #1642 (link). The MD mentioned the threat of large hail and suggested that severe weather would spread across Michigan's Upper Peninsula.

Figure 5: KDLH base reflectivity 6:59 pm (2259Z). (larger image)
Figure 6: KDLH base reflectivity 8:08 pm (0008Z-7/18) (larger image)

The mean wind associated with the cloud layer (a deep layer of the atmosphere) helps to steer Mesoscale Convective Systems. The base of this deep layer is the Level of Free Convection (LFC), while the upper limit is the Equilibrium Level (EL). Meteorologists often use the wind at 500 mb (18,000 feet) as a rough proxy for the cloud layer.

During the initial phase of development in northwestern Wisconsin, the mean cloud layer winds were moderate at approximately 40 knots from the west-northwest. From this direction, the mean cloud layer wind was nearly perpendicular to the initiating surface boundary (warm front) and, in response, the complex took on a west to east orientation slightly to the right of the mean wind.

As the storms moved to the east from 8:06 pm (0006Z) to 9:22 pm (0122Z) (figures 7 and 8), they encountered an environment where the steering currents were more northwesterly. In response to the veering of the cloud layer wind, the overall system pivoted into a more southwest to northeast orientation as the individual imbedded thunderstorms realigned to the direction of the mean wind (radar loop). At the same time, the complex transitioned into a classic forward-building MCS with a squall line containing the strongest storms defining its leading edge and a broad area of lighter precipitation to the rear. A parameter developed at the SPC based upon a combination of deep-layer wind shear and instability and used to predict the future of an organized system indicated this MCS had a 90% chance of persisting (link).

Figure 7: KGRB base reflectivity 8:06 pm (0006Z-7/18) (larger image)
Figure 8: KGRB base reflectivity 9:22 pm (0122Z-7/18) (larger image)

The rapid intensification of the storms associated with this system, and the likelihood of their persistence, prompted the Storm Prediction Center to issue Severe Thunderstorm Watch #655 (link) at 8:25 pm (0025Z). The watch area included northern and northeastern Wisconsin and a small portion of northern Lake Michigan and referenced that 1.5" hail, dangerous lightning and gusts to 70 mph were possible. In the SPC MD (link) that accompanied Watch #655, the forecaster indicated that "the activity is moving southeastward at roughly 40 knots... and ... it would seem likely that this complex of storms will overspread much of Lake Michigan ... over the next several hours."

Michigan Storms
Similar to the region around Duluth, MN, the cold water of Lake Superior and warm temperatures over the land induced lake breezes to form in several areas around Marquette, Michigan. Base reflectivity radar imagery from Marquette (KMQT) and visible satellite imagery confirm that lake breeze circulations were well established by 4:45 PM (2045Z). In addition, the hourly observations at Marquette indicated that the lake breeze front had pressed well inland by 6:00 PM (2156Z).

At nearly the same time that storms appeared southeast of Duluth, a fledgling storm appeared on the Marquette (KMQT) radar imagery a little before 6:00 pm (2200Z) (figure 9). Convective development continued over the next fifty minutes, but at a far slower pace than the storms southeast of Duluth. At 6:40 pm (2242Z) (figure 10), it was apparent that the storms were traveling east-southeast along a subtle lake breeze boundary. As afternoon lake breeze storms often do, these storms largely fizzled out by 7:42 pm (2342Z) rather than intensifying.

Figure 9: KMQT Base reflectivity 5:56 pm (2156Z) (larger image)
Figure 10: KMQT Base reflectivity 6:42 pm (2242Z) (larger image)

Not long after the initial storms dissipated, convection reignited near the western end of the lake breeze front. Instead of dissipating as the previous round of lake breeze storms did, this new development intensified significantly and grew in size between 8:05 pm (0005Z) (figure 11) and 9:24 pm (0124Z) (figure 12), resulting in a multi-cellular cluster of thunderstorms immediately east of the radar site (radar loop). The survival of the second round of lake breeze convection was aided by the extra lift provided by the extension of the frontal boundary into Michigan, and the easterly transition of the 500 mb shortwave trough and an area of divergence at 300 mb -- upper-air features that supported upward motion in the air column.

During this same period, new storms formed a little further east along a lake breeze boundary over the peninsula separating Lake Superior and Lake Michigan. The storms to the east and southeast of Marquette continued to grow and were rapidly approaching the northern shore of Lake Michigan by 10:01 pm (0201Z) (link).

Figure 11: KMQT Base reflectivity 8:05 pm (EST) (0005Z-7/18) (larger image)
Figure 12: KMQT Base reflectivity 9:24 pm (EST) (0124Z-7/18) (larger image)

The Storms Consolidate
By 10:00 pm (0200Z) (figure 13), after nearly four hours of development, there were three separate areas of thunderstorms approaching northern Lake Michigan (annotated radar image). To the south and west was the MCS with its embedded squall line that formed from the original thunderstorms southeast of Duluth, MN. Directly north and north-northeast of the MCS was the cluster of storms that had also formed near Duluth; until this time these had remained separated from the complex as it took on its linear form and began reorienting to the northwesterly deep layer shear. And finally, the third area, characterized by the strongest storms, was located slightly to the east of Marquette, Michigan.

Figure 13: KMQT base reflectivity 10:01 pm (0201Z) (larger image)

Between 10:00 pm (0200Z) (figure 14) and 11:00 pm (0300Z) (figure 15), the intensification of existing thunderstorms and the development of new storms consolidated the previously separate areas into a continuous convective line stretching for nearly 250 miles along the shore of Lake Michigan (radar loop).

Figure 14: KAPX base reflectivity 10:02 pm (0202Z) (larger image)
Figure 15: KAPX base reflectivity 11:01 pm (0301Z) (larger image)

Marine Warnings Are Issued
By 11:00 pm (0300Z), several Special Marine Warnings and Special Marine Statements had already been issued by the National Weather Service offices at Green Bay, WI (KGRB) and Gaylord, MI (KAPX). The first Special Marine Warning was issued by KGRB at 09:49 pm (0149Z) (link) and warned of the potential for 50 knot winds on the central and southern portion of Green Bay. Nearly simultaneously, KAPX published a Marine Weather Statement (link) which alerted mariners to a line of thunderstorms rapidly approaching northern Lake Michigan that were capable of producing wind gusts to 33 knots. KGRB reinforced their initial Special Marine Warning covering most of Green Bay with a Marine Weather Statement at 10:20 pm (0220Z) (link).

At 10:47 pm (0247Z), KAPX issued an additional Marine Weather Statement (link) that expanded the area of concern to the entire portion of Lake Michigan north of Charlevoix, MI. This Statement continued to warn of "winds up to 33 knots". Just one minute later at 10:48 pm (0248Z), KGRB issued a Special Marine Warning (link) that covered Lake Michigan south of Charlevoix, MI and noted the storms were "capable of producing dangerous winds to 50 knots". In addition to suggesting higher wind speeds, the precautionary/preparedness section of the KGRB Warning had a much sterner tone:


Mariners can expect dangerous winds in excess of 50 knots...high waves...dangerous lightning... and heavy rains. Boaters should seek safe harbor immediately.


These storms will likely produce strong gusty winds of 30 knots or higher..and could pose a serious hazard for boaters...Boaters should consider heading for shore before the storm arrives.

Based upon the text in the initial round of Warnings and Statements, NWS forecasters were clearly expecting the southern portion of the approaching system to produce more severe weather.