A Case Study of the April 18, 2004
Severe Weather Event across Upstate New York

By Michael L. Jurewicz, Sr. and Michael Evans

1. Introduction

A mesoscale convective system (MCS) moved across southern Ontario during the early morning hours of April 18, 2004 (1200 to 1500 UTC), producing widespread wind damage.  As the MCS progressed eastward from late that morning into the afternoon (1500 to 1900 UTC), it continued to bring significant wind damage, as well as large hail, to parts of western and central New York State.  In fact, wind gusts of 80 to 100 mph were recorded in the Niagara Falls area and an F0 tornado touched down in the Finger Lakes region in Cayuga county.  By late afternoon, the convection began to weaken over the Catskill Mountains.

An interesting aspect of this event was that the MCS propagated through environments characterized by relatively cool, stable boundary layer conditions during the majority of its life cycle.  Normally, lower-level stability does not promote the downward transfer of strong winds to the surface.  Thus, the magnitude of the winds observed on this day over such a large area and prolonged time period becomes particularly noteworthy.

In this study, we’ll track the MCS from its inception during the evening of April 17th over the Upper Midwest to its eventual demise over eastern New York State late in the afternoon on the 18th.  Synoptic and mesoscale settings will be examined during this period, with an emphasis on factors that likely contributed to convective initiation, intensification, or weakening.  Also, some theories will be proposed on how downdraft intensity was able to be maintained despite marginal, and even unfavorable at times, lower-level conditions.  Radar perspectives are given from both the Buffalo, NY (KBUF) and Binghamton, NY (KBGM) WSR-88D’s, once the convection reached New York State.  Lastly, a summary is presented. 

2. Synoptic Overview

Late in the afternoon on April 17th, a warm front at the surface stretched eastward from a deepening low pressure system in Colorado across the central Plains (Fig. 1).  However, there was little or no convective development across the region at this time.  Note the cool, dry air mass in place across the Dakotas and Minnesota (surface dew points mostly in the 20s and 30s F).  Meanwhile, in the warm sector, a strong southerly flow was advecting higher theta-e air (surface dew points into the 60s F) northward from Oklahoma and eastern Kansas towards the vicinity of the warm frontal boundary across southern Nebraska and Iowa.  A tightening temperature and moisture gradient seemed to be developing along and just north of the warm front.

By 0000 UTC on the 18th, an increasingly favorable environment for convective initiation was developing across portions of Iowa and Minnesota.  At 300 mb (Fig. 2a), the main jet axis extended from the base of an amplified trough over the southwestern United States across parts of the Rockies and northern Plains, then eventually into southern Canada.  Embedded within this jet core was a 100+ kt speed maximum over southern Ontario.   Iowa and Minnesota were located underneath the right-entrance region of the aforementioned jet streak at this time.  At mid-levels, the atmosphere was destabilizing.  Iowa and southern Minnesota were located along the southern edge of the stronger westerlies at 500 mb (Fig. 2b).  Short-wave energy within this belt of more pronounced flow was flattening the large-scale ridge axis and resulting in height falls and cooling at this level.  Meanwhile, at 700 mb (Fig. 2c), a trough extended from the Dakotas southeastward across southern Minnesota and into Iowa.  A tight thermal gradient and strong warm advection were associated with this trough.  At 850 mb (Fig. 2d), the key feature was a west to east oriented warm front, that stretched across Nebraska, northern Iowa, and northern Illinois.  A 30-40 kt low-level jet was transporting higher theta-e air northward and over the warm frontal boundary itself (see the axis of 10-12ºC dew points near the Iowa/Minnesota border).  An initialized Eta horizontal cross-section (Fig. 3) of divergence, omega, and wind from 0000 UTC on the 18th, showed a particularly deep layer of ascent across northern Iowa and southern Minnesota.  The lift in this area appeared to result from a pronounced convergence/divergence couplet.  Strong low-level convergence was noted below 800 mb; likely a result of close proximity to both the surface warm front and nose of the low-level jet.  Mid-level divergence was also seen; likely a result of being located on the tail end of a 60-70 kt jet streak at 500 mb.

As judged from radar, satellite, and lightning analyses, thunderstorms first developed between 0030 and 0100 UTC on the 18th near the Iowa/Minnesota border (Fig. 4).  During the evening, convection slowly organized as it pushed eastward from northern Iowa and southern Minnesota across Wisconsin and Lake Michigan (Fig. 5).  Between 0100 and 0600 UTC, thunderstorms became locally severe.  Severe reports primarily consisted of large hail during this period (Fig. 6).  The comparatively small number of wind reports versus large hail suggested that most of the storms were not surface-based.  Regional stability profiles from that evening implied that most of the convection was indeed elevated in nature (Fig. 7 and Fig. 8).  Another horizontal cross-section (using RUC data this time, Fig. 9), from 0600 UTC on the 18th, continued to show a deep layer of upward vertical motion collocated with the developing MCS over Wisconsin.  Through 0600 UTC on the 18th, it is apparent that the combination of large-scale lift, steep mid-level lapse rates (Figs. 10a and 10b), and pronounced shear north of the surface warm front was able to maintain convective development and organization, despite the presence of low-level stability.  It is also evident, however, that stable boundary layer conditions effectively prevented strong winds from mixing all the way down to the surface within multi-cell or isolated supercellular clusters.

From 0600 to 1100 UTC, the MCS moved steadily eastward across Lake Michigan and  Lower Michigan.  However, very little severe weather was reported during this period (refer back to Fig. 6).  This was an indication that strong winds were still not able to penetrate through to the surface.  Note the relatively cool, stable lower-level conditions that the MCS encountered over Lower Michigan during the pre-dawn hours on the 18th (Figs. 11a and 11b).  However, between 1100 and 1200 UTC, more organized linear structures began to develop on radar across eastern Lower Michigan and southern Lake Huron, with a corresponding increase in lightning activity and cooling trend in cloud top temperatures (Fig. 12 Fig. 13).  It was also during this period that reports of wind damage started to become more prevalent.  In fact, a wind gust of 70 mph was recorded near Michigan’s shoreline with southern Lake Huron just after 1100 UTC.

An inspection of the synoptic-scale environment at 1200 UTC on the 18th showed that the MCS continued to benefit from large-scale forcing.  The upper-level flow became strongly diffluent (Fig. 14a).  The atmosphere also continued to undergo destabilization at mid-levels, just ahead of the MCS.  An approaching 500 mb short-wave trough, moving along the southern fringe of the stronger westerlies, resulted in modest height falls and cooling at that level (Fig. 14b and Fig. 15).  Meanwhile, pronounced warm advection was taking place over the eastern Great Lakes region in the 850-700 mb layer (Fig. 14c and 14d). Close proximity to the nose of the 850 mb jet also supplied an influx of higher theta-e air and promoted strong low-level convergence.

As the MCS continued eastward into and across southern Ontario between 1200 and 1500 UTC, strong winds became the primary severe threat and reports of wind damage were widespread.  As mentioned before, though, little in the way of strong surface winds was associated with the MCS prior to around 1100 UTC.  A common theme during the overnight period (from 0100 to 1100 UTC on the 18th) was that the MCS encountered relatively cool, stable boundary layer environments from northern Iowa all the way across the majority of Lower Michigan.  The authors earlier theorized that the lack of strong winds resulted from their inability to penetrate the stable surface-based layer and mix down to ground level.  Given a similar lower-level environment over eastern Lower Michigan and near Lake Huron around daybreak (surface temperatures generally between 50 and 55F), as compared to areas upstream during the overnight period, why the sudden development of strong, damaging winds with this convective system? 

A possible key to answering the above question lies with the evolution of the mid-level environment along the track of the MCS.  Let’s take a look back at Figs. 7a and 7b.  The initial-hour RUC sounding at this time (0300 UTC) showed that the MCS was encountering an environment characterized by fairly moist mid-levels and dry lower-levels.  Although this "inverted-V" type signature can provide a favorable situation for microbursts, the lack of strong wind reports in the vicinity indicated that, in this particular case, downdrafts were generally not strong enough to reach the surface.  The boundary layer air was likely too cool and the mid-level air not dry enough to enhance any potential downdrafts.  Now let’s view Figs. 8a and 8b once again.  As compared to 0300 UTC, the environment at this juncture (0600 UTC) near the MCS featured cooler, moister lower-levels and slightly drier mid-levels.  Although mid-level lapse rates were steepening with time, the degree of mid-level drying still seemed insufficient to significantly enhance downdraft speeds.  Given the added detractor of pronounced low-level stability, the atmosphere was not conducive to downward momentum transfer.  By 0900 UTC over central Lower Michigan (Figs. 11a and 11b), as compared to 3 hours earlier, more earnest mid-level drying had developed.  Dew point depressions had increased from about 10oC to 13o-14oC in the 600-700 mb layer.  However, no severe weather was occurring near this time, thus indicative of the fact that any downdrafts were still not forceful enough to punch through the surface-based stable zone.  A subjective analysis of the initial-hour RUC sounding (Fig. 11b), using about 650 mb as the level of downdraft origination and minimum wet-bulb potential temperature, showed that the parcel trajectory (following the moist adiabat downward) would ultimately meet the environmental temperature line near the surface. Such an occurrence would presumably diminish fall speeds as the parcel neared ground level.  This reinforces the above notion that mid-level drying, to this point, was still not enough to completely overcome the strength of the low-level inversion, despite increasing Downward Convective Available Potential Energy (DCAPE) aloft.  Now let’s peruse the 1200 UTC sounding (Fig. 16) from Detroit, MI (KDTX).  KDTX was just southwest of where the MCS was located at about 1200 UTC (refer to Fig. 13).  Although a shallow inversion/zone of stability remained below 900 mb, it is evident that the process of mid-level drying we began to see over Lower Michigan about 0900 UTC had further intensified in the last 3 hours.  Dew point depressions in the 600-700 mb layer had now increased to about 17o-18oC.  Utilizing a similar analysis to the one we employed on Fig. 11b, it now appeared that the added mid-level drying would produce enough DCAPE to potentially bring any downdrafts all the way down to ground level, despite lingering cool surface temperatures.  Not coincidentally, perhaps, reports of strong surface winds/wind damage started to increase in frequency just prior to the time the KDTX sounding was taken.

As judged from the 0900 UTC RUC and 1200 UTC DTX soundings (Fig. 11b and Fig. 16), the main thrust of mid-level dry advection occurred around 650 mb.  Using initial-hour model data from the RUC and a series of satellite/radar images (not shown) between 0300 and 1200 UTC on the 18th, the movement/development of the mid-level dry punch was compared to the track of the MCS (Fig. 17a and Fig. 17d).  For much of the night, the main supply of dry mid-level air and the MCS were well separated.  However, the prevailing southwesterly flow advected the dry air aloft from the Ohio Valley into Lower Michigan during the pre-dawn hours.  By 1200 UTC, a layer of well mixed, very dry air at mid-levels had crossed directly into the path of the MCS.  The timing of this interaction seems to further substantiate its potential role in facilitating the downward transfer of strong winds.   

As briefly discussed earlier, once the evolution towards linear development had taken place, an organized squall line produced a swath of wind damage from extreme eastern Lower Michigan through southern Ontario between 1100 and 1500 UTC.  Some of the recent research on quasi-linear convective systems (QLCS) has focused on the relationship between low-level vertical shear and cold pool maintenance.  Such findings suggest that shear magnitudes of 30-40 kt or more in the lowest 2-5 km tend to produce the longest lived squall lines, with deeper, more erect meso-vortices embedded within them (Trapp and Weisman, 2003).  For this event, by 1500 UTC, hourly meso-analyses from the Storm Prediction Center (SPC) showed 0-1 km shear values of 25+ kt (Fig. 18a) extending from southern Ontario towards the Niagara Peninsula of western New York State.  Meanwhile, deeper-layered shear exceeded 50 kt over the same area (Fig. 18b).  In light of the above cited study, such observed shear magnitudes appeared to be more than sufficient to maintain the integrity of any linear convective elements and associated smaller scale vortices.

Fairly long-lived squall lines did, in fact, prevail on this day.  The earlier described swath of severe weather over southern Ontario continued downstream.  Strong winds, wind damage, and/or large hail affected numerous locations across western and central New York State well into the afternoon (Fig. 19).  We’ll look at some of the mesoscale aspects of this event and give a closer radar perspective in the next section.

3. Brief Mesoscale Description and WSR-88D Perspective

The leading edge of the gust front associated with the main squall line crossed into western New York over Niagara county around 1500 UTC.  It was near this time that the convective system, perhaps, reached its peak intensity.  Low-level velocity returns of greater than 60 kt were observed from the KBUF WSR-88D (Fig. 20).  As stated in the introduction, very strong surface winds were experienced throughout this area.  Officially, a wind gust of over 90 mph occurred at the Niagara Airport.  A small sampling of some of the damage inflicted is shown in Fig. 21.

As discussed near the end of section 2, the strong vertical shear in place seemed to be an important contributor in maintaining the gust front intensity over an extended time period.  Another factor may have played a supporting role, as well.  Mesoscale surface analyses at 1600 UTC (Fig. 22) showed the development of an inverted trough, which extended from near Buffalo, NY (KBUF) eastward to just south of Syracuse, NY (KSYR) and Utica, NY (KUCA).  Such a trough is frequently observed in easterly or southeasterly flow situations across western and central New York State.  This troughing is likely a result of downslope flow coming off the Allegheny Plateau and on to the comparatively flat terrain south of Lake Ontario.  Since the track of the MCS across New York State was essentially normal to the orientation of the trough axis, it is possible that the convergence/surface vorticity associated with the trough could have enhanced the lift near the leading edge of the gust front. 

Occurrences of severe weather continued to be associated with the MCS until just past 1800 UTC (refer back to Fig. 19).  The most significant damage within this specific time period occurred over Cayuga county in central New York, where storm surveys later revealed that an F0 tornado briefly touched down.  Some radar images from the KBGM WSR-88D are displayed close to the time of tornado development and associated damage photos are also shown (Fig. 23 and Fig. 24). 

At last, between 1800 and 2100 UTC on the 18th, convection began to wane in intensity as it approached the Catskill Mountains in eastern New York State.  By this time, shear magnitudes had started to decrease.  An initial-hour RUC sounding (Fig. 25) at 1900 UTC, just ahead of the convective system, showed a remaining presence of dry air in the mid-levels, to potentially enhance downdraft strength.  However, this sounding also showed that 0-1 km shear values had weakened to 10-15 kt, while the deeper-layered shear (0-6 km) was down to about 40 kt.  Compare these values to those observed just a few hours earlier (Fig. 18a and Fig. 18b).  Afterwards, no additional severe weather occurred and thunderstorms continued to weaken.

4. Summary

An MCS first began to form near the Iowa/Minnesota border during the evening of April 17, 2004.  Due to favorable large-scale support, the convective system gradually organized itself during the night as it pushed eastward across the Great Lakes region.  Storms were locally severe during the overnight period, but severe reports consisted primarily of large hail.  The authors contend that this was due to the elevated nature of the thunderstorms, given the track of the MCS north of a surface warm front and within a relatively cool, stable lower-level environment.

Just before daybreak on April 18th, a sudden increase in the frequency of strong winds/severe wind reports occurred, despite the continued presence of lower-level stability.  The authors theorized that the interaction between the MCS and a pronounced mid-level dry punch facilitated the downward transfer of strong winds to the surface.  Increased DCAPE associated with the dry mid-level air was finally able to overcome the surface-based inversion and bring downdrafts to ground level.  Thereafter, a swath of wind damage occurred from extreme eastern Lower Michigan across southern Ontario, and then through western and central New York during the morning and early afternoon hours of the 18th.  Once the squall line/gust front became established, a favorable cold pool/shear relationship existed to increase the longevity of the system, given the strong low-level vertical shear in place.

Once the MCS reached western and central New York, the mesoscale environment may have further enhanced the strength of the squall line.  A west to east oriented surface trough developed on the morning of the 18th across upstate New York.  Since the track of the MCS was nearly normal to the orientation of the trough, extra low-level convergence and surface vorticity may have enhanced lifting at the leading edge of the convective line/gust front.

By late afternoon on the 18th, once shear profiles and synoptic scale uplift decreased, convection began to weaken across eastern New York State.

5. FIGURES


Figure 1.  Visible satellite image at 2100 UTC, April 17th.  Surface frontal analysis is in green, MSLP analysis is in tan, and the surface observations are in blue. <<<back>>>

Figure 2a.  300 mb analysis at 0000 UTC, April 18th. <<<back>>>

Figure 2b. 500 mb analysis at 0000 UTC, April 18th. <<<back>>>

Figure 2c.  700 mb analysis at 0000 UTC, April 18th.  <<<back>>>

Figure 2d.  850 mb analysis at 0000 UTC, April 18th.  <<<back>>>

Figure 3.  Eta initialized cross-section at 0000 UTC, April 18th.  The horizontal axis (x-axis) is drawn from eastern Kansas to northern Wisconsin.  Solid blue contours represent units of divergence and dashed blue contours represent units of convergence.  Wind barbs are depicted in green.  Omega is shaded, with warm colors showing ascent and cool colors showing descent. <<<back>>>

Figure 4.  IR satellite image from 0045 UTC, April 18th.  Surface frontal analysis is in tan, MSLP analysis is in blue, and surface observations are in yellow.  Positive cloud-to-ground lightning strokes are depicted with green pluses (+) and negative cloud-to-ground lightning strokes are depicted with orange minuses (-). <<<back>>>

Figure 5.  CONUS-scale composite WSR-88D imagery (2 km resolution) at 0600 UTC, April 18th.  <<<back>>>

Figure 6.  Summary of severe weather reports from 1200 UTC, April 17th until 1200 UTC, April 18th. <<<back>>>

Figure 7a.  IR satellite image at 0315 UTC, April 18th.  Surface frontal analysis is in tan.  Surface-based lifted indices are in yellow (positive values are shown with solid contours and negative values are shown with dashed contours).  Positive cloud-to-ground lightning strokes are depicted by orange pluses (+).  Negative cloud-to-ground lightning strokes are depicted by light blue minuses (-).  <<<back>>>

Figure 7b.  RUC initial-hour point sounding (using point A (labeled in black) from Fig. 7a) at 0300 UTC on April 18th. <<<back>>>
 
Figure 8a.  IR satellite image at 0600 UTC, April 18th.  Surface frontal analysis is in tan.  Surface-based lifted indices are in yellow (positive values are shown with solid contours and negative values are shown with dashed contours).  Positive cloud-to-ground lightning strokes are depicted by orange pluses (+).  Negative cloud-to-ground lightning strokes are depicted by light blue minuses (-). <<<back>>>

Figure 8b.  RUC initial-hour point sounding (using point A (labeled in black) from Fig. 8a) at 0600 UTC on April 18th. <<<back>>>

Figure 9.  RUC initialized cross-section at 0600 UTC, April 18th.  The horizontal axis (x-axis) is drawn from northern Missouri to Lake Superior.  Solid blue contours represent units of divergence and dashed blue contours represent units of convergence.  Wind barbs are depicted in green.  Omega is shaded, with warm colors showing ascent and cool colors showing descent. <<<back>>>

Figure 10a.  RUC initial-hour 700-500 mb lapse rates at 0600 UTC, April 18th.  Fig. 10b depicts these steep lapse rates persisting downstream along the path of the MCS. <<<back>>>

Figure 10b.  SPC hourly meso-analysis of 700-500 mb lapse rates at 1800 UTC, April 18th. <<<back>>>

Figure 11a.  IR satellite image at 0915 UTC, April 18th.  Surface frontal analysis is in tan.  Surface observations are in blue.  Positive cloud-to-ground lightning strokes are depicted by green pluses (+).  Negative cloud-to-ground lightning strokes are depicted by orange minuses (-). <<<back>>>

Figure 11b.  RUC initial-hour point sounding (using point A (labeled with a green star in the upper left of the figure, and by a black dot in  Fig. 11a) at 0900 UTC on April 18th. <<<back>>>

Figure 12.  CONUS-scale composite WSR-88D imagery (2 km resolution) at 1200 UTC, April 18th.  <<<back>>>

Figure 13.  Visible satellite image at 1215 UTC, April 18th.  Surface frontal analysis is in tan and the surface observations are in blue.  Positive cloud-to-ground lightning strokes are depicted by green pluses (+).  Negative cloud-to-ground lightning strokes are depicted by orange minuses (-). <<<back>>>

Figure 14a.  300 mb analysis at 1200 UTC, April 18th.  <<<back>>>
 
Figure 14b.  500 mb analysis at 1200 UTC, April 18th.   <<<back>>>

Figure 14c.  700 mb analysis at 1200 UTC, April 18th.  <<<back>>>

Figure 14d.  850 mb analysis at 1200 UTC, April 18th. <<<back>>>

Figure 15.  Water vapor imagery at 1215 UTC, April 18th.  Solid orange contours represent units of initialized vorticity from the 1200 UTC, 18 April run of the Eta model. <<<back>>>

Figure 16.  Observed sounding from Detroit, MI (KDTX) at 1200 UTC on April 18th. <<<back>>>

Figure 17a.  Initial-hour RUC relative humidity (RH) values (shaded) at 650 mb from 0300 UTC, April 18th.  Warm colors represent dry air (RH values of generally 30% or less) and cool colors represent moist air (RH values of generally 60% or greater).  Radar, satellite, and lightning analyses (not shown here) were used to depict the position of the MCS at this time. <<<back>>>
 
Figure 17b.  Initial-hour RUC RH values (shaded) at 650 mb from 0600 UTC, April 18th.  Warm colors represent dry air (RH values of generally 30% or less) and cool colors represent moist air (RH values of generally 60% or greater).  Radar, satellite, and lightning analyses (not shown here) were used to depict the position of the MCS at this time. <<<back>>>

Figure 17c.  Initial-hour RUC RH values (shaded) at 650 mb from 0900 UTC, April 18th.  Warm colors represent dry air (RH values of generally 30% or less) and cool colors represent moist air (RH values of generally 60% or greater).  Radar, satellite, and lightning analyses (not shown here) were used to depict the position of the MCS at this time. <<<back>>>

Figure 17d.  Initial-hour RUC RH values (shaded) at 650 mb from 1200 UTC, April 18th.  Warm colors represent dry air (RH values of generally 30% or less) and cool colors represent moist air (RH values of generally 60% or greater).  Radar, satellite, and lightning analyses (not shown here) were used to depict the position of the MCS at this time. <<<back>>>

Figure 18a.  SPC hourly meso-analysis of 0-1 km shear vectors at 1500 UTC, April 18th. <<<back>>>

Figure 18b.  SPC hourly meso-analysis of 0-6 km shear vectors at 1500 UTC, April 18th. <<<back>>>

Figure 19a.  Summary of severe weather reports from 1200 UTC, April 18th until 1200 UTC, April 19th.  <<<back>>>

Figure 19b.  Zoomed-in severe weather summary across southern Ontario and New York State from 1200 to 1800 UTC on April 18th. <<<back>>>

Figure 20a.  A 0.5o base reflectivity image from the KBUF WSR-88D just prior to 1500 UTC, April 18th. <<<back>>>

Figure 20b. A 0.5o base velocity image from the KBUF WSR-88D just prior to 1500 UTC, April 18th.   <<<back>>>

Figure 21.  Damage photos from Niagara county, NY on April 18th. <<<back>>>

Figure 22.  Visible satellite imagery from 1600 UTC, April 18th.  Surface frontal analysis is in tan, MSLP analysis is in yellow, and surface observations are in blue.  Positive cloud-to-ground lightning strokes are depicted by green pluses (+) and negative cloud-to-ground lightning strokes are depicted by orange minuses (-).  <<<back>>>
 
Figure 23a.  A 0.5o base reflectivity image from the KBGM WSR-88D just before 1700 UTC on April 18th.  <<<back>>>

Figure 23b.  A 0.5o storm relative motion (SRM) image is displayed from the KBGM WSR-88D just before 1700 UTC on April 18th.     <<<back>>>

Figure 24.  Damage photos from Cayuga county, NY on April 18th. <<<back>>>

Figure 25.  RUC initial-hour point sounding over central New York (just northeast of BGM) at 1900 UTC on April 18th. <<<back>>>