EPA
Uintcu States
Environmental Protection
Agency
Atmospheric Research and Exposuix
Assessment Laboratory
Research Trianele Park, XC 2771 1
Research and Development
Februarv, 1989
PROJECT REPORT
A CLIMATOLOGY OF
TEMPERATURE AND PRECIPITATION
VARIABILITY IN THE UNITED STATES
-------
A CLIMATOLOGY OF TEMPERATURE AND PRECIPITATION VARIABILITY
IN THE UNITED STATES
by
Brian K. Eder
Lawrence E. Truppi
Peter L. Finkelstein
Atmospheric Sciences Modeling Division
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ATMOSPHERIC RESEARCH AND EXPOSURE ASSESSMENT LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
ABSTRACT
This paper examines the seasonal and annual variance and
standardized range for temperature and the seasonal and annual
coefficient of variation and normalized standardized range for
precipitation, on a climatic division level for the contiguous
United States for the period 1895 to 1985.
Examination of the temperature variance reveals a
continentality phenomenon in which the largest variances occur in
the upper midwest section of the country, while the smallest
variances are generally found in coastal regions along the west
coast, the Gulf coast and southeastern states. The winter season
displays roughly twice the amount of seasonal variance as does
spring, and roughly four times that of summer or autumn.
Analysis of the standardized temperature range supports the
continentality phenomenon; however, the transitional seasons,
spring and autumn display the largest amount of within season
variability with winter and summer displaying the least amount.
Examination of the coefficient of variation for
precipitation depicts a propensity for the largest seasonal and
annual variation to occur over the southwestern states from Texas
to California. Conversely, the smallest coefficient of
variations are found over the northeastern sections of the
country from New England into the mid-Atlantic and Great Lakes
states. Analysis of the seasonal and annual standardized
precipitation range reveals that the pattern mimics the
coefficient of variation patterns, but does however, exhibit
less of a gradient, resulting in a smoother pattern. Areas of
greater than normal seasonal and annual precipitation ranges
include the southwestern states from Texas to California, while
areas of less than normal ranges include the northeastern and
Ohio River Valley states.
-------
1.0 INTRODUCTION
Despite the increasing interest shown by the scientific
community in climate and its interactions with the evolution of
ecosystem structures, there continues to be a lack of a consensus
among climatologists and ecologists concerning the future of
global climate and its possible impact upon ecosystems. Policy
makers, and planners as well, need plausible descriptions of
possible long-term changes of such ecologically important
variables as temperature, precipitation, evaporation and soil
moisture conditions on all spatial and temporal scales (Kellogg
and Schware, 1981).
Such descriptions may be found with climatic scenarios,
which are sets of solutions either derived empirically from
observational data (paleoclimatic or instrumental analogues), or
from Global Climate Models (GCMs), often in the form of seasonal
maps showing the range of conditions, or possible variances that
may occur in the future. Climatic scenarios are not meant to be
forecasts of future climates, but rather internally consistent
portrayals of plausible future climates, which can then be used
by other scientists in evaluating possible adverse impacts of
climatic change on man and the ecology, allowing for the
development of alternative strategies in order to mitigate such
impacts (Wigley et al., 1986).
Although research has begun in EPA's Atmospheric Research
and Exposure Assessment Laboratory, the development of climatic
-------
scenarios that have real utility for ecological impact assessment
is still, in somewhat of a rudimentary stage. Subsequently, this
development must be supported by an enhanced understanding of the
climatic sensitivities of a broad range of ecological activities
and of the detailed nature of recent and past climatic patterns
and their variability (Lamb, 1987). Two such variables which
should receive a concentration of research efforts are
temperature and precipitation. From these two measured
variables, numerous derived parameters relevant to local
ecosystems, such as surface moisture stress, duration of rainless
periods, and length of growing season can be calculated. The
development and evolution of ecosystems are as sensitive to the
ranges and variances of temperature and precipitation as they are
to mean conditions. Because of this, ecosystems evolving in
regions which have exhibited little variance in temperature and
precipitation over the years are likely to be more sensitive to
climatic changes than those ecosystems which evolved in regions
exhibiting larger variability. Therefore, a need exists to not
only delineate these regions of differing variance, but to also
establish monitoring networks within both types of regions, which
may provide an understanding of potential ecological responses
toward future climatic change.
Though the delineation of such regions may seem to be
trivial, little if any literature concerning the subject is
available. Cayan et al., (1986) produced an atlas examining the
monthly and seasonal temperature anomalies over the United States
for the period 1930 through 1984. This work however, does not
-------
fulfill all of the needs discussed above, in that full
utilization of the data is not accomplished (accurate records
extend into the last century), nor is the variance of
precipitation analyzed.
This paper therefore represents an initial effort toward the
fulfillment of the requirements mentioned above through the
delineation of areas of the country which experience differing
amounts of temperature and precipitation variability. This is
accomplished through the examination of the variance and
standardized range (as defined in Section 3.0) of temperature
data and the coefficient of variance and standardized range of
precipitation data across the contiguous United States, on a
climatic division level, from the period 1895 through 1985.
Establishment of monitoring networks within these delineated
regions will help provide a new understanding of key ecosystem
processes, as well as their responses to possible climatic
change, which should therefore enhance their treatment in GCM
based scenarios as well as pave their way for their
representation in observationally based scenarios (Lamb, 1987).
This paper is divided into five sections. Following this
introduction is a section discussing the acquisition and
preparation of the data employed in the analysis, which is then
followed by a section examining the statistical techniques used
to prepare the annual and seasonal maps. And finally, the
results of the analysis are discussed followed by a brief
summarization.
-------
2.0 DATA
The monthly temperature and precipitation data employed in
this analysis were obtained from the National Climatic Data
Center (NCDC) located in Asheville, NC. These data, which cover
the period 1895 to 1985, are collected on a climatic division
basis, where each climatic division is designed to represent
regions within a state that are climatically homogeneous or
consistent. Within the contiguous United States, there are 344
such divisions, as depicted in Figure 1 and listed in Table 1.
The areal coverage of the divisions can vary tremendously, with
the largest divisions generally found in the western states and
the smallest found in the east.
Stations used in calculating the divisional monthly averages
of temperature (measured to the nearest tenths in degrees F) and
the monthly totals of precipitation (measured to the nearest
hundreths in inches) include all first order stations and those
cooperative stations which have maintained consistent records.
An equal-weight approach is used for each of the stations located
within the division, the number of which can vary significantly
from one division to the other depending upon the size and
demographics of the division. Figures 2 and 3, which depict on a
state basis the average number of square miles per station for
temperature and precipitation data, respectively, provide a feel
for this density.
Unfortunately, inadvertent bias has been introduced into the
-------
U.S. CLIMATOLOGICAL DIVISIONS
1895 - 1985
J VALID DATA
REJECTED DATA
SUBSTITUTE DATA
Figure 1. U. S. Climatological Divisions
-------
TABLE I
U.S. CLIMATOLOGICAL DIVISIONS
01 - ALABAMA
01 Northern Valley
02 Appalachian Mountain
0) Upper PUini
04 Eutern Valley
05 Piedmont Plateau
09 Prairie
07 Coailal Plain
08 Gulf
OJ - ARIZONA
01 Northwest (R)
0] Northeait
OJ North Central
04 Eut Central (S)
05 Southwell
0« South Central
07 Southeait (S)
0$. ARKANSAS
01 Northwest
0] North Central
0) Northeait
04 Weil Central
05 Central
08 Eul Central
07 Sonthwnt
03 South Central
09 Southeait
04 - CALIFORNIA
01 North Coait Drug.
01 Sacramento Drag.
OS Northeall Inter. Basins
04 Central Coail Dn>|.
05 San Joaquin Drag.
08 South Cout Drng.
07 Southeait Desert Baiini
05 - COLORADO
01 AR Drainage Bruin
01 CO Drainage Daiin (S)
01 KS Drainage Baiin
04 Plalte Drainage Baiin
05 Rio Grande Drng. Balin
OS - CONNECTICUT
01 Northweat
01 Central
03 Coailal
07 - DELAWARE
01 Northern
01 Southern
08 - FLORIDA
01 North.ml
01 North
03 North Central
04 South Central
05 Evergladei U SW Coait
08 Lower Eait Coait
07 Keyi
09. GEORGIA
01 Northweit
02 North Central
03 Northeait
04 Weit Central
05 Central
04 Eait Central
07 Soulhwe»t
08 South Central
09 Southeait
10-IDAHO
01 Panhandle
01 North Central Prairies
03 North Central Canyoni
04 Central Mountain!
05 Southwell Valleys
CX5 Southwell Highland! (R)
07 Central Plaint
08 Northeall Valley!
09 Upper Snake River Plain!
10 Eut Highlandi
11 - ILLINOIS
01 Northweit
01 Northeait
03 Weit
04 Cenlral
05 Eait
CM Welt Southwell
07 Eait Southeait
08 Southwell
09 Southeail
11 - INDIANA
01 Northweit
01 North Central
03 Northeast
04 Wot Central
05 Central
04 Eait Central
07 Southwell
08 South Central
09 Sontheut
1J - IOWA
01 Northwest
01 North Cenlral
03 Northeait
04 Weit Central
05 Central
04 Eait Cenlral
07 Southwest
08 South Cenlral
09 Southeail
14-KANSAS
01 Northweit
01 North Cenlral
03 Northeait
04 West Central
05 Cenlral
04 East Central
07 Soulhweit
08 Sooth Cenlral
00 Soulheail
15 - KENTUCKY
01 Weitern
01 Central
03 Blue Grass
04 Eastern
18 - LOUISIANA
01 Northwest
01 North Central
03 Northeall
04 Weit Central
05 Central
0« Eait Central
07 Southwest
08 South Central
09 Southeail
17- MAINE
01 Northern
01 Southern Interior
03 Coailal
18 - MARYLAND k DC
01 Soulheailern Shore
01 Cenlral Eailern Shore
03 Lower Southern
04 Upper Southern
05 Northeaitern Shore
00 Northern Cenlral
07 Appalachian Mounlain
08 Allegheny Plateau
19 - MASSACHUSETTS
01 Western
02 Central
03 Coaital
10- MICHIGAN
01 Weit Upper
01 East Upper
03 Northwest Lower
04 Northeast Lower
05 Weil Central Lower
08 Central Lower
07 East Central Lower
08 Southwell Lower
09 South Cenlral Lower
10 Soulheail Lower
11 - MINNESOTA
01 Northweit
01 North Central
03 Northeait
04 Weit Cenlral
05 Cenlral
OA Eait Cenlral
07 Soulhweil
08 South Central
09 Southeast
11 - MISSISSIPPI
01 Upper Delta
02 North Central
03 Northeait
04 Lower Delta
05 Central
08 Eait Cenlral
07 Southwest
08 South Central
09 Southeail
10 Coaital
IS - MISSOURI
01 Northweit Prairie
01 Northeait Prairie
03 Weil Cenlral Plain!
04 Wnt Osarki
05 Eait Oiarki
OA Boolheel
24 - MONTANA
01 Weitern
02 Southwestern
03 North Central
04 Central
05 South Central
08 Northeaitern
07 Southeastern
25 - NEBRASKA
01 Panhandle
01 North Central
03 Northeait
05 Central
OA Eait Central
07 Southwest
08 South Cenlral
09 Sonlheait
18 - NEVADA
01 Northwestern
01 Northeulern (S)
01 Sonlh Cenlral (S)
04 Extreme Southern
17 - NEW HAMPSHIRE
01 Northern
01 Southern
28 - NEW JERSEY
01 Northern
01 Southern
03 Coait al
29 - NEW MEXICO
01 Northweitern Plateau
01 Northern Mounlaini
03 Northeailem Plain!
04 Southwestern Mountains
05 Cenlral Valley
08 Central Highland!
07 Sonlheailern Plaint
08 Southern Desert
30 - NEW YORK
01 Western Plateau
01 Eastern Plateau
03 Northern Plateau
04 Coaital
05 Hudson Valley
OA Mohawk Valley
07 Champlain Valley
08 St. Lawrence Valley
09 Great Lakes
10 Central Lakes
31 - NORTH CAROLINA
01 Southern Mountain!
02 Northern Mountain! (S)
03 Northern Piedmont
04 Central Piedmont
05 Southern Piedmont
04 Southern Coailal Plain
07 Central Coaital Plain
08 Northern Coutal Plain
32 - NORTH DAKOTA
01 Northweil
01 North Central
03 Northeait
04 West Central
05 Central
OA Eait Central
07 Southwell
08 South Central
09 Soulheail
3S - OHIO
01 Northwest
02 North Central
03 Northeast
04 West Central
05 Cenlral
OA Cenlral HUli
07 Northeait Hilli
08 Southwell
09 Sonlh Central
10 Southeast
94 - OKLAHOMA
01 Panhandle
01 North Central
03 Northeall
04 Weit Central
05 Central
04 Eut Central
07 Southwest
08 Soilh Central
09 Southeast
35 - OREGON
01 Coastal Area
02 Willametle Valley
03 Southwestern Valleys
04 Norlhem Cascades (S)
05 High Plalean (R)
OA North Central
07 South Central
08 Northeast
09 Southeait
3A - PENNSYLVANIA
01 Pocono Mountain! (R)
02 Eait Central Mounlains
03 Soulheaslern Piedmonl
04 Lower Susquehanna
05 Middle Susquehanna
OA Upper Susquehanna
07 Cenlral Mountains
08 South Central Mountains
09 Southwest Plateau
. 10 Northwest Plateau
37 - RHODE ISLAND
01 All
38 - SOUTH CAROLINA
01 Mountain (R)
01 Northwest
03 North Central
04 Northeait
05 West Cenlral
04 Central
07 Southern
39 - SOUTH DAKOTA
01 Northwest
01 North Central
03 Northeast
04 Black HUli (S)
05 Southwest
04 Central
07 Eait Central
08 South Central
09 Southeail
40 - TENNESSEE
01 Eastern
01 Cumberland Plateau
03 Middle
04 Western
41 - TEXAS
01 High Plains
01 Low Rolling Plain!
OS North Central
04 Eait Texas
05 Tram Pecoi
04 Edwardi Plateau
07 South Central
08 Upper Cout
09 Southern
10 Lower Valley
41 - UTAH
01 Western
01 Dixie (S)
03 North Central
04 South Central (R)
05 Northern Mounlaini
08 Uinla Buin (R)
07 Southeut (S)
43 - VERMONT
01 Northeastern
01 Weslern
03 Southeutem
44 - VIRGINIA
01 Tidewater
01 Eastern Piedmont
03 Western Piedmont
04 Northern
05 Central Mountain
08 Southwestern Mountain
45 - WASHINGTON
01 West Olympic Coastal
01 NE Olympic San Juan
03 Pugel Sound Lowlands
04 E Olymp Cascade Foolh
05 Cascade Mountain! Weil
08 Eut Slope Cascades
07 Okanogan Big Bend
08 Central Buin
09 Northeulern
10 PalouM Blue Mountains
48 • WEST VIRGINIA
01 Northwestern
01 North Central
03 Southwestern
04 Central
05 Southern
08 Northeastern
47-WISCONSIN
01 Northwesl
01 North Central
03 Northeait
04 Weil Central
05 Central
08 Eut Central
07 Southwest
08 South Central
09 Sonlheut
48 - WYOMING
01 Yellowstone Drainage
01 Snake Drainagi
03 Green and Bear Drainag
04 Big Horn
05 Powdr.Ltl Mo.Tongne
08 Belle Fourth! Drainage
07 Cheyenne & Niobrara
08 Lower PlatU
09 Wind Hirer
10 Upper Plalte
Table 1. U.S. Climatological Divisions
-------
HISTORICAL CLIMATE DATA 1895 - 1985
TEMPERATURE STATION COVERAGE (SO MI/STATION)
00
COVERAGE
+ 2000
1500 TO 1000
LESS THAN 500
2000 TO 1500
1000 TO 500
Figure 2. Temperature Station Coverage (Square Mile/ Station)
-------
HISTORICAL CLIMATE DATA 1895 - 1985
PRECIPITATION STATION COVERAGE (SQ MI/STATION)
COVERAGE
+ 2000
1500 TO 1000
LESS THAN 500
2000 TO 1500
1000 TO 500
Figure 3. Precipitation Station Coverage (Square Mile/Station)
-------
data set, which has resulted in numerous problems. One such
problem is that the actual number of stations, as well as their
spatial distribution within each division, has varied over the
ninety-one year period from 1895 to 1985. Station changes such
as these can introduce sampling (not climatic) variability into
the data set, especially in those climatic divisions which have
large geographic variability. Additional bias was also
inadvertently introduced when the observation time at the
cooperative stations changed from late afternoon to early
morning.
For the most part, these potential errors and biases have
been estimated then systematically removed from the data set
(Karl et al., 1986); however, of the 344 climatic divisions used
in this study, 17 still contained an unacceptable amount of bias.
The majority of these divisions were located in mountainous
areas, as seen again in Figure 1. For ten of these problem
divisions, (classified as Substitute divisions and indicated by
the slashed lines) the NCDC was able to substitute proxy records
by obtaining data from one or two consistent stations within
that division. Unfortunately, suitable replacements were not
available for the remaining seven divisions, which were
classified as rejected and indicated by the cross-hatching. For
this analysis, the temperature and precipitation data in the
rejected divisions were replaced by taking an average of the data
collected from surrounding divisions, so there would not be data
gaps or holes in this analysis. Results for these seven
divisions must be treated with caution.
10
-------
The analyses in this report are displayed graphically on a
climatic division level, using software developed by SAS, Inc.
(Statistical Analysis Systems, 1985). Unfortunately, this
software system only recognizes state and county boundaries, and
does not recognize climatic division boundaries. For the
overwhelming majority of divisions this presented no problem as
most are defined in terms of county boundaries. However, there
are divisions, most notably in the Rocky Mountain states, where
county lines do not exactly define climatic divisions; therefore,
some division boundaries have been approximated from the county
boundaries.
11
-------
3.0 METHODOLOGY
The seasonal and annual variability of both temperature and
precipitation were examined in order to better understand the
variability of climate within the contiguous U.S. For
temperature data this consisted of examining the variance from
season to season, and by examining the range within season
(standardized over the United States). Because precipitation
inherently has more variance, the coefficient of variation was
used to examine season to season variability, while the
normalized ranges were used to examine the within season
variability (also standardized over the United States).
3.2 Temperature
For each climatic division the variance of temperature was
calculated for the annual average as well as for the seasonal
averages for each of the four seasons. For simplicity, only the
annual average temperature will be used in defining the
statistical procedure. The variance, (S2), is defined as
follows:
N
- X)^
(1)
N - 1;
where X^ is the temperature averaged over the 12 month period for
each year, for each climatic division, and X is the average for
12
-------
that division over the j = 1 to 91 year period. Calculation of
the annual variance indicates the variability of the temperature
that occurs between years. Calculation of the seasonal variances
is accomplished similarly and indicates the variability of the
temperature that occurs between seasons (i.e. between winters).
Another way of examining the annual variability of
temperature, is to examine the standardized range that occurs
within each year, which provides a feel for the within year
variability. Standardization of the temperature range allows for
direct comparison between individual climatic divisions and the
country as a whole. The standardization was performed across the
i = 1 to 344 climatic divisions as seen below.
N _
V (Rji - R)
Standardized 1 ^~\
Temperature Range/jx = (2)
N SR
where for climatic divisions i and year j , R^-; is the temperature
range exhibited within a specific year (the maximum monthly
average temperature minus the minimum monthly average
temperature) for the i = 1 to 344 divisions and j = 1 to 91
years. R is the average range over the 344 climatic divisions
and 91 seasons or years and S is the standard deviation of the
Rji's over the same divisions and time periods.
3.3 Precipitation
Due to the tremendous range in normal precipitation
13
-------
exhibited over the United States, a different approach was
necessary for the seasonal and annual precipitation analysis.
Rather than take the variance, which would be biased towards
areas of high precipitation, the coefficient of variation was
examined which "normalizes" the variance as seen in the equation
below:
C. V. = S/ X; (3)
where S is the standard deviation of the precipitation data for a
particular climatic division and X is the mean precipitation over
the 91 year period for that division.
Similarly, calculation of the standardized range also
considered this extreme variability in precipitation and was
therefore calculated using a normalized version of equation (2)
above, as seen below:
Standardized 1 N / R^-; R
Precipitation Range/JN =
E
N
J "*• "J (4)
SR
where RJ^J is the precipitation range for the i = 1 to 344
climatic divisions and j = 1 to 91 years. R is the average
precipitation range over the 344 climatic divisions and 91 time
periods, and SR is the standard deviation over the same divisions
and period. T^^ is the total precipitation for division i and
time period j, and T is the average total precipitation over all
divisions and periods.
14
-------
4.0 RESULTS
Results of the analyses are presented in Figures 4 to 23,
with the first map in each of the four series depicting the
annual analyses, and the subsequent maps depicting the winter,
spring, summer and autumn analyses. Different hatching types are
used to display the ranges of the different analyses. Whenever
possible, consistent ranges were used across seasons and plots;
however, due to the varying nature of the variables investigated
this often proved to be infeasible.
4.1 Temperature Variance
Examination of Figure 4, which depicts the annual
temperature variance reveals several interesting features. Most
notable of these features is the tendency for the largest
variance to occur in the upper midwest portions of the country,
especially in North and South Dakota and eastern Montana, where
the annual temperature variance exceeds 3° F. A trend toward
decreasing annual variance is exhibited as climatic divisions
approach coastal regions. This pattern is depicted especially
well along the west coast from Washington and Oregon to
California, and again along the Gulf coast and southeastern
states, where the annual temperature variance reaches a minimum
of less than 0.5° on the southern Florida peninsula.
This phenomenon of large variances in the center of the
country and smaller variances near coastal areas is a direct
15
-------
consequence of a region's continentality, and the subsequent
differences found between the heat capacity of land and ocean
masses. Coastal areas tend to experience modified, maritime
climates, generally free of temperature extremes, whereas
interior areas experience continental type climates where
temperature extremes are more common.
Examination of the seasonal variances (Figures 5 through 8)
also reveals this continentality phenomenon; however it is
interesting to note that the temperature variance exhibited
during the winter is much stronger than during the other
seasons. In fact the winter variance, which ranges from 5 to
20°, is roughly twice that for the spring, which ranges from 2 to
10 and four times that of the summer and autumn, which range
from 1 to 5° and from 2 to 6°, respectively. It is also worth
noting that the area of maximum variance shifts southward during
the summer, from the northern to the central plains. The maps do
however, depict a tendency towards consistency between time
periods, in that the range of variance within each map is roughly
a factor of four (from the minimum variance found on the map to
the maximum variance) for each season and the year.
16
-------
U.S. ANNUAL TEMPERATURE VARIANCE
1895 - 1985
I LESS THAN 0.51
1.01 TO 1.50
2.01 TO 2.50
MORE THAN 3.00
TO 1.00
TO 2.00
TO 3.00
Figure 4. Annual Temperature Variance (°F)
-------
U.S. WINTER TEMPERATURE VARIANCE
1895 - 1985
oo
J LESS THAN 5.01
10.01 TO 15.00
MORE THAN 20.00
KXXXI 5.01 TO 10.00
15.01 TO 20.00
Figure 5. Winter Temperature Variance (°F)
-------
U.S. SPRING TEMPERATURE VARIANCE
1895 - 1985
j LESS THAN 2.01
4.01 TO 6.00
8.01 TO 10.00
NNNN1 2.01 TO 4.00
6.01 TO 8.00
MORE THAN 10.00
Figure 6. Spring Temperature Variance (°F)
-------
U.S. SUMMER TEMPERATURE VARIANCE
1895 - 1985
[S3
o
! LESS THAN 1.01
2.01 TO 3.00
4.01 TO 5.00
1.01 TO 2.00
3.01 TO 4.00
MORE THAN 5.00
Figure 7. Summer Temperature Variance (°F)
-------
U.S. AUTUMN TEMPERATURE VARIANCE
1895 - 1985
1 LESS THAN 2.01
3.01 TO 4.00
5.01 TO 6.00
KXXX1 2.01 TO 3.00
4.01 TO 5.00
MORE THAN 6.00
Figure 8. Autumn Temperature Variance (op)
-------
4.2 Standardized Temperature Range
Figure 9, which depicts the annual standardized temperature
range exhibits, in a somewhat different manner the same
continentality as seen with the variance figures. Since the data
are now represented in a standardized format, values above and
below a mean of zero are plotted. Assuming that the standardized
temperature range data are normally distributed, roughly 20% of
the climate divisions would have standardized ranges within (+/-)
0.25, while 55% would have ranges within (+/-) 0.75, and 78%
would have ranges within (+/-) 1.25, and finally 92% would have
standardized ranges within (+/-) 1.75. Consistent with the
annual map, the largest seasonal standardized ranges occur in the
upper midwest, especially in the states of North and South Dakota
and Minnesota. A trend toward decreasing seasonal ranges are
found near the coastal areas, especially along the Pacific Coast
states and the Gulf Coast states.
A narrow zone of "normal" standardized ranges (between +/-
0.25), depicted by the absence of hatching, can be found
extending from the southern New England coast through the Ohio
River Valley into the lower midwest and into the Rocky mountain
states. This transitional zone separates areas of higher than
"normal" seasonal ranges from areas of lower than "normal"
ranges.
Unlike the seasonal variance maps which depicted winter as
the season having the most variance, the seasonal standardized
range maps (Figures 10 through 13) depict the transitional
22
-------
seasons, spring and autumn as exhibiting the most variability
within their seasons. This phenomenon is not unexpected since
the range of monthly temperature would be greater during the
transitional seasons than during winter or summer.
It is also interesting to note that the transitional zone
from negative to positive anomalies maintains the position seen
earlier with the annual map. The size of this zone however
increases with the seasonal analyses.
23
-------
U.S. ANNUAL STANDARDIZED TEMPERATURE RANGE
1895 - 1985
! LESS THAN -1.75
-0.76 TO -1.25
3 -0.25 TO 0.25
0.76 TO 1.25
MORE THAN 1.75
„_ -1.26 TO -1.75
1NNNSI -0.26 TO -0.75
0.26 TO 0.75
1.26 TO 1.75
Figure 9. Annual Standardized Temperature Range
-------
U.S. WINTER STANDARDIZED TEMPERATURE RANGE
1895 - 1985
LESS THAN -0.75
-0.25 TO 0.25
MORE THAN 0.75
-0.26 TO -0.75
0.26 TO 0.75
Figure 10. Winter Standardized Temperature Range
-------
U.S. SPRING STANDARDIZED TEMPERATURE RANGE
1895 - 1985
M
I LESS THAN -1.75
-0.76 TO -1.25
] -0.25 TO 0.25
0.76 TO 1.25
MORE THAN 1.75
-1.26 TO -1.75
-0.26 TO -0.75
0.26 TO 0.75
1.26 TO 1.75
Figure 11. Spring Standardized Temperature Range
-------
U.S. SUMMER STANDARDIZED TEMPERATURE RANGE
1895 - 1985
NJ
-J
LESS THAN -0.75
-0.25 TO 0.25
MORE THAN 0.75
-0.26 TO -0.75
0.26 TO 0.75
Figure 12. Summer Standardized Temperature Range
-------
U.S. AUTUMN STANDARDIZED TEMPERATURE RANGE
1895 - 1985
to
CO
1 LESS THAN -1.75
-0.76 TO -1.25
3 -0.25 TO 0.25
0.76 TO 1.25
MORE THAN 1.75
-1.26 TO -1.75
[NNSX1 -0.26 TO -0.75
0.26 TO 0.75
1.26 TO 1.75
Figure 13. Autumn Standardized Temperature Range
-------
4.3 Precipitation Coefficient Of Variation
Examination of Figure 14, which depicts the precipitation
coefficient of variation (%) reveals several interesting
features. Unlike the temperature analysis, which indicated a
north-south gradient, the precipitation analysis depicts somewhat
of an east-west gradient. This is supported by the propensity
for the largest coefficient of variation to occur over the
southwestern states from Texas to California, where the values
exceeds 25.9%, while the smallest variation generally occur over
the eastern sections of the country from the mid-Atlantic and
Great Lake States into New England, where values are less than
14.0%. In some respects, interpretation of the precipitation
maps is more complicated than the temperature maps in that the
anomaly patterns are not as smooth as those seen for temperature.
This is especially true of the Rocky Mountain states, where large
ranges in variations occur over adjacent climatic divisions.
The maps depicting the coefficient of variation for
seasonal precipitation (Figures 15 through 18) are, with only a
few exceptions, similar to the annual map. Most notable of
these exceptions is the extension or shift of high variations
into the lower midwestern states during the winter season, and
into the Pacific coast states during the summer season. Although
the summer season seems to exhibit somewhat less variation on a
nationwide basis than the other seasons, this decrease is small
when compared to the changes seen in the seasonal temperature
variances. In general, the coefficients of variation range from
25 to 55% for each of the seasons.
29
-------
U.S. ANNUAL PRECIPITATION COEFFICIENT OF VARIANCE (%)
1895 - 1985
OJ
o
3 Less Than 14.0
18.0 To 21.9
More Than 25.9
V//7A 14.0 To 17.9
22.0 To 25.9
Figure 14. Annual Precipitation Coefficient of Variation ($)
-------
U.S. WINTER PRECIPITATION COEFFICIENT OF VARIANCE (%)
1895 - 1985
Less Than 25.0
35.0 To 44.9
More Than 54.9
1////A 25.0 To 34.9
45.0 To 54.9
Figure 15. Winter Precipitation Coefficient of Variation
-------
U.S. SPRING PRECIPITATION COEFFICIENT OF VARIANCE (%)
1895 - 1985
OJ
N3
I J Less Than 25.0
35.0 To 44.9
More Than 54.9
25.0 To 34.9
45.0 To 54.9
Figure 16. Spring Precipitation Coefficient of Variation (%)
-------
U.S. SUMMER PRECIPITATION COEFFICIENT OF VARIANCE (%)
1895 - 1985
OJ
j Less Than 25.0
35.0 To 44.9
More Than 54.9
25.0 To 34.9
45.0 To 54.9
Figure 17. Summer Precipitation Coefficient of Variation (%)
-------
U.S. AUTUMN PRECIPITATION COEFFICIENT OF VARIANCE (%)
1895 - 1985
\ _l Less Than 25.0
35.0 To 44.9
More Than 54.9
25.0 To 34.9
45.0 To 54.9
Figure 18. Autumn Precipitation Coefficient of Variation (%)
-------
4.4 Standardized Precipitation Range
Examination of the annual standardized precipitation range
map (Figure 19), reveals patterns similar to those of the
precipitation coefficient of variation. The southwestern states
from New Mexico to California tend to have larger annual ranges
when compared to the rest of the country. Another area exhibiting
annual ranges which are greater than "normal" is found in the
upper midwest from North and South Dakota into Montana. Areas
exhibiting smaller than "normal" annual ranges include the New
England and Appalachian Mountain states. Areas which tend to
exhibit "normal" amounts of annual ranges are generally scattered
throughout the country and include some of the Rocky Mountain and
mid Mississippi Valley states.
Figures 20 through 23 which depict the standardized
seasonal ranges of precipitation again somewhat mimic the annual
map; the patterns, however, tend to be somewhat flatter,
indicating less within seasonal variability. Areas of greater
than "normal" precipitation ranges include the southwestern
states from Texas to California, while the eastern states,
especially those in New England and the Ohio River Valley, tend
to exhibit less than "normal" ranges.
As was seen with the precipitation coefficient of variation,
which exhibited less variance from season to season than did the
temperature variance, the standardized range of precipitation
exhibits less variability between seasons when compared to the
standardized temperature range.
35
-------
U.S. ANNUAL STANDARDIZED PRECIPITATION RANGE
1895 - 1985
3 Less Than -1.25
-0.75 To -0.26
0.26 To 0.75
More Than 1.25
-1.25 To -0.76
-0.25 To 0.25
0.76 To 1.25
Figure 19. Annual Standardized Precipitation Range
-------
U.S. WINTER STANDARDIZED PRECIPITATION RANGE
1895 - 1985
u>
1 '
Less Than -0.75
-0.25 To 0.25
More Than 0.75
-0.75 To -0.26
0.26 To 0.75
Figure 20. Winter Standardized Precipitation Range
-------
U.S. SPRING STANDARDIZED PRECIPITATION RANGE
1895 - 1985
00
Less Than -0.75
3 -0.25 To 0.25
More Than 0.75
INSXSl -0.75 To -0.26
0.26 To 0.75
Figure 21. Spring Standardized Precipitation Range
-------
U.S. SUMMER STANDARDIZED PRECIPITATION RANGE
1895 - 1985
Less Than -0.75
-0.25 To 0.25
More Than 0.75
_____ -0.75 To -0.26
IXXXXJ 0.26 TO 0.75
Figure 22. Summer Standardized Precipitation Range
-------
U.S. AUTUMN STANDARDIZED PRECIPITATION RANGE
1895 - 1985
Less Than -0.75
I -0.25 To 0.25
More Than 0.75
-0.75 TO -0.26
0.26 To 0.75
Figure 23. Autumn Standardized Precipitation Range
-------
5.0 SUMMARY
Because there continues to be no consensus among
climatologists and ecologists concerning climate change and its
possible impact upon ecosystems, the development of climatic
scenarios will be necessary in order to assist scientists in
evaluating possible adverse effects of climatic change on the
ecology. Unfortunately, the development of such scenarios as a
utility in assessing this impact is still somewhat in a
rudimentary stage, and therefore must be supported by an enhanced
understanding of recent and past climatic patterns and their
variability. In an initial attempt to assist in this
understanding, this paper has examined the seasonal and annual
variance and standardized range for temperature and the seasonal
and annual coefficient of variation and normalized standardized
range for precipitation, on a climatic division level for the
contiguous United States for the period 1895 to 1985.
Examination of the temperature variance revealed a
continentality phenomenon in which the largest variance occurred
in the upper midwest section of the country, while the smallest
variance were generally found in coastal regions along the west
coast, the Gulf coast and southeastern states. The winter season
displayed roughly twice the amount of seasonal variance as did
spring, and roughly four times that of summer or autumn.
Analysis of the standardized temperature range supports the
continentality phenomenon; however, the transitional seasons,
spring and autumn displayed the largest amount of within season
41
-------
variability with winter and summer displaying the least amount.
Examination of the coefficient of variation for
precipitation depicted a propensity for the largest seasonal and
annual variation to occur over the southwestern states from Texas
to California. Conversely, the smallest coefficient of
variations were found over the northeastern sections of the
country from New England into the mid-Atlantic and Great Lakes
states. There is less of a seasonality effect with the
precipitation maps when compared to the temperature maps, in that
the relative variations do not changes as much from season to
season.
Analysis of the seasonal and annual standardized
precipitation range reveals that the pattern mimics the
coefficient of variation patterns, but does however, exhibit
less of a gradient, resulting in a smoother pattern. Areas of
greater than normal seasonal and annual precipitation ranges
include the southwestern states from Texas to California, while
areas of less than normal ranges include the northeastern and
Ohio River Valley states.
Successful climate scenarios, whether derived from climate
models or analogue techniques, should duplicate the patterns
produced in this paper as well as the simple mean patterns.
Present models are, for the most part, unable to do this. The
design of ecological monitoring networks, both for base line
stations, which require some climatic stability, and for stations
where a range of climatic conditions is required should also be
cognizant of the information developed in this and similar
studies.
42
-------
6.0 REFERENCES
Cayan, D. R. , Ropelewski, C. F. and T. R. Karl (1986). An Atlas
of United States Monthly and Seasonal Temperature Anomalies
December 1930 - November 1984.
Karl, T. R. , Williams, Jr., C. N. , Young, P. J. and W. M.
Wendland (1986). A Model To Estimate The Time Of Observation
Bias Associated With Mean Monthly Maximum, Minimum, And Mean
Temperature For The U. S. J. Clim. & Appl. Meteor., 25.
Kellogg, W. W. , and R. Schware (1981). Climatic Change And
Society: Consequences Of Increasing Atmospheric Carbon
Dioxide. Westview Press, Boulder, CO, 178 pp.
Lamb, P. J. (1987) . On The Development Of Regional Climatic
Scenarios For Policy Oriented Climatic Impact Assessment.
Bull. Amer. Meteor. Soc., 68.
SAS Institute (1985). Statistical. Analysis System User's Guide:
Statistics, Version 5 Edition, SAS Institute, Inc., Gary, NC
Wigley, T. M. L., Jones, P. D. and P. M. Kelly (1986). Empirical
Climate Studies: Warm World Scenarios And The Detection Of
CO2 Induced Climatic Change Induced By Radiatively Active
Gases. Chapter 6, The Greenhouse Effect, Climate Change,
and The Ecosystems. John Wiley & Sons, Chichester, 271-322.
(B. Bolin et al., eds)
43
------- |