The
Anthony R. Lupo1 Patrick S. Market1 F. Adnan Akyüz1,2 Patrick E. Guinan1,2
Janelle
1Department of Atmospheric Sciences 2
389 Mc ReynoldsHall 365 Mc Reynolds Hall
Submitted to: Electronic Journal of Operational
Meteorology June 2003
*Corresponding author address:
Anthony R. Lupo, LupoA@missouri.edu.
The heat island effect is a well known feature in the microclimate of large urban areas,
but only a few studies have addressed the heat island effect for smaller
cities. Here we examined the combined impact of
1. Introduction
The effect of urban environments on
local temperature and precipitation distributions have been examined in the
past (e.g., Changnon, 1981; Segal and Arritt, 1992; Karl and Knight, 1997; Melhuish
and Pedder, 1998; Pinho and
Manso-Orgaz, 2000; Rozoff
and Cotton, 2001; Changnon, 2003). These studies have
usually focused on cities that have large populations, however, Melhuish and Pedder (1998) and Pinho and Manso-Orgaz (2000)
examine the heat island effect in smaller urban areas. The "heat-island
effect" produced by cities can have a profound impact, sometimes adverse,
on the well-being of its residents (e.g., Karl and Knight, 1997). Weather
elements that impact on human health and comfort are
routinely included in local forecasts and/or forecast discussions or are
accounted for by forecasters in their county warning areas.
The heat island effect is produced
by many factors which result in a change in the
underlying energy budgets in the boundary layer due to urbanization. These
include effects such as (e.g., Oke, 1982); an increase in sensible heating (e.g., due to changes in
surface albedoes), an increase in thermal storage
capacity of the underlying surface, decreased evapo-transpiration,
and heat given off (generated) by urban structures. These processes then can
have an impact on the temperature field (see references above) and the
precipitation field (e.g., Shephard et al., 2002; Rozoff et al. 2003; Changnon,
2003). A few studies examined also the climatological
(long-term) impact of heat islands including their variance by season (e.g., DeMarrais, 1975; Ackerman, 1985).
There is published work (e.g., Melhuish and Pedder, 1998; Pinho and Manso-Orgaz, 2000)
demonstrating that medium-sized and small urban areas may also be responsible
for heat-island effects. While the heat islands associated with these smaller
urban areas would not be expected to be as pronounced as those of larger
cities, the heat island effect in the latter study was shown to be quite
substantial (up to 7.5o C for individual days).
The main objective for
this work was to demonstrate the temporal and spatial extent to which Columbia,
Missouri (a smaller urban area), and the University of Missouri campus produce
a heat-island effect that could be incorporated into forecasts using the latest
technology and made available to the public (e.g., the National Digital
Forecast Database or NDFD, see Glahn and Ruth (2003)
and references therein).
2.
Data and Methodology
a. Data
Volunteer participants provided the temperature data, which were measured using a Radio Shack® Indoor/Outdoor
Maximum-Minimum thermometer (Item #63 - 1014). These instruments resided
indoors and included a 10-ft probe, which was deployed
outdoors. The
b. Methodology
For our purposes,
The terrain surrounding the
Faculty, staff, and students (22
volunteer participants in all - 17 were observers and 7
analyzed or archived the data) in the Atmospheric Sciences Program (and some
outside the program) were invited to participate in this project. Enlisting
volunteer participants to measure local variations in climatic parameters has
produced successful results in other locations (e.g., Doeskin and Weaver,
2000). Those who deployed instruments were ultimately selected on the basis of their location in the Columbia region, and
their ability to accommodate the proper deployment techniques of the
instrument(s). Students were given explicit
instructions on how to deploy the instrument. Also included in the site
selection was an attempt to concentrate some instruments in the south-central
part of Columbia, which has less green-space in comparison to other regions of
the city due to recent development.
In order to determine if the heat island
effect was detectable given the fact that each Radio Shack®
instrument did not read the same values despite being subject to the same
conditions, the instruments were compared to a
standard instrument. The standard deviation among the set thermometers was
calculated. The range in the set was 1.0 oF
(1.3 oF) at room temperature (in an ice
bath), and the standard deviation was 0.35 oF
in the set for both trials. Thus, any heat-island effect would have to be
significantly larger than the standard deviation after correcting the data to
the standard. Also, a Radio-Shack® instrument was tested in real
time against an electronic thermometer, HMP35C, used by the automated weather
stations, and there was remarkable agreement between the two instruments (this
automated instrument would fall somewhere in the middle of our max/min.
instrument sample). Rigorous statistical testing other than the informal test
described above was not performed since the small
sample precludes producing statistically robust results. In spite of this
problem, meaningful results can be obtained (e.g., Nicholls, 2001) and compared
to studies which found similar results.
The participants collected the maximum
and minimum temperature once daily at 0400 UTC (10:00 pm LST). These data were recorded and then averaged, with the goal of determining if
the heat island existed in the mean data field. The strength of the heat
island effect is defined as:
HI = Tic - Tos (1)
where Tic is the mean temperature recorded by
the "inner city" units (defined as the square area on Fig. 1) and Tos is
the mean temperature recorded by the instruments more than 1 mile outside the
city limits. The mean temperatures produced by this instrumentation network in
these regions are compared in order to examine the
distribution of the heat island effect.
3.
Season-by-season results
using monthly means
The analysis of the COHIX project data
started with July 2000. Table 1 and Fig. 3 show the results after examining the data from
a. July and August 2000 results
The monthly mean temperature for
July (August) was below (above) normal when comparing the mean at the COU
airport with the 30-year normals (Table 2). As shown
in Table 1, there was a difference of 2.7 and 2.8 oF between the mean of the inner city and
outside city stations (HI) for the maximum and minimum temperatures for July,
respectively. All the inner city stations, in general, recorded monthly mean
temperatures that were higher than the highest means recorded outside the city
for maximum or minimum temperatures. The largest difference between the warmest
individual inner city station and the coolest outer city station was 3.3 oF and 4.7 oF
for the maximum and minimum temperatures, respectively (Table
1, HImax). During August (Fig. 4), the heat island effect was stronger for the
maximum temperatures than that found for July (3.4 oF),
while the minimum temperatures produced a weaker signal (1.9 oF). The largest differences between individual
stations were 4.8 oF for maximum
temperatures and 3.3 oF for the minimum
temperatures. The warmest individual stations were inner city stations, while
the coolest stations were outside the city.
b.
September - November 2000 results
The
HI values for the each of the fall months were smaller for the maximum
temperatures than for the minimum temperatures (Table
1). For September and October, the maximum temperatures were slightly less
than 1 oF in the
city of Columbia as compared to the outside, while the minimum temperatures
were nearly 2.5 oF warmer in the city.
These values are smaller than the comparable values for the July and August
period. During November, however, the heat island effect was comparable to that
of August despite cloudier conditions, with the minimum temperatures showing
the stronger signal. An examination of the differences
between the warmest and coldest individual stations (Table
1) reveal that these values are comparable to those of the warmer
months. This suggests that the coverage of the heat island effect may have
shrunk in area coverage and weakened during the cooler months, and examining
contour plots of August (Fig. 4) versus October (Fig. 5) supports this hypothesis.
c.
December
2000 - February 2001 results
The heat island effect for December
was as strong as that for the summer months (Table 1),
but like the fall season, the region of
d. March - June 2001 results
The strength of the heat island for
the spring months was similar to that of the other months when examining HI or
taking the difference between the warmest inner city station and the coldest
station outside the city (Table 1). However, there was a difference in the area coverage of the heat
island as the effect expanded during these months and by May and June the area
coverage was similar to that of July and August of 2000 (not shown). Also, the strength of the heat island effect was quite large
during June, and the effect was larger for the maximum temperatures than for
the minimum temperatures. Table 1 supports the assertion
of an expanding heat island during the spring season when comparing the values
of Tb (temperatures at stations inside the city limits but not in the inner
domain) to those of the inner (Tic) and outer (Tos)
city stations. During the latter part of the fall and throughout the winter
months, the values of Tb were closer to those of Tos.
Then during the spring season, these two values were closer to Tic as they were
during July and August of 2000.
e. Discussion
An examination of the data reveals
that when the monthly average of inner city stations are
compared to those outside the city (Fig. 3), there
is a discernable urban influence in the local temperature fields on the order
of 2 - 3 oF. This difference grows to 3 -
6 oF when comparing the monthly means of
the warmest inner city station versus the coldest station outside the city.
These values are consistent with those found by Pinho
and Manso - Orgaz (2000)
for a smaller city, and are a little less than those which
might be expected for a city of Columbia's size (see Aguado
and Burt, 2001, ch. 14). Thus, the investigators are
confident that their result is robust even though no rigorous statistical
testing was performed due to the small sample size. It
should also be noted that the heat island effect found here is larger than the
spread in the instrument sample, the standard deviation of the sample, and even
the precision of the instruments used (+/- 1o C or 1.8 oF for the Radio-Shack® instrument).
That the heat island effect is not
of the magnitude expected for a city of Columbia's size may be partially due to the fact that Columbia has made an effort to increase
the amount of green-space within city limits over the last 15 years. The
assertion that green-space can reduce the heat island effect is supported by Table 1 when comparing the values of Ts (stations in
the southern part of the city where there has been more intensive development
and decreasing green-space) to those of Tic, Tos, and
Tb. The values of Ts are generally more similar to Tic than
those of Tb or Tos. However, another possible
reason for the results found here may be that no
instruments were deployed in the center of town where there are more buildings
and more concrete and asphalt covered surfaces. No instruments were deployed in this area since proper instrument
deployment, data collection, and instrument integrity could not be guaranteed.
The heat island itself does vary
with the seasons as is shown by Table 1, Figs. 4,5
and the discussion above. The heat island effect does expand in area extent
during the warmer months and contracts during the colder months. This
contradicts the commonly held belief that heat islands expand during the cold
season. The contraction of the heat island here may be due to several factors
including, increased cloudiness during the cold season, or the low sun angle. Also, the Columbia region does not have the construction
density of larger cities, thus it is likely that the regional surface may be of
more uniform character in terms of surface albedoes
after vegetation dies off in the fall and grows again in the spring.
The HI values
are similar for all months whether the means of all the inner city and stations
outside the city are used, or the warmest (coldest) stations from the former
(latter) group are compared. It also appears that the
heat island effect is stronger in the maximum (minimum) temperatures during the
summer (winter) months. Finally, December 2000 stands out as a month in which
the heat island effect was strongest. This may be due, at least partially, to
the fact that this month was the second coldest December in the history of
Columbia, and was associated with an unusually persistent snow cover during
that month. The persistent snow cover would fundamentally alter the regional
surface radiation balance as snow cover is well known
to be a strong reflector (emitter) of shortwave (longwave)
radiation. Also, snow cover in the regions outside the
city would be expected to stay fresher for a longer period of time, while snow
is removed from large portions of Columbia's surface area. What snow remains
becomes dirtier more quickly in Columbia since the city maintenance department
liberally spreads black cinders on the roads to improve vehicle traction on snow covered roads and absorb more sunlight. However, we
acknowledge that the December heat island may also be partially due to the
increased need for heating in the city as suggested by Oke
(1982) and others. Nonetheless, since the areal
coverage of the strong heat island was similar to that of the other fall
months, and did not expand as other studies have shown, the former explanation
regarding the change in albedo due to snow cover is
plausible.
An examination of individual days shows that for 54
(31) of the daily maxima (mininma), the temperatures
were 5 – 10 oF greater inside the city
than outside the city. Most of these were associated with mostly clear skies
and winds of less than 10 kts. Most of these daily
maxima (minima) occurred during the warm (cold) season, which reflects the
seasonal changes in monthly means described above. While these represent a
small percentage of the days during the year, this kind of information, for
example, could make a difference in the urban area forecasts on a digital map
for extremely warm days, or during, for example, freezing rain events (Changnon, 2003).
4. Summary and Conclusions
Many publications have shown the impact
on small-scale regional surface temperatures as caused by urbanization or
agricultural activities. The heat island effect has been
studied extensively for larger cities, but there are fewer studies
examining this effect for smaller urban areas. For this study, 17 thermometers were distributed throughout the Columbia, MO, region to
examine the impact of the city and the University of Missouri campus on the
surface temperature fields. Daily data was gathered from
1 July 2000 to 30 June 2001.
We examined mean monthly data in order
to determine if the heat island effect is detectable in the region's
microclimate, and all 12 months exhibited a clear "heat island
effect" as the mean temperature of the inner city sites exceeded those of
the sites outside the city. The heat island effect was much larger than both
the standard deviation of the 20 individually purchased (and deployed)
instruments or their range when they were tested under
"uniform" conditions. This suggests that the Columbia, MO, heat
island effect is a significant feature in the local microclimate. The heat
island effect was larger in area during the warm season with a stronger effect
shown in the maximum temperatures during the summer months and in minimum
temperatures during the winter months. Also,
fundamentally altering the surface type such as adding green-space or a
persistent snow cover is shown to influence the strength of the heat island.
When examining the strength of the heat island as calculated by the difference
between the monthly means of the warmest individual station inside the city and
the coolest station outside the city revealed temperature differences of 3 - 6 oF.
With this type of knowledge about urban areas in a
forecaster’s CWA, this kind of temperature information could be included in
routine forecasts which utilize new technologies for
displaying temperature forecast information.
5. Acknowledgements
The
authors would like to thank the University of Missouri Alumni Association for
their financial support of this project. The authors would also like to thank
Dr. Stephen Mudrick, Dr. Milon George, Dr. Robert Pastoret,
Dr. Bruce Cutter, Dr. David Larsen, and Mr. Christopher Ratley, for their
participation in the experiment. We would also like to thank those
undergraduate (majors and non-majors) students who also participated in the
collection, analysis, and archival of the data or in lecture sessions related
to instrumentation and experimentation procedures. These include; Mr. Daniel
Robinson, Mr. Eric Kelsey, Ms. Sarah Thompson, Mr. Thaddeus Glynn, Ms. Megan
Ainsworth, Ms. Kelly Donohue, Ms. Jill Ahders, Mr.
Nicholas Mikulas, Mr. Brian Oravetz, Mr. Christopher Schimmer, and Mr. Daniel
Keating. Additionally, the authors would like to thank sincerely Mr. Robert J.
Ricks, Jr. (NWS WFO New Orleans/Baton Rouge) for his thorough review of this
contribution. We would also like to thank the United States Geological Survey
for providing Figure 2.
6. References
Ackerman, B., 1985: Temporal march of the Chicago heat island. J. Clim. And Appl. Meteor., 24, 547 – 554.
Aguado, E., and
J.E. Burt, 2001: Understanding Weather
and Climate, 2nd ed., Prentice Hall, Inc., 505 pp.
Changnon, S.A.,
2003: Urban modification of freezing rain events. J. Appl.
Meteor., 42, 863 – 870.
Changnon, S.A.,
1981: METROMEX: A review and summary,
Meteor. Monogr. No. 40,
Amer. Meteor. Soc., 181 pp.
DeMarrais,
G.A., 1975: Nocturnal heat island intensities and relevance to forecasting
mixing heights. Mon. Wea.
Rev., 103, 235 – 245.
Doeskin, N. J., and J.K.
Weaver, 2000: Microscale rainfall variations as
measured by a local volunteer network. Preprints
of the 12th conference on Applied Climatology, 8 - 11 May,
2000, Asheville, NC.
Glahn, H.R., and D.P. Ruth, 2003:
The new digital forecast database of the National Weather Service. Bull. Amer. Meteor. Soc., 84, 195 – 202.
Karl, T.R., and R.W. Knight, 1997: The 1995 Chicago
heat wave: How likely is a recurrence? Bull.
Amer. Meteor. Soc., 78, 1107 -
1120.
Melhuish, E., and
M. Pedder, 1998: Observing an urban heat island by
bicycle.
Weather, 53, 121 - 128.
Nicholls, N., 2001: The insignificance of
significance testing. Bull. Amer. Meteor.
Soc., 82, 981 - 986.
Oke,
T.R., 1982: The energetic basis of the urban heat island. Atmos. Envir., 7, 769 – 779.
Pinho, O.S.,
and M.D. Manso - Orgaz,
2000: The urban heat island in a small city in coastal Portugal. Int. J. Biomet., 44, 198 - 203.
Rozoff, C.M., W.R. Cotton, and
J.O. Adegoke, 2003: Simulation of St. Louis, Missouri, land use impacts on
thunderstorms. J. Appl. Meteor., 42, 716 – 738.
Rozoff, C.M., and W.R. Cotton,
2001: METROMEX revisited. Preprints of
the 15th Conference on Planned and Inavertent Weather
Modification, Albuquerque, NM, 14 - 18 January 2001.
Segal, M., and R.W. Arritt,
1992: Nonclassical mesoscale
circulations caused by surface sensible heat flux gradients. Bull. Amer. Meteor. Soc., 73, 1593 – 1604.
Shepherd,
J.M., H. Pierce, and A. Negri, 2002: Rainfall
modification by major urban areas from spaceborne
rain RADAR on the TRMM Satellite: J. Appl.
Meteor., 41, 689 – 701.