severe wind-driven hail events: dependence on...
TRANSCRIPT
SEVERE WIND-DRIVEN HAIL EVENTS: DEPENDENCE ON CONVECTIVE
MORPHOLOGY AND LARGER-SCALE ENVIRONMENT
NICHOLAS D CARLETTAˡ
Mentors: William A. Gallus Jr.ˡ, M. A. Fowle² and D. J. Miller³
Department of Geological and Atmospheric Sciences
Iowa State University, Ames, IAˡ National Weather Service Aberdeen, SD²
National Weather Service Duluth, MN³
ABSTRACT
Severe wind events and hail events are well covered in various studies
but relatively little study has been done on wind driven hail. Both
wind and hail can cause significant damage but together they can cause
even more damage. Thus a further examination of the storms that
produce these events is warranted. By comparing the times and
locations of reported severe hail and wind, the locations of wind driven
hail can be identified. These locations can be compared to the storm
morphologies identified in Gallus et al (2008) and Duda and Gallus
(2010) to determine a specific morphology associated with these
events. Severe storms producing either severe wind or hail are often
associated with high values of CAPE and strong shear. Identifying
how closely these parameters correlate to wind driven hail events can
assist the forecasting of these events. Determining which parameters
are present in wind driven hail events that are not present in just a
severe hail event, or just a severe wind event, could prove vital to
forecasting these events. Squall lines, as well as isolated cells and
clusters of cells, were the most frequent morphologies associated with
wind-driven hail. CAPE played a role in the strength of the wind-driven
hail events as higher CAPE resulted in larger hail and/or faster wind. If
an event occurs with a supercell, the hail will be larger and the wind
will be faster for wind-driven hail events and the CAPE will also be
higher in supercells.wwwwwwwwwwwwwwwwwwwwwwwwwwww
1. Introduction
A wind-driven hail event is defined here as an
event where severe hail greater than 1 inch in
diameter occurs simultaneously with severe
wind in excess of 50 knots. Prior studies have
examined both the near-storm environments
(e.g. Rasmussen and Blanchard 1998) and
convective morphologies, most likely to be
associated with either severe wind events or hail
events (e.g., Duda and Gallus 2010), but
relatively little work has been done to
understand the conditions favorable for severe
wind-driven hail events. A study that did look at
simultaneous wind and hail was Lemon and
Parker (1996). In that study they made mention
of the importance of shear in the presence of
severe wind in a storm. Mention is also made to
a strong updraft being required for large hail to
be present. Some recent cases, such as the
Eldora Iowa event in August 2009 where hail of
3 inch diameter was blown by winds exceeding
90 knots, have demonstrated how tremendously
damaging and dangerous a thunderstorm can be
when severe wind and large hail occur together.
The present study uses the approach followed in
Gallus et al. (2008) and Duda and Gallus (2010)
to examine the convective morphologies
associated with severe wind-driven hail events.
In addition to this analysis of these events,
CAPE (Convective Available Potential Energy),
0-2 km SRH (Storm Relative Helicity), and 0-3
km SRH present for each event was also
examined. The events are also compared to 30
wind-only, 30 hail-only, and 14 wind-driven
hail events from the National Weather Service
at Aberdeen, South Dakota that have occurred
over the past decade, hereafter called known
events. This all leads to the hypothesis of this
study: Wind-driven hail events occurred most
frequently with certain storm morphology and
these storms had high values of CAPE and
SRH, as expected of a high end severe event
like wind-driven hail.
2. Methods
The present study follows the approach of Duda
and Gallus (2010), using storms from 2007 and
assigning morphologies based on Gallus et al.
(2008). The morphologies used were IC –
isolated cells, CC – cluster of cells, BL –
broken line of cells, NS – squall line with no
stratiform precipitation, TS – trailing stratiform
precipitation, PS – parallel stratiform
precipitation, LS – leading stratiform
precipitation, BE – bow echo, and NL – non-
linear convective system with storm motion
from left to right. Examples of these
morphologies are illustrated in Fig. 1. Unlike
the earlier studies that looked at flooding,
tornadoes, hail, and wind alone, the present
study focuses on severe wind-driven hail events.
To determine a wind-driven hail event, the
National Climatic Data Center’s (NCDC) Storm
Data publications were used. Since Duda and
Gallus (2010) used 2007’s storm reports and
classified the morphologies of all convective
systems in a 10-state region from April through
August, that same dataset was used for this
study. Wind-driven hail events were noted if
severe hail and severe wind were reported
within five miles of each other and within the
Figure 1: The morphologies used in the present study, taken from Duda and Gallus (2010)
same thirty minute period. In 2007, 0.75 inch
diameter hail was used to define a severe
hailstone, but that minimum criterion has
changed, so we used the new one inch diameter
criterion.
In all, 69 wind-driven hail events were found in
the 10 state region during the 2007 warm
season, occurring on 33 different days. Then
for the 2002 season, in the same region over the
same time period as 2007, 69 wind-driven hail
events were found over 38 different days. After
this dataset of cases was built, the location and
date of the events were compared to the dataset
of convective morphologies from Duda and
Gallus (2010) to assign a morphology. The
Duda and Gallus (2010) study also mentioned if
a storm was a supercell, so a note was made if
the event matched a supercell, as well. For the
69 2002 cases, data morphologies were from
Gallus et al (2008) with the same methods
above applied.
In both the Duda and Gallus (2010) and Gallus
et al (2008) studies, morphologies were
assigned based on date and state(s). For cases
where multiple morphologies were found at one
date in one state, the wind and hail reports listed
were used to identify the true morphology. If
there were still multiple events that had hail and
wind at a certain date and time, then the radar
composite archive provided by UCAR
(University Cooperation for Atmospheric
Research) was used online.
The next part of the project determined the
CAPE (Convective Available Potential Energy),
0-2 km SRH (Storm Relative Helicity), and 0-3
km SRH present, using RUC output from Iowa
State University’s online meteorology archive.
For each event a date, time, and location was
provided with the storm data so the
corresponding zero hour forecast closest to the
event was downloaded from the meteorology
archive. This data was then loaded in the
program nsharp, included in the GEMPAK
package of programs provided on the Iowa State
Meteorology computers. This program can read
the model upper air sounding data from the file
and compute CAPE, 0-2 km SRH, and 0-3 km
SRH. The latitude and longitude for the events
to be used in the nsharp program to obtain the
data were obtained with an online mapping tool.
In 2007, the online archive had RUC files every
three hours starting daily at 0000 UTC , while in
2002 there were RUC files every 6 hours
starting at 0000 UTC every day.
Then for comparison to the results from the
wind-driven hail cases, the previous methods
were used to obtain data of significant wind-
only and hail-only events in the 2007 warm
season period due to availability of archived
RUC model data. Events were taken from the
same NCDC storm data used for the previous
section. To qualify, the storm had to only have
wind or hail reports in it, not just wind and hail
either more than five miles apart or with thirty
minutes or more between them. The same
method explained above, with nsharp, was then
used.
Another point of comparison is the known
wind-driven hail events from the NWS.
Morphologies were determined with a
combination of the previously mentioned
UCAR image archive and the level II radar
archive that NCDC maintains. Then the method
with nsharp was used to determine CAPE, 0-2
km SRH, and 0-3 km SRH.
The last point of comparison used the hail and
wind breakdowns established in Gallus et al
(2008) and Duda and Gallus (2010) of hail one
to two inches and hail greater than two inches as
well as wind 50 knots to 65 knots and greater
than 65 knots. If a storm met the lower wind
speed and smaller hail category, it was
considered a category 1 event. If it met the
higher wind speed but not the larger hail it was
considered a category 2 event. If it met the
larger hail criteria but not the higher wind speed
it was considered a category 2 event. Last, if it
met both the higher wind speed and the larger
hail, it was considered a category 3 event. All
box charts were produced using the fit y by x
functions in the software package JMP.
Figure 2: Severe hail reports (sorted by size) as a function of convective morphology, taken from
Duda and Gallus (2010).
Figure 3: Severe wind reports (sorted by speed of wind in knots) as a function of morphology,
taken from Duda and Gallus (2010).
Figure 4: The percentage of convective morphologies in the data sample from 2002 and 2007 from Duda and Gallus (2010) on the
left, and the percentage of convective morphologies for wind driven hail events in 2002 and 2007 on the right.
3. Results
a. Wind and Hail from Duda and Gallus (2010)
In Duda and Gallus (2010) storm reports for
wind, hail, tornado, and flooding were classified
according to the convective morphology of the
system responsible for producing the report.
For large hail events, (Fig. 2) PS was the largest
producer in 2002 while BL was largest in 2007.
The number of NS events was much larger in
2007 than 2002. Otherwise, CC and IC also
contributed significantly to reports of hail larger
than two inches in diameter. For all hail in
2002, the top three morphologies were BL, BE,
and PS. In 2007 BL, CC, and PS were the top
three in 2007. For high wind events greater than
65 knots, BE events were largest in 2002
followed by TS. While in 2007, BE events again
were largest but followed by BL (Fig. 3). For
all severe wind events, BE were most followed
by TS, and both BL and PS were close.
b. Morphologies for 2007 and 2002 Wind-
Driven Hail
i) Morphologies for 2007
For cases of wind-driven hail, squall lines
contributed the greatest share, with TS and NS
combined accounting for 30% of all events. IC
and CC events also contributed substantially to
the total, with 19% and 16% respectively (Fig.
4). As for changes in percentage compared to
the 2007 non wind-driven hail events, TS had a
twelve point increase, NS a one point increase,
and IC a nine point increase. These results
imply squall lines have a relatively higher risk
of severe wind-driven hail compared to other
morphologies than they do for hail or wind
alone. On the other hand, CC events had a
seventeen percent decrease with BL and NL
having a one percent increase.
ii) Morphologies for 2002
For cases in 2002, clusters of cells and isolated
cells were more prevalent for wind-driven hail
events. Together they represent 58% of the
events in 2002. The average hail size and wind
speed in the 2002 data was greater, which may
be part of the reason for this difference in
morphologies. As far as changes from all the
events in 2002, CC had a seventeen point
increase, IC had a two point increase, and NS
had a two point increase (Fig. 4). All other
morphologies had decreases with the most
significant being for NL and PS, who both fell
by six points.
c. Comparisons
i) Comparing 2002 to 2007
The two years studied produced very different
storms with some similarities. For
morphologies, the difference between the two
seasons is ten points for TS, no change for NS,
no change for IC, twenty-three points for CC,
four points for PS, seven points for NL, six
points for BE, eleven points for BL, and one
point for LS. So over the two seasons, CC, IC,
and TS play the largest roles with combined
2002 and 2007 season percentages of 27%,
19%, and 18% respectively.
The values of CAPE and SRH between the two
years were very similar. For CAPE in 2002, the
Figure 5: Box plots of CAPE values from the 2002 and
2007 season. Also plotted is the set of base known cases as
well as cases in 2007 where only sever wind or only severe
hail was reported
Figure 6: Box plots of 0-3 km SRH values from the 2002
and 2007 season. Also plotted is the set of base known cases
as well as cases in 2007 where only sever wind or only severe
hail was reported
average value was 2678 J/kg and in 2007 the
average value was 2823 J/kg, but both years had
a standard deviation around 1500 J/kg. So both
years averaged a high amount of CAPE but with
significant variability in the data due to the large
standard deviations present. For 0-3 km SRH,
2002 averaged 124 m2/s
2 with a standard
deviation of 110.44 m2/s
2 and 2007 averaged
127 m2/s
2 with a standard deviation of 123.12
m2/s
2. Once again both values have a large
standard deviation because of a wide range of
values from negative to over 600 m2/s
2. As
shown in figures 5 and 6, the two years have a
similar spread of data and there is not a
statistically significant difference (Table A1).
ii) Comparing to Severe Wind-only and
Severe Hail-only Events
The wind-only events averaged CAPE of 2630
J/kg which is comparable to the values of 2002
and 2007 wind-driven hail. The hail-only
events averaged a higher 3066 J/kg of CAPE
and is also close to the average value in 2007
wind-driven hail. The average 0-3 km SRH,
though for wind-only cases, is 192 m2/s
2 and for
hail-only cases it is 116 m2/s
2. Comparing
wind-only and hail-only values to the report
based cases; only SRH between the wind-only
cases and the report based cases has a
statistically significant difference (Table A1).
iii) Comparing to known Wind-Driven Hail
Events
These known events all occurred within the past
ten years and are all in the Midwest. These
cases have an average CAPE of 2515 J/kg and
an average 0-3 km SRH of 225 m2/s
2 with once
again high standard deviations. These events
presented a CAPE similar to the 2002 cases and
not far from 2007 cases. The difference is the
SRH, which is around 100 m2/s
2 larger than
both the 2002 and 2007 average values. There
is a statically significant difference between the
known cases and the storm report based cases of
wind-driven hail (Table A1). Some of this can
be attributed to comparing the 14 cases from the
NWS to the 138 cases from storm reports.
iv) Comparing Parameters between
Morphologies
A change in CAPE and SRH was also present
across the different morphologies.. When
comparing these data sets, the greatest
differences were noticed with BE morphologies
that produced wind-driven hail events, because
they had statistically larger values of both 0-2
km and 0-3 km SRH than the other
morphologies with the exception of BL (Table
A1). The other difference was that the IC
morphologies had statistically higher CAPE
than the BE, NS, LS, and PS events. The BL,
NL, and IC events had the highest average
CAPE (Fig. 8). While the BE, BL, and CC
events had the highest SRH (Fig. 7). There is
variability in this though, since the highest
CAPE value was a CC event with 7905 J/kg
although the highest 0-3 km SRH was a BL
event with 602 m2/s
2.
Figure 7: Box plots of 0-2 km and 0-3 km SRH plotted for the different morphologies from both the 2002 and 2007 season.
The morphology of the event also seemed to
play a role in size of the hail and the speed of
the wind present. Figure 9 shows this with the
number of events for each category with respect
to hail size and wind speed across each
morphology. Based in figure 9, the CC, IC, and
TS events were the events that had a category
three event with hail over 2 inches and wind
over 65 knots. The CC, NL, IC, BL, TS, and
NS all had category two events with either hail
over 2 inches or wind over 65 knots. Category
one seems to follow a trend of the most frequent
occurrence of wind-driven hail events for the
2002 and 2007 data so that is insignificant.
d. Supercells and Dependence on Hail Size and
Wind Speed
i) Comparing Supercell Events to Non-
supercell events
Figure 10: Chart of morphologies associated the supercells
with wind-driven hail from the 2007 season.
Supercells play a role in wind-driven hail events
even though a supercell is not required for an
event to occur; however, it can increase the
Figure 8: Box plots of CAPE
plotted for the different
morphologies from both the 2002
and 2007 season.
Figure 9: Graph illustrating the
different number of cases for each
morphology that was a category one, two, or
three event based on hail size and wind
speed as described in the text above.
Figure 8: Box plots of
CAPE plotted for the different
morphologies from both the 2002
and 2007 season.
strength of the event. The average hail size for
a non-supercell is 1.26 inches and the average
wind speed is 55.53 knots. For a supercell, the
average hail size is 1.47 inches with wind of
60.16 knots. These differences are statistically
significant as well for the hail and the wind
(Table A2). Certain morphologies were also
more frequent for supercell events, CC had 38%
and BL had 31% of the events. . CC already
had a large amount of the cases, but BL did not
(Fig. 10). Following these two were IC (19%),
NL (6%), and BE (6%). So if a there is a
supercell, if it is A BL or a CC it has a better
chance of wind-driven hail occurring.
The size of the hail and speed of the wind were
not the only factors influenced by supercells,
because the CAPE was also higher. The
average CAPE for a supercell was 3140 J/kg
which is higher than the 2007 average of 2823
J/kg and the non-supercell average value of 258
J/kg. Through a t-test this was also proved
significant because when comparing supercells
with wind-driven hail to non-supercell with
wind-driven hail the value is 0.05. The SRH is
also higher for supercells with 147 m2/s
2 versus
109 m2/s
2. The CAPE is a significant difference
and the SRH is not (Table A1).
ii) Comparing Events based on Hail Size
and Wind Speed
The categories, as defined in the methodology,
have average CAPE of 2609 J/kg for one, 3188
J/kg for two, and 3487 J/kg for three. For 0-3
km SRH the average was 121 m2/s
2 for one, 141
m2/s
2 for two, and 119 m
2/s
2 for three. There is
a generally increasing trend in the value of
CAPE as the severity of the event increases but
this does not hold true with the SRH. This is
Figure 11: Box plots of CAPE and 0-3 km SRH from the 2007 season comparing supercell cases to non-supercell cases.
Figure 12: Box plots of CAPE and 0-3 km SRH from the 2002 and 2007 seasons separated based on the category determined by hail
size and wind speed.
supported by a t-test in which comparing the
CAPE of category three and two events to
category one events (Table A1). So CAPE can
be an indicator for the potential of more severe
wind-driven hail.
4. Conclusions
For severe wind-driven hail, both morphology
and CAPE seem to play some role in
determining the strength of the event. For all
the wind-driven hail events in the 2002 and
2007 seasons, the NS, TS, IC, and CC
morphologies were the most common. The
CAPE and SRH over these two periods were
very similar and both had a high variance. A
difference was noted, when the 2002 and 2007
cases were compared to the wind-only cases,
that there was a higher value for both the 0-2 km
and 0-3 km SRH in the wind-only cases. The
CAPE was higher for the hail-only events but it
was not statistically significant.
In comparing the different morphologies, they
seemed to have varying levels of CAPE and
SRH. The BL, NL, and IC morphologies
produced the highest values of CAPE. The BE,
BL, and CC events produced the highest amount
of SRH. Then the CC, IC, and NL events
produced the largest number of hail over 2
inches and/or wind over 65 knots.
Supercells with wind-driven hail produced
statistically larger hail and higher wind speeds
than cases that were not in a supercell. These
supercells were often CC, BL, or IC
morphologies. The CAPE of supercells were
larger but the SRH, while larger on average, it
was not statistically significant. For the
category breakdown of wind-driven hail events,
it was the CAPE that had a positive correlation
with the events. The CAPE had a statistically
significant difference between the top two
categories and category one, with the CAPE
getting larger as the hail size and wind speed
increased. CAPE in a number of instances
seemed to have an impact on the occurrence of
severe wind-driven hail and certain favorite
morphologies presented themselves. The high
value of variance in all of the data warrants
further research into this topic to see if these
points hold. For either, there is a high amount
of variability in wind-driven hail or there needs
to be another step beyond simultaneous wind
and hail reports at the same location to
determine an event.
5. Acknowledgements
I would like to thank all three of my mentors,
William Gallus, Michael Fowle, and Daniel
Miller, for their support and guidance on this
project. I would especially like to thank
William Gallus for helping me layout my plan
for this project and providing the morphology
data from Duda and Gallus (2010) and Gallus et
al (2008).
REFERENCES
Das, P., 1962: Influence of the wind shear on the
growth of hail. J. Atmos. Sci., 19, 407–
414.
Donavon, R. A. and K. A. Jungbluth (2007).
"Evaluation of a Technique for Radar
Identification of Large Hail across the
Upper Midwest and Central Plains of the
United States." Wea. Forecasting 22, 244-
254.
Duda, J. D. and W. A. Gallus (2010). "Spring
and Summer Midwestern Severe Weather
Reports in Supercells Compared to Other
Morphologies." Wea. Forecasting 25, 190-
206.
Gallus, W. A., Jr., E. V. Johnson, and N. Snook,
2008: Spring and summer severe weather
reports over the Midwest as a function of
convective mode: A preliminary study.
Wea. Forecasting, 23, 101-113.
Lemon, and S. Parker, 1996: The Lahoma deep
convergence zone: its characteristics, and
role in storm dynamics and severity.
Preprints, 18th
Conf. on Severe Local
Storms, Boston, Amer. Meteor. Soc., 70-
75.
Morgan Jr., G. M. and N. G. Towery, 1976: On
the role of strong winds in damage to
crops by hail and its estimation with a
simple instrument. J. Appl. Meteor., 15,
891–898.
Nelson, S. P., 1983: The influence of storm flow
structure on hail growth. J. Atmos. Sci., 40,
1965–1983.
Rasmussen, E. N., and D. O. Blanchard, 1998:
A baseline climatology of sounding-
derived supercell and tornado forecast
parameters. Wea. Forecasting, 13, 1148-
1164.
APENDIX A
CAPE 0-2 km SRH 0-3 km SRH
2007 to 2002 0.572603 0.97449736 0.89110934
2007 to hail-only 0.477844 0.9678285 0.70814482
2007 to wind-only 0.541664 0.00260713 0.01891734
2007 to known 0.497312 0.01730106 0.01740117
2002 to hail-only 0.257879 0.9470008 0.76915778
2002 to wind-only 0.879883 0.00145136 0.00899765
2002 to known 0.717265 0.01113094 0.00886651
cat3 to 2 hail 0.299293 0.2144603 0.19308014
cat3 to 2 wind 0.416776 0.47271931 0.45253635
2 hail to 2 wind 0.325067 0.14284887 0.14398163
cat3 to 2007 0.45388 0.92952655 0.91325496
cat3 to 2002 0.359376 0.91337438 0.9370357
cat3 to wind 0.270725 0.27331569 0.34282419
cat3 to hail 0.67937 0.95359017 0.97546847
cat3 to control 0.307617 0.37698678 0.35664867
2 hail to 2007 0.797966 0.27037759 0.25608803
2 hail to 2002 0.605733 0.22770241 0.1870065
2 hail to wind 0.519964 0.50051994 0.7899235
2 hail to hail 0.87997 0.37055904 0.21092231
2 hail to control 0.497451 0.51693008 0.52262651
2 wind to 2007 0.243846 0.96099316 0.98286383
2 wind to 2002 0.123701 0.93932857 0.9036966
2 wind to wind 0.122916 0.02261875 0.07335663
2 wind to hail 0.666651 0.99346976 0.7580156
2 wind to control 0.185726 0.06541757 0.06774443
BE to BL 0.141003 0.29253034 0.29004228
BE to CC 0.390067 0.0531993 0.06583375
BE to IC 0.095503 0.05058691 0.039641
BE to NL 0.197853 0.02438653 0.01537897
BE to NS 0.376949 0.0504234 0.07803075
BE to PS 0.321476 0.08577723 0.06366815
BE to TS 0.375305 0.07456396 0.07631107
BL to CC 0.140166 0.26849656 0.27211247
BL to IC 0.463928 0.1708542 0.13923156
BL to NL 0.404524 0.16427055 0.12209907
BL to NS 0.058189 0.24156555 0.27689468
BL to PS 0.099293 0.2597694 0.20812559
BL to TS 0.163132 0.27698864 0.28169573
CC to IC 0.103516 0.18462438 0.14551136
CC to NL 0.205155 0.15724273 0.10957496
CC to NS 0.239309 0.26554202 0.35382638
Table A1: P-values for all
comparisons with respect to CAPE and
SRH with green being less than 0.1 and
red being greater than 0.1
CC to PS 0.267565 0.1963697 0.14335798
CC to TS 0.473196 0.43284578 0.4063685
IC to NL 0.413441 0.46961592 0.43998047
IC to NS 0.027423 0.4561835 0.34696244
IC to PS 0.053457 0.38363146 0.32532431
IC to TS 0.123817 0.26697833 0.23753357
NL to NS 0.093895 0.41600004 0.26796521
NL to PS 0.14187 0.36586963 0.30130522
NL to TS 0.231182 0.23612252 0.18884107
NS to PS 0.406612 0.32399608 0.21706594
NS to TS 0.2286 0.33795975 0.43226179
PS to TS 0.255011 0.25550958 0.19988308
Y to N 0.051217 0.14426647 0.10857432
cat 3 to cat 2 0.377764 0.19230439 0.18430051
cat 3 to cat 1 0.08878 0.27482532 0.29063652
cat 2 to cat 1 0.032483 0.18600044 0.13739561
cat 3 to cat 2 wind 0.416776 0.24347163 0.23833047
cat 3 to cat 2 hail 0.299293 0.10881367 0.0945726
cat 2 wind to cat 1 0.021778 0.23394739 0.17655923
cat2 hail to cat 1 0.139263 0.18860821 0.1520781
hail wind
SC Hail and wind 0.078852 0.00455
Table A2: P-values for the comparison
of size of hail and speed of wind between
supercell and non-supercells.