An Assessment of Short-term Synoptic Air Mass Modification through Land-Atmosphere Interactions by Daniel J. Vecellio, B.S. A Thesis In Atmospheric Science Submitted to the Graduate Faculty of Texas Tech University in Partial Fulfillment of the Requirements for the Degree of MASTERS OF SCIENCES Approved Dr. Jennifer Vanos Committee Chair Dr. Eric Bruning Dr. David Hondula Mark Sheridan Dean of the Graduate School May, 2015
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
An Assessment of Short-term Synoptic Air Mass Modification throughLand-Atmosphere Interactions
by
Daniel J. Vecellio, B.S.
A Thesis
In
Atmospheric Science
Submitted to the Graduate Facultyof Texas Tech University in
Partial Fulfillment ofthe Requirements for
the Degree of
MASTERS OF SCIENCES
Approved
Dr. Jennifer VanosCommittee Chair
Dr. Eric Bruning
Dr. David Hondula
Mark SheridanDean of the Graduate School
May, 2015
Copyright 2015, Daniel J. Vecellio
Texas Tech University, Daniel J. Vecellio, May, 2015
ACKNOWLEDGEMENTS
I would like to acknowledge:
Dr. Jennifer Vanos, for bringing me into her research group early into her
career here at Texas Tech and a year into my graduate studies, an inopportune time
for the both of us. I’d like to thank her for all the knowledge she has imparted on
me and all the connections she has allowed me to make in the short year we have
worked together that will benefit me for a lifetime. And finally, I’d like to give her
my sincerest gratitude for providing the path that allowed me to reconnect with the
passion I had for atmospheric science and academia in general.
My two other committee members, Dr. Eric Bruning and Dr. David Hondula.
Dr. Bruning has not only been a resource for feedback on research ideas, but also a
constant help with questions and concerns ranging from coding to the research
process itself. It has all been truly appreciated. Dr. Hondula came onto my
committee without even knowing what he was getting into with me, but I hope that
he has not regretted the decision. Thank you, Dave, for all the help and I hope to
continue working with you in the future.
Trent Ford and Dr. Steven Quiring of Texas A&M University for all of their
help with soil moisture data and their comments which improved the methodology
employed in this study.
The Texas Tech Climate Science Center and, specifically, Ian Scott-Fleming for
help with data acquisition and manipulation.
The NOAA Air Resources Laboratory (ARL) for providing reanalysis data and
answering my numerous questions on running the HYSPLIT model.
The Earth Observing System at NASA for radiation data from the CERES
project.
ii
Texas Tech University, Daniel J. Vecellio, May, 2015
The International Research Institute at Columbia University for NOAA OI sea
surface temperature data.
Dr. Jon Nese, my undergraduate advisor at Penn State University and still my
academic mentor today. I’ve always taken something from each of our talks over the
past seven years in Walker Building. Thank you for all you did during my time in
the Happiest of all Valleys and even more so after you were rid of me.
Nick Smith, my officemate for two years and Aaron Hill, my roommate for one.
Thanks for dealing with my antics since we started in August of 2012 by either
going to grab a beer with me or simply telling me to shut up.
Tony Reinhart, for the reasons listed above as well as helping me out with
numerous computer issues throughout the past two years.
My two best friends from undergraduate studies, Greg Ferro and Simone
Gleicher, for our monthly Google Hangouts which were always a welcome break
from the world of constant work. I’ll meet you two at Cafe 210 for teas once this
thesis passes.
Kevin Horne, Ryan Beckler, Devon Edwards, Julia Kern, Jessica Tully and
Anna Orso who all befriended me soon after my return to State College in January
2012 and got me through what was certainly the most stressful and lost periods of
my life. #DMT and “Hey, Jude”, y’all.
All others I have written with throughout the years at Onward State and Black
Shoe Diaries, especially Davis Shaver, Chase Tralka, Eli Glazier, Evan Kalikow, Dan
McCool, Bill DiFilippo, Chris Grovich, Jeff Junstrom, Mike Pettigano, Jared
Slanina and Cari Greene.
Ginuwine, as without his musical masterpiece “Pony,” I may not have written a
single line of code correctly over the past two years.
My parents, who have never stopped believing in me since the day I was born.
iii
Texas Tech University, Daniel J. Vecellio, May, 2015
Whether it was making sure I had a TV Guide to read when I was two, getting me
to soccer or basketball practice during grade school or listening to my problems,
both academically and personally, throughout college, I don’t know what I would
have done without all the love you’ve provided.
The rest of the Texas Tech Atmospheric Science Group and the National Wind
Institute, everyone else that I have met while in Lubbock, the rest of my friends and
family in Bradford and State College and everyone else who has supported my
journey from home to Penn State to Florida State to not knowing what was going
to happen to my eventual landing spot here at Texas Tech.
iv
Texas Tech University, Daniel J. Vecellio, May, 2015
Pre-grouping trajectories by weather type may be a mitigating factor in the
effectiveness of clustering. Hondula et al. (2010) found that while each SSC type
had its own general flow pattern, there was still overlap between trajectory groups
of different SSC types, hence, knowing the SSC type at the trajectory’s terminal
location was not enough to know the source region of any individual parcel on a
27
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure 3.1: Map of Lexington, KY MT trajectories
daily time-scale. Hondula et al. (2010)’s study had a focus on a full back-trajectory
climatology with SSC typing of the trajectories being a complimentary component.
They even stated that they expected the overlap that they revealed in their
analysis. Due to the overlap, they reclassified trajectories, a step that was not taken
in this study as the pre-grouping by SSC type was the primary focus of the study.
The argument that the datasets were not large enough for significant natural
groupings to be disseminated is one that can be made when explaining the
clustering algorithm’s seemingly poor performance. The largest of this study’s
datasets contains 105 trajectories (Huntsville MT and Raleigh-Durham MT) while
Davis et al. (2010) and Hondula et al. (2010) used ten years worth of trajectories to
28
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure 3.2: Map of Oklahoma City, OK MT trajectories
compete their clusters using the same technique. Short-term climatological patterns
begin to become apparent over such a time period. There is less of a chance that
the current study’s relationships can represent any significant trend using only the
warm-season period over five years of record. However, the Oklahoma City MT
scenario discussed above presents two established trajectory patterns that appear
despite the limited amount of data (see Figure 3.2). Without knowing how many
trajectories are used, it is not unfathomable that someone may see a potential
climatology of air mass positions entering Oklahoma City. Due to this, strength is
lacking in the small sample size argument, although the technique has worked
satisfactorily in previous studies with more trajectories. Further analysis is needed
to explain the algorithm’s trouble with the current data.
29
Texas Tech University, Daniel J. Vecellio, May, 2015
3.1.2 Evapotranspiration Over Water
Values of maximum potential evapotranspiration (MPET) are found to be
anywhere between 10-1000+ kg m-2 for each 12-hour period (comparable to values
found in Borma et al. (2009) which investigated ET values over a floodplain in
Amazonia), which is intuitive as parcels of air over the ocean have a limitless water
supply to extract from. Such values relative to the land values from GLDAS-1 are
very high. GLDAS-1 values are, at most, on the order of 100 kg m-2. As a example of
this, sample data from each set of evapotranspiration data taken from text files can
be seen in Figure 3.3 for comparison. Negative values in Figure 3.3 can be ignored
as evaporative processes are not present due to the lack of solar radiation at night.
Figure 3.3. Sample evapotranspiration data from Huntsville DT trajectories. Theseventh column shows values taken directly from GLDAS-1 data. The eighth columnshows calculated values using the Priestley-Taylor equation.
Due to the fact that differences between land-based evapotranspiration and
30
Texas Tech University, Daniel J. Vecellio, May, 2015
MPET values can be as much as three orders of magnitude, statistical comparisons
between different trajectories are rendered meaningless as the direct, land-based
values become negligible and calculations are too heavily weighted on the MPET
values. Hence, calculated measurements of MPET were not used in this study
although they provided valuable insight into the controlling mechanisms of a large
body of water resulting in moist air masses on land.
***
The MATLAB clustering algorithm and process as well as the values of
evapotranspiration over water were only two of the methods that, although
hypothesized to provide critical information, were deemed unsuitable for analysis for
this study. These two hinderances and corresponding methods are listed here so
that future researchers addressing this topic are aware as they delve deeper into
these factors. If able to improve upon the thoughts presented here, they may create
a new methodology of their own that acknowledges the limitations if they choose to
attempt to include them in consequent studies.
3.2 Weather Type and Modification Frequency
The number of events for each of the fifteen location-weather type combinations
are listed in Table 2. Also shown are numbers for each modification scenario, listed
in column 1, based on the starting, 96-hour SSC weather type heading each column.
The main takeaway from the Table 2 is that air mass modification – examined using
the SSC – occurs more often than not, therefore, validating the need for
understanding the physical nature of these modifications within this research.
Modification of any air mass into a DT weather type is the most common change
when compared to MT and MT+ ending weather types. The highest percentage of
31
Texas Tech University, Daniel J. Vecellio, May, 2015
DT presence at both the start and end of a trajectory occurred in Lexington at
13.3% frequency. It is important to note that Lexington DT is by far the study’s
most limited dataset and that DT-to-DT presence at the other four target locations
was well under 10%. MT weather type presence at both the start and end of 96-hour
back trajectories was far more common than that of DT (19.8% at Wilmington
being the minimum and and 48.6% at Oklahoma City being the maximum), but
over the five target locations, modification still occurred more than half of the time.
MT/MT+ and DM are the most common starting SSC weather types across
the fifteen scenarios studied. The southern cities close to the coast (Huntsville and
Raleigh-Durham) each see 34% of their resulting DT-type associated air masses
start off in the moist tropical regime. This number nearly doubles for Oklahoma
City DT cases ( 64%), most likely caused by air from the Gulf of Mexico reaching
the city due to the climatological presence of anticyclones across the lower Midwest
and Ohio River Valley during the warm season (Harman, 1987). This allows for
southeasterly flow to develop from the Gulf into Oklahoma City. DM sources
represent a moderate percentage of events for all but the Oklahoma City cases. As
discussed in Section 2.1, DM weather types have no distinct source region, but are
rather normally modified themselves from a number of previous weather types. The
most common way for a DM weather type to become present – through west-to-east
flow descending off of the Rocky Mountains – would give credence to the fact that
DM is present ninety-six hours ahead of time for the events in this dataset that are
well off to the east in locations, such as Huntsville and Raleigh-Durham. This also
explains the relative lack of DM presence in the previous 96 hours of Oklahoma City
events given its relative vicinity to the lee of the Rockies.
32
Texas Tech University, Daniel J. Vecellio, May, 2015
Table
2:
Fre
quen
cyof
modifi
cati
onfo
rea
chSSC
wea
ther
typ
eat
each
loca
tion
.E
ndin
gSSC
wea
ther
typ
esan
d
loca
tion
are
list
edin
each
row
.Sta
rtin
gSSC
wea
ther
typ
ear
elist
edin
each
colu
mn.
(‘T
otal
’M
TP
valu
eis
asu
bse
tof
‘Tot
al’
MT
valu
e)(H
SV
-H
unts
ville
,IG
L-
Wilm
ingt
on,
LE
X-
Lex
ingt
on,
RD
U-
Ral
eigh
-Durh
am,
OK
C-
Okla
hom
a
Cit
y).
DM
DP
DT
MM
MP
MT
MT
PT
RN
AT
otal
HSV
DT
1432
.56%
24.
65%
24.
65%
613
.95%
00.
00%
1227
.291
%3
6.98
%1
2.33
%3
43M
T27
25.7
1%1
0.95
%5
4.76
%13
12.3
8%3
2.86
%32
30.4
8%9
8.57
%8
7.62
%7
105
MT
P2
5.41
%1
2.70
%2
5.41
%2
5.41
%0
0.00
%19
51.3
5%8
21.6
2%3
8.11
%0
37
IGL
DT
618
.75%
825
.00%
26.
25%
412
.50%
39.
38%
39.
38%
13.
13%
26.
25%
332
MT
2930
.21%
66.
25%
44.
17%
1717
.71%
44.
17%
1919
.79%
11.
04%
44.
17%
1296
MT
P5
16.6
7%0
0.00
%1
3.33
%2
6.67
%0
0.00
%14
46.6
7%4
13.3
3%2
6.67
%2
30
LE
XD
T4
26.6
7%1
6.67
%2
13.3
3%0
0.00
%3
20.0
0%3
20.0
0%0
0.00
%1
6.67
%1
15M
T15
17.6
5%11
12.9
4%2
2.35
%15
17.6
5%2
2.35
%27
31.7
6%3
3.53
%5
5.88
%5
85M
TP
421
.05%
15.
26%
210
.53%
210
.53%
00.
00%
736
.84%
15.
26%
15.
26%
119
RD
UD
T12
20.6
9%8
13.7
9%2
3.45
%5
8.62
%3
5.17
%15
25.8
6%5
8.62
%2
3.45
%6
58M
T11
10.4
8%11
10.4
8%5
4.76
%14
13.3
3%2
1.90
%32
30.4
8%10
9.52
%8
7.62
%12
105
MT
P5
10.4
2%3
6.25
%1
2.08
%6
12.5
0%0
0.00
%20
41.6
7%8
16.6
7%4
8.33
%1
48
OK
CD
T3
7.69
%1
2.56
%1
2.56
%3
7.69
%2
5.13
%17
43.5
9%8
20.5
1%2
5.13
%2
39M
T11
15.7
1%6
8.57
%3
4.29
%2
2.86
%2
2.86
%34
48.5
7%7
10.0
0%4
5.71
%1
70M
TP
13.
57%
00.
00%
517
.86%
27.
14%
00.
00%
1242
.86%
725
.00%
00.
00%
128
33
Texas Tech University, Daniel J. Vecellio, May, 2015
In addition to the quantitative nature of this air mass modification study based
on evapotranspiration, a qualitative case-study is performed using one of the fifteen
city-weather type scenarios to determine if the synoptic environmental factors that
play an important role in the air mass’ modification. The goal of this case-study is
to examine the indirect factors outside of the direct path of the calculated trajectory
that impact the modification of the air mass encapsulated by the single trajectory.
As described in the frequency table at the beginning of this chapter, there were
forty-three instances of Huntsville, Alabama incurring a warm-season, DT weather
type event of one or more days during the period of study. Of these forty-three
instances, only two starting weather types made up more than 20% of modified air
masses based on weather typing: DM (32.56%, 14 instances) and MT/MT+
(34.88%, 15 instances) – refer to Table 2. It is deemed unlikely any patterns in
synoptic conditions would be valid given the limited dataset of the other starting
weather-type sub-scenarios. Thus, this case study focuses on the DM and MT
sub-scenarios alone. Archived daily synoptic weather maps made available by
NOAA’s Weather Prediction Center (WPC) are compared with HYSPLIT back
trajectories from each event and used to diagnose the synoptic setup during each
five-day event leading up to the presence of DT conditions in Huntsville.
3.3.1 MT-to-DT Modification
In Huntsville, there were fifteen instances of moist-tropical-to-dry-tropical
weather type modification, where moist tropical includes both MT and MT+
weather types. Certain synoptic patterns are apparent at points during the event’s
duration. Of these fifteen events, thirteen of them can be split into one of two main
patterns that emerged. The first of these patterns is characterized by a center of
34
Texas Tech University, Daniel J. Vecellio, May, 2015
high pressure either moving into or building up in the southeastern United States.
The second is distinguished by a frontal passage during the event, still leaving high
pressure in its wake, but with a weak pressure gradient associated with it. Only two
of the events were qualitatively classified as outliers and not addressed in this case
study.
3.3.1.1 Southeastern U.S. High Pressure Center
The first of these synoptic patterns is characterized by a high pressure center
taking hold in the general vicinity over the Southeastern United States by the end
of the four-day modification event, such as that displayed in Figure 3.4. This
synoptic setup takes place seven times in this dataset. The event depicted in Figure
3.4, which took place between the days of June 4-8, 2008, provides a textbook
example of the case. On Day 0 of the event (-96 hours), a low pressure system and
associated front is present over the midwestern United States, while a large high
pressure system is situated well off the east coast with only a portion of the closed
isobar present in this surface analysis. As the event progresses, the high pressure
system over the Atlantic retrogrades back towards the east coast of the United
States, eventually forming a 1020-millibar closed center that settles over the
Georgia/South Carolina border. This occurs on Day 4 of the event and brings
relatively calm conditions to the Huntsville area, providing conditions for
stagnation. This pattern is also seen during the MT-to-DT modification event of
July 19-23, 2010 (Figure 3.5). In this case, while a completely-closed-off high
pressure center is not present in the surface analysis presented, a large high pressure
system is present in the southeastern United States for the entire five-day period,
shifting its position daily, but always affecting the region.
A hypothesis can be formed from these sub-scenario results on how synoptic
35
Texas Tech University, Daniel J. Vecellio, May, 2015
conditions may dry out the air mass once it arrives in Huntsville. When a center of
high pressure is situated over the southeastern United States, the clockwise flow
around it brings air from the southern Texas and Mexico region into the southeast.
Winds from Huntsville rarely come from the southwest when DT conditions are
present (winds are commonly southerly or southeasterly on Day 4 at Huntsville),
but the wind direction at stations to Huntsville’s west normally have a
southwesterly component on Day 3 or 4. Therefore, advection of drier air from
common DT source regions (i.e. the desert Southwest and Mexico) into the
Huntsville region are postulated to be partially the reason for DT conditions at
Huntsville on Day 4 when air mass trajectories begin and travel through MT source
regions, as seen in Figures 3.4 and 3.5.
3.3.1.2 “No Man’s Land” High Pressure Presence After Frontal Passage
The second of these synoptic patterns found in the case-study for MT-to-DT
modification is classified as a “No Man’s Land” situation in Huntsville as high
pressure is present, but there is no substantial center or gradient of the measured
high pressure in the region based on the surface analysis. The high pressure is found
to frequently follow a cold frontal passage through the region which provides a clue
into how the region may dry out as the originally moist parcel of air makes its way
to the target location. This sub-scenario of the MT-to-DT modification occurred six
times out of fifteen within the five-year dataset at Huntsville.
Figure 3.6 shows an example of this situation spanning between August 2-6,
2008. At the beginning of the period, a low pressure system was located over the
Northeastern United States with its associated cold front stretching through the
Ohio River Valley and into the midwest. The front becomes stationary by Day 2
(August 4) and moves through the Southeastern United States on Day 3. High
36
Texas Tech University, Daniel J. Vecellio, May, 2015
pressure builds soon afterwards which is strong enough at mid- and upper levels to
push Tropical Storm Edouard (Brown et al., 2010), which was occurring and located
in the Gulf of Mexico at that time, towards Louisiana and eventually Texas. By
Day 4, centers of high pressure are found over western Canada as well as the
upper-midwestern United States. While a prototypical center was not found in the
southeastern United States, Huntsville’s pressure still read near 1020 millibars on
August 6, comparable to the center over the midwest at the time.
A similar event is depicted in Figure 3.7, which took place between the dates of
July 28-August 1, 2011. An initial stationary front located over the northern United
States began sliding southward as a cold front on Day 1 (July 29) of the event. This
continued until it was positioned along the southeastern seaboard at the end of the
period. A disorganized high pressure system was located behind the advancing front
which once again left Huntsville in this area of high pressure with no gradient
present.
The hypothesized subsequent drying-out of the Huntsville region in this
sub-scenario of the MT-to-DT modification is also part of the basis of synoptic
meteorology. Cold frontal passages typically bring drier air from the north along
with it. In the six events in this subset of the MT-to-DT modification data, a cold
front made its way through the Huntsville area at some point in or right before the
ninety-six hour event period. That synoptically-forced phenomenon, coupled with
stagnant conditions after the front’s passage through the region, not allowing for
new, moist air to advect into the area, is a simple hypothesis for partial reasoning
for the presence of DT conditions by Day 4 when MT air is modified to DT in
Huntsville.
***
37
Texas Tech University, Daniel J. Vecellio, May, 2015
While briefly touched upon and somewhat assumed above, it is important to
note that low wind speeds at Huntsville on Day 4 are consistently prevalent
throughout the case-study. In the thirteen cases described in the two sub-scenarios
of MT-to-DT modification discussed above, the highest wind speed present at
Huntsville on Day 4 was eight miles per hour, most of the cases having recorded
wind speeds between 3-5 miles per hour. These wind speeds are certainly light
enough for stagnation of air, effectively maintaining dry conditions once air of that
character enters the region (either by southwest flow or a frontal passage, as
exhibited).
3.3.2 DM-to-DT Modification
The DM-to-DT weather type modification story is very different from that of
MT-to-DT modification as a temperature change (rather than moisture) is the
variable of interest. In North America, temperature normally has a strong
latitudinal dependence. This train of thought is confirmed in the trajectories that
comprise the DM-to-DT sub-scenario. Of the fourteen events examined, nine have
air parcel trajectory paths that traverse into Huntsville from the north. Another
four of the events start at a latitude comparable to Huntsville’s, leaving only one
outlier that started at a position deep in the Gulf of Mexico. The outliers become
an MT after 12 hours and experiences an environment much like the one described
in Section 3.3.1.1.
Apart from modification relative to the MT-to-DT sub-scenario, where events
(advection and frontal activity) outside of the parcel’s path play a large part in the
modification, there is a synoptic story present in the DM-to-DT events. In 13 of 14
DM-to-DT modifications, air parcel trajectories follow the path around an
anticyclone present in the eastern half of the continent into the target location of
38
Texas Tech University, Daniel J. Vecellio, May, 2015
Huntsville. While the anticyclone is always present, air parcel trajectories have no
common source region. Within the dataset, air parcel trajectories begin at locations
such as British Columbia and Ontario, Canada, Washington, Minnesota, Iowa and
Wisconsin to the north, the Gulf of Mexico to the south and even Alabama itself.
The fourteenth case, what can be considered an outlier, involved 2008’s Hurricane
Kyle forcing an air parcel, well ahead of the center of the storm, into the target
around its vast low pressure system.
Examples of this synoptic pattern are shown in Figures 3.8 and 3.9. Figure 3.8
depicts an event which took place between May 2-6, 2008. On Day 0, a low-pressure
system is located over the midwestern United States while a high pressure system is
starting to stretch into the Rocky Mountain region of the United States. Over the
four-day period, the large North American anticyclone center moves into the
Midwest and eastward until covering the eastern third of the United States by Day
4. The air parcel follows along the northern and eastern side of the anticyclone on
its entire path to Huntsville from western Canada across the Rockies and Midwest.
The random assortment of source regions for DM-to-DT modification is
apparent when comparing the previous event to the August 15-19, 2008 event shown
in Figure 3.9. Widespread high pressure from the east coast to the Rocky
Mountains is present throughout the four-day period. A distinct anticyclone center
is not present until Day 4 when it takes hold over the Great Lakes, but weak
clockwise circulations are found throughout the event’s period. The air parcel
associated with this event makes a much shorter trip into Huntsville when compared
to the previous event. It starts in South Carolina and rides along the weak pressure
gradient at the lower periphery of the vast high-pressure system that is present
during the ninety-six hour period.
39
Texas Tech University, Daniel J. Vecellio, May, 2015
3.4 Effect of Evapotranspiration on Modification
To quantitatively describe the process of air mass modification, specifically in
the framework of the moisture content of an air parcel, evapotranspiration along
trajectory paths are examined. Once it was established that evapotranspiration over
bodies of water would unevenly weight results towards trajectories that traversed
water at some point during their journey to a target location (Section 3.1.2), a novel
method to compare evapotranspiration values was surmised. For each city and each
modification scenario, trajectories are split into two groups: a group denoted by
trajectories being over land and having evapotranspiration values throughout the
entire 96-hour period (a total of nine values along the path), and a group of
“partial” trajectories. For these partial trajectories, the evapotranspiration values
from hour 0 (the target location) and stepping back every twelve hours are used in
analysis. However, partial trajectories are denoted as being found over water at
some point in the five days. Evapotranspiration values (every twelve hours) up until
the point where the trajectory is over water are used for analysis. For example, for
the trajectory presented in Figure 3.6, evapotranspiration values from hour 0
through hour 48 are considered valid for analysis as the trajectory is over land for
these time steps before it moves over the Gulf of Mexico.
Based on the partial and full groupings, the simple hypothesis is formed: For
instances of air mass modification where the characteristics of a weather type’s
moisture parameter is switched (i.e. moist to dry or vice versa), there should be a
noticeable difference between the groups’ average evapotranspiration values. For
example, in a moist-to-dry modification, it is predicted that the partial trajectory
group has a much lower average evapotranspiration than the full trajectory group
due to the fact that the partial trajectory spent time over water previously where
the intake of moisture into the air parcel was much greater than an parcel that only
40
Texas Tech University, Daniel J. Vecellio, May, 2015
traveled over land.
Table 3.1 displays average evapotranspiration values for both full and partial
trajectories for each of the five target locations during MT-to-DT modification
scenarios. The final row of the table displays the five-city average of those average
evapotranspirations. In order to sufficiently dry out the air parcel to become a DT
weather type relative to each target location, one would expect the partial
trajectory average to be much lower than its full counterpart according to our
hypothesis. In this modification scenario, partial trajectory evapotranspiration
values are lower than their full trajectory counterparts in Huntsville (slightly),
Wilmington and Oklahoma City. Taking the average ET over the five cities, average
full trajectory evapotranspiration comes out to be 1.16 kg m-2 and partial trajectory
evapotranspiration is 1.07 kg m-2. While the partial group is lower by 0.09 kg m-2,
the difference between the two groups is two orders of magnitude lower than the
values themselves and not statistically significant. This result does not agree with
the hypothesis of a large difference between the two values.
Table 3.1. Average evapotranspiration (kg m-2) values for full and partial trajectoriesin MT-to-DT modification scenarios.City Full Avg. Amt. Partial Avg. Amt. P-value
Huntsville, AL 1.34 3 1.28 12 0.86Wilmington, DE 1.69 2 1.28 2 0.38Lexington, KY 0.75 2 1.05 1 –Oklahoma City, OK 0.89 3 0.60 22 0.39Raleigh-Durham, NC 1.13 7 1.15 13 0.91
Five-City Average 1.16 17 1.07 50 0.68
The opposite situation is examined with results displayed in Tables 3.2 and 3.3.
Both DT-to-MT and DM-to-MT modification scenarios are examined due to the
41
Texas Tech University, Daniel J. Vecellio, May, 2015
minimal amount of data contained in the former’s scenario. The results in both
analyses agree closer to the initial hypothesis stated, showing full trajectories to
have higher evapotranspiration values on average than their partial counterparts.
On a city-by-city basis, the only scenario that does not fit the trend is the Huntsville
DT-to-MT modification scenario. The five-city averages also agree with the given
hypothesis as the differences between the full and partial trajectory groups are 0.28
and 0.42 kg m-2 for DT-to-MT and DM-to-MT scenarios, respectively.
Table 3.2. Average evapotranspiration values (kg m-2) for full and partial trajectoriesin DT-to-MT modification scenarios.City Full Avg. Amt. Partial Avg. Amt. P-value
Huntsville, AL 1.24 3 1.38 2 0.77Wilmington, DE 1.43 2 0.79 2 0.61Lexington, KY 1.57 2 0.00 0 –Oklahoma City, OK 0.78 1 0.48 2 –Raleigh-Durham, NC 1.36 3 1.36 2 0.94
Five-City Average 1.28 12 1.00 8 0.15
Table 3.3. Average evapotranspiration values (kg m-2) for full and partial trajectoriesin DM-to-MT modification scenarios.City Full Avg. Amt. Partial Avg. Amt. P-value
Huntsville, AL 1.28 12 0.89 15 0.09Wilmington, DE 1.60 8 0.75 21 0.01**Lexington, KY 1.23 12 1.13 3 0.53Oklahoma City, OK 1.20 5 0.71 6 0.12Raleigh-Durham, NC 1.07 5 0.81 6 0.47
Five-City Average 1.28 42 0.86 51 0.01**
42
Texas Tech University, Daniel J. Vecellio, May, 2015
Additional evapotranspiration figures may be seen in the Appendix A of this
document.
43
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure 3.4. Surface analyses from Day 0-4 of event taking place June 4-8, 2008. A:Day 0. B: Day 1. C: Day 2. D. Day 3: E. Day 4: Subfigure F shows the trajectoryinto Huntsville for the four-day event. (Source: NOAA WPC)
44
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure 3.5: Same as Figure 3.4 but for July 19-23, 2010 event.
45
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure 3.6: Same as Figure 3.4 but for August 2-6, 2008 event.
46
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure 3.7: Same as Figure 3.4 but for July 28-August 1, 2011 event.
47
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure 3.8: Same as Figure 3.4 but for May 2-6, 2008 event.
48
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure 3.9: Same as Figure 3.4 but for August 15-19, 2008 event.
49
Texas Tech University, Daniel J. Vecellio, May, 2015
CHAPTER 4
DISCUSSION AND CONCLUSIONS
4.1 Synopsis of Results
This project is meant to be an establishment of the methods that can be used
in future air mass modification projects. The results show that the use of
evapotranspiration is not a strong determinant in air mass modification analyses as
hypothesized. A reason weak relationships are found is due to the relativity of the
Spatial Synoptic Classification system and its contrast with the absolute nature of
evapotranspiration. As stated in Sections 1.3 and 2.2, the SSC is a spatial- and
temporal-relative weather typing system that bears unique quantifying
characteristics for each reporting station at different times of the year.
Evapotranspiration, on the other hand, is dependent on three separate factors:
1. Temperature, which can be described as spatially- and temporally-relative,
but not on the scales of one weather station or the two-week stepping that is
used to develop the SSC. There is a latitudinal dependence and seasonality
encompassed in the variability of temperature, but over larger time-steps and
spatial areas than the SSC.
2. Radiation, which once again has a latitudinal dependence, but may also vary
daily based on cloud cover. It is a variable that also has a dependence on land
use and land cover which may change over time.
3. Soil moisture, which is dependent on precipitation which is a highly variable
process, not leading to any relativity.
The MT-to-DT hypothesis presented in Section 3.4 failed as there was not a
statistically significant difference between average ET values for the full and partial
50
Texas Tech University, Daniel J. Vecellio, May, 2015
trajectory groups. While DM-to-MT and DT-to-MT results given in Section 3.4
accurately describe and confirm the hypothesis stated in Section 3.4 on the surface,
less importance should be placed on results pertaining to dry-to-moist
modifications. Partial trajectories in these scenarios reach water and ingest a large
amount of moisture before returning to land where this average evapotranspiration
analysis begins. This is opposite of what was thought in the previous moist-to-dry
analysis where the evapotranspiration, or lack thereof, of a parcel moving from
water to land was quite important in the drying-out process of the air mass. With
this in mind, due to the lackluster results seen in moist-to-dry modification
scenarios as detailed in Section 3.4, it is stated that using evapotranspiration as a
quantitative measure of air mass modification is weak and inconclusive.
An alternate surface moisture parameter that could be considered is discussed
in the next section.
However, this project has shown that the relativity of the Spatial Synoptic
Classification system has both its advantages and disadvantages, the former having
been discussed primarily to this point with regards to its use in applicative studies.
The spatial and temporal relativity is a unique feature allows an individual station
to have certain criteria distinguish its classification. However, when comparing two
stations (i.e. the target location and the location ninety-six hours before arrival at
the target location), the distinguishable criteria can become lost in the translation.
For example, take an air mass in the month of June that begins in Bismarck, North
Dakota as a moist tropical weather type and traverses to Huntsville, Alabama in
ninety-six hours where it is classified as dry tropical. Based on classification, it is to
be expected that the parcel dried out as it made its way south. However, in June, a
Bismarck MT has a characteristic dewpoint of 62 degrees Fahrenheit while a
Huntsville DT has a characteristic dewpoint of 61 degrees Fahrenheit. In a relative
51
Texas Tech University, Daniel J. Vecellio, May, 2015
sense, the air mass is moist for its location in Bismarck while it is dry in Huntsville.
However, in the absolute sense, the change in the moisture characteristics of the
parcel is almost negligible which brings the use of the term “modification” into
question which, in turn, casts doubt on the SSC being the best option for weather
station comparison.
There is no question that to most accurately describe air mass modification,
both qualitative and quantitative methods must be undertaken. The atmosphere is
a chaotic mechanism and while scientists have been able to describe it adequately
with a set of equations, they are riddled with unrealistic assumptions, namely
isolating a parcel of air from the rest of the atmosphere and treating it separately, a
process exhibited by the output trajectories from the HYSPLIT model. From the
results of the case studies performed in this research, it is apparent that the
surrounding environment has a large impact on the final state of an air mass in
conjunction with the air along its direct path to the target location. Advection of
additional air masses along other paths during the time period of any situation that
is being examined should be expected given the fluid medium that is the Earth’s
atmosphere. For instance, in Figure 3.6, a cold front that pushed down from the
north is hypothesized to be the agent that dried out conditions in Huntsville.
However, this is not apparent from the given trajectory that moves north from the
Gulf of Mexico and shows no sign of being affected by the frontal passage. The
trajectory does not show the full advective transport into the Huntsville area, but
merely the transport along one individual streamline. Additionally, the HYSPLIT
model does not account for any thermodynamic properties, leaving radiation fluxes,
adiabatic processes and sources and sinks of moisture (all sources of modification)
to be determined through other means.
52
Texas Tech University, Daniel J. Vecellio, May, 2015
4.2 Future Work
There are many directions for future work in this research area. First and
foremost, finding a variable or set of variables to help quantitatively describe air
mass modification should be a focus. While evapotranspiration seemingly provides a
dead end in modification analysis, using a moisture parameter to describe air mass
modification should not be dismissed entirely. The Standardized Precipitation Index
(SPI) (McKee et al., 1993) calculates the probability of precipitation on many
different monthly time scales at different stations to provide a statistical
representation of precipitation deficits or surpluses. This can be used as a proxy for
soil moisture deficits or surpluses. In its calculation, the SPI is normalized which
allows for locations with differing climatological standards in precipitation to be
compared against each other. This provides a similar spatial- and temporal-relative
system akin to the SSC. The SPI has already been used to forecast drought and
heat waves with much success (Cancelliere et al., 2007; Mueller and Seneviratne,
2012) and may be used in the future as a way to predict modifications in SSC
weather type (Ford and Quiring, 2014b). The SPI was not originally used for this
study as it did not represent a proxy for the land-atmosphere interaction focus of
this project. Once the predictive variables are decided upon and confirmed through
studies, year-round prediction should be the main focus of future development.
The focus of this research consisted of warm-season studies as snowpacks
during the cold season, especially with trajectories reaching into Canada during the
winter, impacted ET analyses. Although outside of the project’s scope, this can be
resolved in a future project and is needed due to cold-related mortality. It is already
known that evapotranspiration is essentially negligible when there is snow on the
ground, but interactions between the ground and air become much more complex
once the snowpack begins to melt in spring. This is another reason against the use
53
Texas Tech University, Daniel J. Vecellio, May, 2015
of evapotranspiration as the predictand for air mass modification studies. Extension
of these methods to the cold season should not prove to be difficult once a method
for snow cover analysis and its interactions with the air above it is appropriately
handled.
In addition to the needs for future development of this research already stated,
progression in other facets of research in the field would be helpful to fully
implement the ideas put forth by this project.
1. Currently, the North America Soil Moisture Database (Quiring, 2014), housed
at Texas A&M University, provides historical soil moisture quantities for
stations across the United States, Canada and Mexico. Datasets of varying
periods of record at these sites are available for research based upon past
events. However, the system does not presently have real-time capabilities.
Improvements to the temporal acquisition of data as well as the addition of
new stations to enhance spatial coverage of the network would provide for a
more robust dataset to work with. This would help to provide a sufficient air
mass modification forecasting tool, not to mention how it may be used in
current forms of numerical weather prediction models in their surface
parameterizations.
2. In addition to the back trajectories output by the HYSPLIT model in this
research, the model also has capabilities to compute forward trajectories based
on a given input. Hence, the ever-sought-after challenge to create the best
forecasting model possible should continue to have an importance placed upon
it, especially for the search of better boundary layer turbulence
parameterizations to better investigate the movement of air masses with time,
so that five-day forecasts produced by the GFS, WRF or ECMWF models
54
Texas Tech University, Daniel J. Vecellio, May, 2015
may be included in trajectory forecasting with desired confidence. HYSPLIT
capabilities and techniques to resolve boundary layer motions, something
touched upon by Stohl et al. (2002), should also be continued to be improved
upon.
3. In the end, a traditional SSC numerical weather prediction model, one with
the ability to produce SSC forecasts multiple days out for the United
States/North America, may be the greatest advancement that can be made in
applied synoptic meteorology. However, it would certainly be deemed a large
undertaking due to the amount of data that would need to be assimilated into
said model because of the relative nature of the SSC. If such progress was
made, there would be two methods of predicting SSC type: a dynamical
method and a statistical method. This is much like what is seen in traditional
weather prediction today as described in the introduction of this paper. There
are many other improvements to our numerical weather prediction models
currently being undertaken to allow them to produce better results in the
context of its current output, but the science and the data needed for the
improvements presented here are certainly available for synopticians and
modelers.
4.3 Implications and Applications
Once extension of this work in the development of an air mass modification
prediction process is completed, there will be many opportunities to apply the
results to better applied research already completed with the SSC. The SSC is used
in many studies within the field of biometeorology when determining relationships
between weather with mortality and morbidity while also being used to understand
the weather’s effects on human health.
55
Texas Tech University, Daniel J. Vecellio, May, 2015
The prediction of these oppressive air masses will become of greater importance
as society surges into a future with an ever-increasingly warming climate. According
to the International Panel on Climate Change’s (IPCC) Fifth Assessment released
in 2013 (Stocker et al., 2013), relative to the period between 1986 and 2005,
temperatures could increase by almost five degree Celsius on average across the
globe by the end of the 21st century. Knight et al. (2008) confirmed this trend in
the historical data with respect to the SSC, finding that MT air masses had
generally increased over a majority of the United States with no preference to
season. Vanos and Cakmak (2014) confirmed this to a greater extent. As a result, it
would be expected that moist tropical weather type frequencies, and to a lesser
extent, dry tropical weather type frequencies, will continue be on the rise as time
moves to the future. Kalkstein and Greene (1997) explored mortality relationships
in forty-four large United States cities before narrowing their discussion down to the
cities of Chicago, Illinois, New York, New York and Los Angeles California. Using
three different general circulation models, the pair found large increases in MT
frequency in Chicago and New York as well as significant increases in DT frequency
in Los Angeles as parts of their future 2020 and 2050 climates. As a general
conclusion, they state that during summer, hot and dry DT and very warm and
humid MT consistently appear as “high-risk”. The spatial presence across the
country differs greatly, hence varying which regions will be affected more by each air
mass. In the three cities selected, as well when totaling across the studied cities, the
team found that excess mortality during the average summer season could triple as
a result of climate change.
The team of Sheridan et al. (2012a,b) performed similar research, but focused
on the state of California, a state with different climate zones, each that will feel the
effects of climate change in unique ways. Using future climate projections, they
56
Texas Tech University, Daniel J. Vecellio, May, 2015
found that inland locales such as Fresno and Sacramento will experience more
frequent DT weather types while cities along the coast such as Miramar and El Toro
will see increases in MT weather types. Both situations are associated with an
uptick in oppressive air mass types and, coinciding with them, an uptick in
projected mortality due to heat stress. Vanos and Cakmak (2014) looked at the past
climate in 30 different Canadian weather stations, finding a summertime increase of
moist tropical air masses in the majority of stations across the country with an
upward trend that looked to continue increasing into the future. In addition to
heat-related stress, they also noted, along with research completed by Health
Canada (Seguin and Berry, 2008), that air pollution episodes will become more
severe and longer-lasting in a projected warmer climate, negatively impacting those
living in those regions if adaptive measures are not taken.
Heat-health warning systems have become more and more prevalent in urban
areas over the past two decades (Sheridan and Kalkstein, 2004; Michelozzi et al.,
2010) as major heat waves, such as those in the northeastern United States in 1993,
in Chicago in 1995 and across Europe in 2003, have proven to be disastrous in terms
of loss of human life. Kalkstein et al. (1996a) developed one such system for the city
of Philadelphia, Pennsylvania in 1995 based on the SSC’s predecessor, the Temporal
Synoptic Index (Kalkstein et al., 1987). Using MOS forecasts, the system was able
to predict the arrival of an oppressive air mass, which was considered to be dry
tropical or maritime tropical for the city of Philadelphia, 48 hours before it arrived.
The system used an algorithm to determine when a health watch, health alert or
health warning should be issued based on the prediction of TSI category type (for
watch and alert) and estimated mortality (for warning). Later research found that
between the years of 1995 and 1998, the Philadelphia Hot Weather-Health
Watch/Warning System (PWWS) saved an estimated 2.6 lives on average in those
57
Texas Tech University, Daniel J. Vecellio, May, 2015
age 65 or older, resulting in a $468 million net benefit for the city during that time
(Ebi et al., 2004). A similar system was set up in Phoenix, Arizona in 2002 (NOTE:
This SSC-based system has since been replaced). Kalkstein and Sheridan (2007)
surveyed residents of the area and gauged how they perceived the warnings put out
by the National Weather Service (NWS) office. Over 86% of respondents said that
they were aware that warnings or advisories were issued, yet only 49.7% said that
they changed their daily routine on days of issuance. If, based on the results of this
research, better air mass type forecasts were able to be issued with more advanced
notice, mortality and morbidity figures would be expected to decrease even further
as more time and preparation would be available to the public and policy-makers.
Cold-related illness relationships can also be predicted by the SSC. Kalkstein
(2013) confirmed a heightened mortality in winter for the entire United States when
compared to summer, especially in the southwestern United States. Some of his
later research highlighted the presence of influenza outbreaks in the wake of dry
polar air mass types moving into regions of the southwestern United States due to
the cold, dry and particularly dusty conditions (Kalkstein and DeFelice, 2014). The
relationship is not constricted to the Southwest. Davis et al. (2012) found
relationships between influenza and pneumonia mortality and dry and cold weather
conditions in New York City, however, relationships with the actual DP weather
type were not statistically significant. Yet, a signal between atmospheric conditions
and human health was once again found in the data.
Being able to predict SSC type will help out in each of these previous studies as
well as many others that show dependence on weather type. In short-term projects
such as heat-health warning systems and influenza mortality prevention, simple
predictive probability will allow for numerous lives to be saved in events that take
place over the course of a few days. However, much larger research questions may
58
Texas Tech University, Daniel J. Vecellio, May, 2015
also be explored when breaking down the factors behind air mass modification.
Numerous papers have investigated how SSC frequency will change under future
climate scenarios, many of them cited in this work. But there remains a literature
gap in what factors will cause these frequency shifts. In studying the processes that
modify air masses, one or multiple atmospheric variables may arise as being the
drivers of this modification. Then, those factors may be probed in future climate
scenario analyses to confirm previous results based solely on SSC-type
characteristics as originally laid out by Sheridan (2002).
This may also have potential to be applied to other areas of future research and
new method application:
• Climate change and Earth’s warming are already apparent. However, research
has shown that while statistically significant risks of heat-related mortality
have remained, adaptation to higher temperatures have decreased heat-related
mortality and mortality risk in recent years (Bobb et al., 2014). Ebi et al.
(2004) also found that long-term adaptation-favorited processes, such as
improved healthcare (as hospitalizations during extreme heat events
increased), were at least partially responsible for declines in mortality.
Questions remain as to how projected future temperature increase will affect
the adaptation to heat that has been found in current studies. Voorhees et al.
(2011) used the IPCC A1B emissions scenario to model future temperature
change (2048-2052) and heat-related mortalities (3,700-3,800 for all-cause,
3,500 for cardiovascular disease and 21,000-27,000 for non-accidental) with no
adaptation or mitigation strategies accounted for. Stone et al. (2014),
however, incorporated vegetation and albedo enhancement mitigation
techniques into their analysis, revealing an offset of heat-related mortality by
40-99%. Discussed in Section 4.1, previous work has already been done using
59
Texas Tech University, Daniel J. Vecellio, May, 2015
the Standardized Precipitation Index (SPI) to predict extreme heat events
(Mueller and Seneviratne, 2012). This work has also been done using a simple
soil moisture method (Ford and Quiring, 2014a). Work combining the SPI and
SSC to predict future extreme heat events, whether by maximum temperature
of percent of hot days, using the characteristic factors of modification into
oppressive air masses highlighted by the framework of this and subsequent
research, can be examined in the future (Ford and Quiring, 2014b).
• In addition to heat, drought is becoming an increasingly prevalent problem
facing the United States, specifically in the western and midwestern portions
of the country (Peterson et al., 2013; Kam et al., 2014), with years such as
2011 standing out in recent memory. The coupling of heat extremes with the
severe lack of precipitation will have affects on health in the short- and
long-term. Sources of drinking water may begin to become scarce if these
conditions exacerbate in the future. Also, food shortages can result from
future water shortages as crops will not be able to be watered and livestock
will not receive the nutrients needed to provide acceptable meat for sale. An
investigation into large-scale SSC and modification factors in previous severe
drought conditions may help to provide a clue on atmospheric factors to look
for on preceding seasonal or yearly timescales.
• Plant phenology is emerging as a significant topic in the biometeorology field,
including how climate change is affecting the timing of the beginning of
growing seasons as well as early-season cold snaps which may affect a crop for
the rest of the year. A climatological study of SSC type specific to these
events and, once again, examining the characteristics of modification leading
up to the climatological mean may provide farmers a tool to protect their
60
Texas Tech University, Daniel J. Vecellio, May, 2015
harvest and livelihood.
• In the field of microbiology, some microbes have been found to thrive in
certain temperature and moisture conditions. As they are so small, they are
able to be picked up and transported within an air mass to a new location.
This applicative study would include both a HYSPLIT component to find
where these microbes are traveling from as well an SSC study to see which
weather types harbor populations of whatever organism is begin studied
(San Francisco, 2014).
There are hundreds-to-thousands of applications that a completed modification
framework can lead to. In the end, it’s a project that will increase the ability for
humans to adapt to their living conditions, both in the present and in the future.
The implications of this research and the subsequent follow-up studies are
significant, having a hand in human health, policy-making, agriculture, culture,
customs, society and human livelihood as a whole.
4.4 Final Conclusions
This project’s main findings can be summarized by three main points: the
inability of evapotranspiration to become the predictive variable in dealing with air
mass modification, the distinct disadvantage of using the SSC to describe the
characteristics of an air mass on an extended journey and the physics that the
HYSPLIT model masks or does not take into account. Overall, the goal of this
project was not achieved, but important takeaways from its failures were deduced
and discussed in Section 4.1.
With the realization that the relativity of the SSC may actually be hinderance
in modification studies, a greater importance should be placed on the realization of
modification in numerical weather prediction models to more accurately predict
61
Texas Tech University, Daniel J. Vecellio, May, 2015
meteorological variables and, in turn, SSC weather types more than a couple of days
in advance. If air mass modification must be disseminated with an absolute point of
view in order to calculate and predict SSC weather types days in advance, the
smaller-scale parameterizations within numerical weather prediction models must
continue to improve. This project examined air mass modification from a
synoptic-scale point of view while most of the fluxes, whether it be radiative and
moisture, work on the scale of the boundary layer or smaller.
To sum up, air mass modification occurs in our atmosphere, however,
attempting to quantify it is a complex problem. Additionally, use of the spatially-
and temporally-relative SSC to compare one air mass as it moves between two
different locations provides its own challenges, even though the classification system
is used in many meteorological contexts. The integration of the use of the
HYSPLIT model along with numerical weather prediction models may provide for
better modification or, at the very least, SSC weather type prediction, which has
been deemed important for knowledge in biometeorological applications. With this
in mind, broadening the spectrum of the weather scales at which researchers and
forecasters examine in the attempt to detect air mass modification is warranted.
62
Texas Tech University, Daniel J. Vecellio, May, 2015
BIBLIOGRAPHY
Arthur, D. and S. Vassilvitskii, 2007: K-means++: The Advantages of CarefulSeeding. Proceedings of the Eighteenth Annual ACM-SIAM Symposium onDiscrete Algorithms , Society for Industrial and Applied Mathematics, 1027–1035.
Artuso, F., P. Chamard, S. Piacentino, A. Di Sarra, D. Meloni, F. Monteleone,D. Sferlazzo, and F. Thiery, 2007: Atmospheric Methane in the Mediterranean:Analysis of Measurements at the Island of Lampedusa During 1995–2005.Atmospheric Environment , 41, 3877–3888.
Bergeron, T., 1928: Uber die dreidimensional verknupfende Wetteranalyse.Cammermeyer in Komm.
— 1930: Richtlinien einer dynamischen klimatologie. Meteorologische Zeitschrift ,47, 246–262.
Bobb, J., R. Peng, M. Bell, and F. Dominici, 2014: Heat-Related Mortality andAdaptation in the United States. Environmental Health Perspectives , 130739.
Bonan, G. B., 2002: Ecological Climatology: Concepts and Applications . CambridgeUniversity Press.
Borma, L., H. Da Rocha, O. Cabral, C. Von Randow, E. Collicchio,D. Kurzatkowski, P. Brugger, H. Freitas, R. Tannus, L. Oliveira, et al., 2009:Atmosphere and Hydrological Controls of the Evapotranspiration Over aFloodplain Forest in the Bananal Island Region, Amazonia. Journal ofGeophysical Research: Biogeosciences (2005–2012), 114.
Bower, D., G. R. McGregor, D. M. Hannah, and S. C. Sheridan, 2007: Developmentof a spatial synoptic classification scheme for western europe. InternationalJournal of Climatology , 27, 2017–2040.
Brazel, A., P. Gober, S.-J. Lee, S. Grossman-Clarke, J. Zehnder, B. Hedquist, andE. Comparri, 2007: Determinants of changes in the regional urban heat island inmetropolitan Phoenix (Arizona, USA) between 1990 and 2004. Climate Research,33, 171.
Brown, D. P., J. L. Beven, J. L. Franklin, and E. S. Blake, 2010: Atlantic HurricaneSeason of 2008. Monthly Weather Review , 138, 1975–2001.
Burke, C. J., 1945: Transformation of polar continental air to polar maritime air.Journal of Meteorology , 2, 94–112.
Cancelliere, A., G. Di Mauro, B. Bonaccorso, and G. Rossi, 2007: Droughtforecasting using the Standardized Precipitation Index. Water ResourcesManagement , 21, 801–819.
63
Texas Tech University, Daniel J. Vecellio, May, 2015
Covert, D., V. Kapustin, T. Bates, and P. Quinn, 1996: Physical Properties ofMarine Boundary Layer Aerosol Particles of the Mid-Pacific in Relation toSources and Meteorological Transport. Journal of Geophysical Research:Atmospheres (1984–2012), 101, 6919–6930.
Cunningham, K. M. and J. C. Ogilvie, 1972: Evaluation of hierarchical groupingtechniques: A preliminary study. The Computer Journal , 15, 209–213.
Davis, R. E. and L. S. Kalkstein, 1990a: Development of an automated spatialsynoptic climatological classification. International Journal of Climatology , 10,769–794.
— 1990b: Using a spatial synoptic climatological classification to assess changes inatmospheric pollution concentrations. Physical Geography , 11, 320–342.
Davis, R. E., C. P. Normile, L. Sitka, D. M. Hondula, D. B. Knight, S. P. Gawtry,and P. J. Stenger, 2010: A comparison of trajectory and air mass approaches toexamine ozone variability. Atmospheric Environment , 44, 64–74.
Davis, R. E., C. E. Rossier, and K. B. Enfield, 2012: The impact of weather oninfluenza and pneumonia mortality in New York City, 1975–2002: a retrospectivestudy. PloS One, 7, e34091.
Dayan, U., 1986: Climatology of back trajectories from Israel based on synopticanalysis. Journal of Climate and Applied Meteorology , 25, 591–595.
Draxler, R. and G. Hess, 2010: Description of the HYSPLIT 4 Modeling System.URL http://www.arl.noaa.gov/documents/reports/arl-224.pdf.
Draxler, R. and G. Rolph, 2006: Hybrid Single-particle Lagrangian IntegratedTrajectory (HYSPLIT) Model, Version 4.8, via NOAA ARL Website(http://www.arl.noaa.gov/ready/hysplit4.html). NOAA Air ResourcesLaboratory, Silver Spring, Maryland. ESRI. 2006. ArcMap GeographicInformation System (GIS).
Ebi, K. L., T. J. Teisberg, L. S. Kalkstein, L. Robinson, and R. F. Weiher, 2004:Heat watch/warning systems save lives: estimated costs and benefits forPhiladelphia 1995-98. Bulletin of the American Meteorological Society , 85,1067–1073.
Falkovich, A. H., E. Ganor, Z. Levin, P. Formenti, and Y. Rudich, 2001: Chemicaland Mineralogical Analysis of Individual Mineral Dust Particles. Journal ofGeophysical Research: Atmospheres (1984–2012), 106, 18029–18036.
Fleming, Z. L., P. S. Monks, and A. J. Manning, 2012: Review: Untangling theinfluence of air-mass history in interpreting observed atmospheric composition.Atmospheric Research, 104, 1–39.
64
Texas Tech University, Daniel J. Vecellio, May, 2015
Ford, T. W. and S. M. Quiring, 2014a: In situ soil moisture coupled with extremetemperatures: A study based on the Oklahoma Mesonet. Geophysical ResearchLetters , 41, 4727–4734.
— 2014b: Personal communication.
Freedman, J. M. and D. R. Fitzjarrald, 2001: Postfrontal airmass modification.Journal of Hydrometeorology , 2, 419–437.
Freitag, S., A. Clarke, S. Howell, V. Kapustin, T. Campos, V. Brekhovskikh, andJ. Zhou, 2014: Combining airborne gas and aerosol measurements withHYSPLIT: a visualization tool for simultaneous evaluation of air mass history andback trajectory consistency. Atmospheric Measurement Techniques , 7, 107–128.
Glahn, H. R. and D. A. Lowry, 1972: The use of model output statistics (MOS) inobjective weather forecasting. Journal of Applied Meteorology , 11, 1203–1211.
Harman, J. R., 1987: Mean monthly North American anticyclone frequencies,1950-79. Monthly Weather Review , 115, 2840–2848.
Hawkins, D., M. Muller, and J. Krooden, 1982: Topics in Applied MultivariateAnalysis . Cambridge University Press, Cambridge, 303–356 pp.
Hayhoe, K., S. Sheridan, L. Kalkstein, and S. Greene, 2010: Climate change, heatwaves, and mortality projections for Chicago. Journal of Great Lakes Research,36, 65–73.
Hogan, T. F., T. E. Rosmond, and R. Gelaro, 1991: The NOGAPS forecast model:A technical description. Technical report, DTIC Document.
Hondula, D., J. Vanos, and S. Gosling, 2014: The SSC: a decade of climate–healthresearch and future directions. International Journal of Biometeorology , 58,109–120.
Hondula, D. M., R. E. Davis, D. B. Knight, L. J. Sitka, K. Enfield, S. B. Gawtry,P. J. Stenger, M. L. Deaton, C. P. Normile, and T. R. Lee, 2013: A respiratoryalert model for the Shenandoah Valley, Virginia, USA. International Journal ofBiometeorology , 57, 91–105.
Hondula, D. M., L. Sitka, R. E. Davis, D. B. Knight, S. D. Gawtry, M. L. Deaton,T. R. Lee, C. P. Normile, and P. J. Stenger, 2010: A back-trajectory and air massclimatology for the Northern Shenandoah Valley, USA. International Journal ofClimatology , 30, 569–581.
Kalkstein, A. J., 2013: Regional Similarities in Seasonal Mortality across the UnitedStates: An Examination of 28 Metropolitan Statistical Areas. PloS One, 8,e63971.
65
Texas Tech University, Daniel J. Vecellio, May, 2015
Kalkstein, A. J. and T. P. DeFelice, 2014: Predicting the Flu: The Impact of CoolDry Air on Hospital Admissions in the Southwest United States. AMHCRConference Proceedings .
Kalkstein, A. J. and S. C. Sheridan, 2007: The social impacts of the heat–healthwatch/warning system in Phoenix, Arizona: assessing the perceived risk andresponse of the public. International Journal of Biometeorology , 52, 43–55.
Kalkstein, L. S., 1979: A syntopic climatological approach for environmentalanalysis. Middle States Division: Association of American Geographers , 68–75.
Kalkstein, L. S., P. C. Dunne, and R. S. Vose, 1990: Detection of climatic change inthe western North American Arctic using a synoptic climatological approach.Journal of Climate, 3, 1153–1167.
Kalkstein, L. S. and J. S. Greene, 1997: An evaluation of climate/mortalityrelationships in large us cities and the possible impacts of a climate change.Environmental Health Perspectives , 105, 84.
Kalkstein, L. S., J. S. Greene, D. M. Mills, A. D. Perrin, J. P. Samenow, and J.-C.Cohen, 2008: Analog European heat waves for US cities to analyze impacts onheat-related mortality. Bulletin of the American Meteorological Society , 89, 75–85.
Kalkstein, L. S., P. F. Jamason, J. S. Greene, J. Libby, and L. Robinson, 1996a:The Philadelphia hot weather-health watch/warning system: development andapplication, Summer 1995. Bulletin of the American Meteorological Society , 77,1519–1528.
Kalkstein, L. S., M. C. Nichols, C. D. Barthel, and J. S. Greene, 1996b: A newspatial synoptic classification: application to air mass analysis. InternationalJournal of Climatology , 16, 983–1004.
Kalkstein, L. S., S. C. Sheridan, and D. Y. Graybeal, 1998: A determination ofcharacter and frequency changes in air masses using a spatial synopticclassification. International Journal of Climatology , 18, 1223–1236.
Kalkstein, L. S., G. Tan, and J. A. Skindlov, 1987: An evaluation of three clusteringprocedures for use in synoptic climatological classification. Journal of Climate andApplied Meteorology , 26, 717–730.
Kalnay, E. and M. Cai, 2003: Impact of Urbanization and Land-Use Change onClimate. Nature, 423, 528–531.
Kalnay, E., M. Kanamitsu, and W. Baker, 1990: Global numerical weatherprediction at the National Meteorological Center. Bulletin of the AmericanMeteorological Society , 71, 1410–1428.
66
Texas Tech University, Daniel J. Vecellio, May, 2015
Kam, J., J. Sheffield, and E. F. Wood, 2014: Changes in Drought Risk Over theContiguous United States (1901–2012): The Influence of the Pacific and AtlanticOceans. Geophysical Research Letters , 41, 5897–5903.
Kanamitsu, i., J. Alpert, K. Campana, P. Caplan, D. Deaven, M. Iredell, B. Katz,H.-L. Pan, J. Sela, and G. White, 1991: Recent changes implemented into theglobal forecast system at NMC. Weather and Forecasting , 6, 425–435.
Kanamitsu, M., 1989: Description of the NMC global data assimilation and forecastsystem. Weather and Forecasting , 4, 335–342.
Klein, W. H., B. M. Lewis, and I. Enger, 1959: Objective prediction of five-daymean temperatures during winter. Journal of Meteorology , 16, 672–682.
Knight, D. B., R. E. Davis, S. C. Sheridan, D. M. Hondula, L. J. Sitka, M. Deaton,T. R. Lee, S. D. Gawtry, P. J. Stenger, F. Mazzei, et al., 2008: Increasingfrequencies of warm and humid air masses over the conterminous United Statesfrom 1948 to 2005. Geophysical Research Letters , 35.
Lawrence, D. M., P. E. Thornton, K. W. Oleson, and G. B. Bonan, 2007: ThePartitioning of Evapotranspiration Into Transpiration, Soil Evaporation, andCanopy Evaporation in a GCM: Impacts on Land-Atmosphere Interaction.Journal of Hydrometeorology , 8, 862–880.
Lee, C. C., 2014a: The Development of a Gridded Weather Typing ClassificationScheme. Ph.D. thesis, Kent State University.
— 2014b: The development of a gridded weather typing classification thedevelopment of a gridded weather typing classification scheme. InternationalJournal of Climatology .
Lee, C. C. and S. C. Sheridan, 2012: A six-step approach to developing futuresynoptic classifications based on GCM output. International Journal ofClimatology , 32, 1792–1802.
Lund, I. A., 1963: Map-pattern classification by statistical methods. Journal ofApplied Meteorology , 2, 56–65.
Lynch, P., 2008: The origins of computer weather prediction and climate modeling.Journal of Computational Physics , 227, 3431–3444.
McKee, T. B., N. J. Doesken, and J. Kleist, 1993: The relationship of droughtfrequency and duration to time scales. Proceedings of the 8th Conference onApplied Climatology , American Meteorological Society Boston, MA, volume 17,179–183.
67
Texas Tech University, Daniel J. Vecellio, May, 2015
Metzger, K. B., K. Ito, and T. D. Matte, 2010: Summer heat and mortality in NewYork City: how hot is too hot? Environmental Health Perspectives , 80–86.
Michelozzi, P., F. K. De’Donato, A. M. Bargagli, D. D’Ippoliti, M. De Sario,C. Marino, P. Schifano, G. Cappai, M. Leone, U. Kirchmayer, et al., 2010:Surveillance of Summer Mortality and Preparedness to Reduce the Health Impactof Heat Waves in Italy. International Journal of Environmental Research andPublic Health, 7, 2256–2273.
Molinari, R. L., 1987: Air mass modification over the eastern Gulf of Mexico as afunction of surface wind fields and Loop Current position. Monthly WeatherReview , 115, 646–652.
Mueller, B. and S. I. Seneviratne, 2012: Hot days induced by precipitation deficitsat the global scale. Proceedings of the National Academy of Sciences , 109,12398–12403.
Penman, H. L., 1948: Natural evaporation from open water, bare soil and grass.Proceedings of the Royal Society of London. Series A. Mathematical and PhysicalSciences , 193, 120–145.
Peterson, T. C., R. R. Heim Jr, R. Hirsch, D. P. Kaiser, H. Brooks, N. S.Diffenbaugh, R. M. Dole, J. P. Giovannettone, K. Guirguis, T. R. Karl, et al.,2013: Monitoring and Understanding Changes in Heat Waves, Cold Waves,Floods, and Droughts in the United States: State of Knowledge. Bulletin of theAmerican Meteorological Society , 94, 821–834.
Priestley, C. and R. Taylor, 1972: On the assessment of surface heat flux andevaporation using large-scale parameters. Monthly Weather Review , 100, 81–92.
Quiring, S. M., 2014: North American Soil Moisture Database. URLhttp://soilmoisture.tamu.edu/.
Rainham, D. G., K. E. Smoyer-Tomic, S. C. Sheridan, and R. T. Burnett, 2005:Synoptic weather patterns and modification of the association between airpollution and human mortality. International Journal of Environmental HealthResearch, 15, 347–360.
Reynolds, R. W., N. A. Rayner, T. M. Smith, D. C. Stokes, and W. Wang, 2002:An improved in situ and satellite SST analysis for climate. Journal of Climate,15, 1609–1625.
Reynolds, R. W. and T. M. Smith, 1994: Improved global sea surface temperatureanalyses using optimum interpolation. Journal of Climate, 7, 929–948.
68
Texas Tech University, Daniel J. Vecellio, May, 2015
Rodell, M., P. Houser, U. e. a. Jambor, J. Gottschalck, K. Mitchell, C. Meng,K. Arsenault, B. Cosgrove, J. Radakovich, M. Bosilovich, et al., 2004: The globalland data assimilation system. Bulletin of the American Meteorological Society ,85, 381–394.
San Francisco, M., 2014: Personal communication.
Seguin, J. and P. Berry, 2008: Human Health in a Changing Climate: A CanadianAssessment of Vulnerabilities and Adaptive Capacity: Synthesis Report . HealthCanada.
SethuRaman, S., 1976: Air mass modification due to change in surfacecharacteristics. Monthly Weather Review , 104, 1040–1043.
Sheridan, S. C., 2002: The redevelopment of a weather-type classification scheme forNorth America. International Journal of Climatology , 22, 51–68.
— 2003: North American weather-type frequency and teleconnection indices.International Journal of Climatology , 23, 27–45.
Sheridan, S. C., M. J. Allen, C. C. Lee, and L. S. Kalkstein, 2012a: Future heatvulnerability in California, part II: projecting future heat-related mortality.Climatic Change, 115, 311–326.
Sheridan, S. C. and A. J. Kalkstein, 2010: Seasonal variability in heat-relatedmortality across the United States. Natural Hazards , 55, 291–305.
Sheridan, S. C. and L. S. Kalkstein, 2004: Progress in heat watch-warning systemtechnology. Bulletin of the American Meteorological Society , 85, 1931–1941.
Sheridan, S. C., L. S. Kalkstein, and J. M. Scott, 2000: An evaluation of thevariability of air mass character between urban and rural areas. Biometeorologyand Urban Climatology at the Turn of the Millennium, 487–490.
Sheridan, S. C., C. C. Lee, M. J. Allen, and L. S. Kalkstein, 2012b: Future heatvulnerability in California, part I: projecting future weather types and heatevents. Climatic Change, 115, 291–309.
Skamarock, W., J. Klemp, and J. Dudhia, 2001: Prototypes for the WRF (WeatherResearch and Forecasting) model. Preprints, Ninth Conf. Mesoscale Processes,J11–J15, American Meteorological Society, Fort Lauderdale, FL.
69
Texas Tech University, Daniel J. Vecellio, May, 2015
Stocker, T. F., D. Qin, G.-K. Plattner, M. Tignor, S. K. Allen, J. Boschung,A. Nauels, Y. Xia, V. Bex, P. M. Midgley, et al., 2013: Climate Change 2013. ThePhysical Science Basis. Working Group I Contribution to the Fifth AssessmentReport of the Intergovernmental Panel on Climate Change-Abstract fordecision-makers. Technical report, Groupe d’experts intergouvernemental surl’evolution du climat/Intergovernmental Panel on Climate Change-IPCC, C/OWorld Meteorological Organization, 7bis Avenue de la Paix, CP 2300 CH-1211Geneva 2 (Switzerland).
Stohl, A., S. Eckhardt, C. Forster, P. James, N. Spichtinger, and P. Seibert, 2002: Areplacement for simple back trajectory calculations in the interpretation ofatmospheric trace substance measurements. Atmospheric Environment , 36,4635–4648.
Stohl, A., M. Hittenberger, and G. Wotawa, 1998: Validation of the lagrangianparticle dispersion model flexpart against large-scale tracer experiment data.Atmospheric Environment , 32, 4245–4264.
Stone, B., J. Vargo, P. Liu, D. Habeeb, A. DeLucia, M. Trail, Y. Hu, andA. Russell, 2014: Avoided Heat-Related Mortality Through Climate AdaptationStrategies in Three US Cities. PloS one, 9, e100852.
Tan, J., L. Kalkstein, J. Huang, S. Lin, H. Yin, and D. Shao, 2004: An operationalheat/health warning system in Shanghai. International Journal ofBiometeorology , 48, 157–162.
Vanos, J. and S. Cakmak, 2014: Changing air mass frequencies in Canada: potentiallinks and implications for human health. International Journal of Biometeorology ,58, 121–135.
Vanos, J. K., C. Hebbern, and S. Cakmak, 2014: Risk assessment for cardiovascularand respiratory mortality due to air pollution and synoptic meteorology in 10Canadian cities. Environmental Pollution, 185, 322–332.
Voorhees, A. S., N. Fann, C. Fulcher, P. Dolwick, B. Hubbell, B. Bierwagen, andP. Morefield, 2011: Climate Change-Related Temperature Impacts on WarmSeason Heat Mortality: A Proof-of-Concept Methodology Using BenMAP.Environmental Science & Technology , 45, 1450–1457.
Wielicki, B. A., B. R. Barkstrom, E. F. Harrison, R. B. Lee III, G. Louis Smith, andJ. E. Cooper, 1996: Clouds and the Earth’s Radiant Energy System (CERES):An Earth Observing System Experiment. Bulletin of the American MeteorologicalSociety , 77, 853–868.
Yarnal, B., 1993: Synoptic Climatology in Environmental Analysis: A Primer .Belhaven Press London.
70
Texas Tech University, Daniel J. Vecellio, May, 2015
APPENDIX A
A.1 Finding closest station to a given latitude/longitude
Texas Tech University, Daniel J. Vecellio, May, 2015
m.plot(X[d,e:e+11],Y[d,e:e+11],’k’)
m.plot(X[d,e],Y[d,e],’k’,linewidth=0.01)
else:
m.plot(X[d,e:e+11],Y[d,e:e+11],’k:’)
m.plot(X[d,e],Y[d,e],’k’,linewidth=0.01)
elif e == 96:
if SSCtype[c] == 1:
m.plot(X[d,e],Y[d,e],color=’#FFA500’)
elif SSCtype[c] == 2:
m.plot(X[d,e],Y[d,e],’y’)
elif SSCtype[c] == 3:
m.plot(X[d,e],Y[d,e],’r’)
elif SSCtype[c] == 4:
m.plot(X[d,e],Y[d,e],’c’)
elif SSCtype[c] == 5:
m.plot(X[d,e],Y[d,e],’b’)
elif SSCtype[c] == 6:
m.plot(X[d,e],Y[d,e],’g’)
elif SSCtype[c] == 66:
m.plot(X[d,e],Y[d,e],color=’#006400’)
elif SSCtype[c] == 67:
m.plot(X[d,e],Y[d,e],color=’#006400’)
elif SSCtype[c] == 7:
m.plot(X[d,e],Y[d,e],’k’)
else:
continue
plt.title(’’+a+’ ’+b+’ Trajectories’)
plt.show()
Additional code, including a script to run the HYSPLIT model as well as otherdata manipulation framework, may be made available upon request.
75
Texas Tech University, Daniel J. Vecellio, May, 2015
APPENDIX B
Figure B.1. Average evapotranspiration (kg/m2) values for each modification scenarioof the Huntsville, AL DT dataset. Full and partial designations are described inSection 3.4
Figure B.2: Same as Figure 4.1 but for Huntsville, AL MT dataset
76
MISCELLANEOUS FIGURES
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure B.3: Same as Figure 4.1 but for Wilmington, DE DT dataset
Figure B.4: Same as Figure 4.1 but for Wilmington, DE MT dataset
77
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure B.5: Same as Figure 4.1 but for Lexington, KY DT dataset
Figure B.6: Same as Figure 4.1 but for Lexington, KY MT dataset
78
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure B.7: Same as Figure 4.1 but for Raleigh-Durham, NC DT dataset
Figure B.8: Same as Figure 4.1 but for Raleigh-Durham, NC MT dataset
79
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure B.9: Same as Figure 4.1 but for Oklahoma City, OK DT dataset
Figure B.10: Same as Figure 4.1 but for Oklahoma City, OK MT dataset
80
Texas Tech University, Daniel J. Vecellio, May, 2015
Figure B.11. Same as Figure 4.1 but for a five-city average of DT-resultant modifiedweather types
Figure B.12. Same as Figure 4.1 but for a five-city average of MT-resultant modifiedweather type