CHANGING THERMAL TOPOGRAPHY BALTIMORE-WASHINGTON
OF THE
CORRIDOR:
1950-1979
ARTHUR VITERITO
Department of Geography, The George Washington University, Washington, D.C. 20052, U.S.A.
Abstract. A study of the temperature field for the Baltimore-Washington region reveals that since 1950 there has been the development of an urban "heat corridor'. Trend surface analysis shows that there has been an inversion of the thermal topography in the region as a 'saddle' of rising temperatures has emerged, replacing a trough of lowered temperatures through three decades. Steady population growth throughout the area is seen to be the most important contributor. The strengthening of the heat corridor is best expressed in the summer months and weakest during the winter months. As population continues to grow in this region, the thermal topography is sure to be modified even further. It is recommended that further study be devoted to temperature and precipitation changes in regions experiencing urban growth.
1. Introduction
Numerous studies have amply confirmed that urbanization significantly modifies the overlying atmosphere. The best documented effect is the urban heat island phenomenon. Specifically, urban areas exhibit higher temperatures than their rural counterparts and this effect is due to a number of factors, most notable being the radically different radiative and thermal characteristics between urban and rural surface cover. The intensity of the heat island is highly variable and is a function of urban size, topography, season, wind speed, humidity and cloud cover, population density, and amount of industrial activity (Landsberg, 1974). With regard to urban size, strong positive relationships between urban population and heat island intensity have been established. For example, Oke (1973) has derived a log-linear relationship for North American cities of the form:
dZ(u-r)max = 3.06 LOG P - 6.79,
where d T is the maximum urban/rural temperature difference and P is urban population. Landsberg (1975) corroborates the model's validity in his study of population growth and the strength of Columbia, Maryland's heat island. Similar studies by Mitchell (1961), Bomstein (1968), Moffltt (1972), Karl (1985), and Kukla et al. (1986), clearly establish the heat island intensity/population link. Climatic Change 14: 89-102, 1989. 9 1989 Kluwer Academic Publishers. Printed in the Netherlands.
90
Arthur Viterito
Fig. 1. The Baltimore-Washington Corridor.
BALTIMORE CO.
I I
HOWARD
MONTGOMERY
-- -- -9
I L .
0 L-
10 I
~ .
.
.
Grid Boundary Station
I o------ 9
-J
20 I
mi
Fig. 2. Station locations and grid boundary.
A related issue in urban climatology which has received scant attention is the thermal modifications of entire regions due to large scale urbanization. As much of today's urban growth is occurring in the rural/suburban continuum away from city centers, theory would predict that such extra-urban development should give rise to expanded heat islands for isolated cities and to 'heat
91
Changing Thermal Topography of the Baltimore- Washington Corridor: 1950-1979
corridors' between neighboring cities. This 'heat corridor' effect would evolve from the expansion and merger of proximal urban heat islands, thus producing an inter-urban zone of temperatures intermediate in value of the terminal urban centers which bound the corridor and the outlying rural areas. One such corridor which is emerging is the area between Baltimore, Maryland and Washington, D.C.. The Baltimore-Washington corridor (BWC) will be defined in this study as the region bounded by Annapolis, MD to the east, Washington, D.C. to south, Towson, MD to the north and Great Falls, MD to the west (see Figures 1 and 2). Population in the corridor has experienced rapid development between 1950 and the present and Table I provides information on the population characteristics in the corridor from 1950 to 1980. As evidenced by the statistics, population in the region has grown steadily throughout the period. TABLE I: Population statistics for the Baltimore-Washington corridor. County
1950
Anne Arundel Baltimore Baltimore City Howard Montgomery Prince Georges District of Columbia Total
1960
1970
1980
117,392 270,273 949,708 23,119 164,401 194,182 802,178
206,634 492,428 939,024 36,152 340,928 357,395 763,956
297,539 621,077 905,757 61,911 522,809 660,564 756,492
370,775 655,615 786,775 118,572 579,053 665,071 638,333
2,521,253
3,136,526
3,826,149
3,814,194
The hypothesis of this study is that the substantial population growth in the BWC has caused significant modifications to the region's thermal topography for the period 1950-1979. That is, the development of a heat corridor is evident and will continue to intensify as population growth continues. 2. Data and Methodology
For this study the mean monthly temperatures for 27 stations in the corridor were analyzed (Figure 2). The selected network is composed of first and second order stations that have a minimum of 10 yr temperature data. Table II gives station locations and site characteristics. One problem which is readily apparent in assessing the thermal topography of the region is the uneven spatial coverage these stations provide. In order to compensate for this difficulty an evenly spaced grid of interpolated values was prepared using an inverse distance routine (Unwin, 1981). A 15 by 16 grid of temperature data was generated for each month in the period. The value of each point in the grid was determined as:
92
Arthur Viterito TABLE II: Station locations and site characteristics. Station
Lat.
Long.
Site
Annapolis Baltimore (airport) Baltimore (city) Beltsville Cheltenham Clarksville College Park Dalecarlia Reservoir District Heights Dundalk Fort Meade Glenn Dale Bell Station Great Falls Laurel Lutherville National Arboretum Riverdale Rockville Silver Spring Suitland Takoma Park Towson Upper Marlboro U.S. Soldiers Home Washington W.B. City Wheaton Regional Park Woodstock
38 59 39 15 39 17 39 02 38 44 39 15 38 59 38 56 38 51 39 15 39 07 38 58 39 00 39 06 39 22 38 54 38 58 39 05 39 00 38 51 38 59 39 24 38 52 38 55 38 54 39 04 39 20
76 28 76 32 76 37 76 53 76 51 76 56 76 56 77 07 76 54 76 30 76 46 76 48 77 15 76 54 76 37 76 58 76 56 77 09 77 00 76 56 77 01 76 36 76 47 76 54 77 03 77 02 76 52
Urban Rural Urban Rural Rural Rural Suburban Urban Urban Urban Rural Rural Rural Suburban Suburban Urban Suburban Suburban Suburban Suburban Suburban Suburban Suburban Urban Urban Suburban Rural
T(i,j) = ~. (t(k)/D(k) k=l
(1/D(k)), Ik=l
where T(L j) is the interpolated temperature value of the ith by jth point, D(k) is the Squared distance between the ith j t h grid point and the kth nearest station, and t(k) is the temperature of the kth nearest station. For this study, the maximum value of k was set at 3. The next step in the analysis was to average the interpolated grids into decadal averages. This was accomplished for each month for each of the three decades of study. The data base which results is the mean interpolated temperature for January for the years 1950-1959, the mean interpolated temperature for January for the years 1960-1969, etc.. Each of the resultant grids (39 in all) were analyzed via trend surface analysis. This technique is a specialized application of the general linear model in which geographic coordinates (x and y) serve as the independent variables and, in this case, temperature is the dependent variable. As with the linear model a number
Changing Thermal Topography of the Baltimore- Washington Corridor." 1950-1979
93
of higher orders can be specified with each order explaining a certain percentage of the total variance. The surfaces of most value in the analysis are the first and second order (linear and quadratic, respectively) in that these are the approximations which explain the greatest variance for the interpolated temperature distributions. The linear surface defines a plane which slopes from a high temperature surface to a low temperature surface. The parameters of the linear model also define the gradient of the slope and the orientation of the resultant surface. The linear surface, while useful, is too simplistic for the purpose of defining the complexities of regional temperatures fields. The quadratic surface defines topographic complexities which most closely approximate the temperature field of an 'urban corridor'. As described by Johnston (1978), the quadratic surface is characterized by two high points with an intervening trough of lowered values or, as a 'saddle' shaped phenomenon. Intuitively, it is this type of surface which most closely approximates an urban heat corridor. Higher order surfaces (e.g. quartic, quintic) will refine the picture given by the first and second order surfaces (i.e. explain a greater percentage of the total variation) but will not be used here because they explain insignificant amounts of the variance. In addition, higher order surfaces are difficult to inteq9ret (Unwin, 1983). The form of the linear surface is defined as: X(ij) :- aoo + a lo Ui -~- aol Vj + eij ,
where X(ij) is the predicted temperature value of the ith by jth grid point, U(i) and V(j) are the X and Y coordinates of the ith by jth grid point, e is an error term and a(00), a(10), and a(01) are the coefficients of the linear model (Krumbein and Graybill, 1965). Similarly, the form of the quadratic surface is defined as: X(i,j) = a(00) + a(10) U(i,j) + a(01)V(i,j) + a(20)f~(i,j) 2 2 + a(ll) U(i,j)V(i,j ) + a(o2)V[i,j ) + e.
For both surfaces for each of the decade months analyzed, the predicted surface was derived. In addition, the percent reduction sums of squares for both types of surfaces was also determined.
3. Results
Tables III through V present the results of the trend surface analysis. Examination of the data reveals a number of clearly established trends. Foremost, is the significant strengthening of the percent variance explained by the quadratic surface through time. The most striking features to emerge are
94
Arthur Viterito TABLE III: Percent reduction sums of squares for linear and quadratic surfaces for the years 1950-1959. Month
% Var. Linear
% Var. Quadratic
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
29.6 31.5 37.6 37.3 30.2 13.6 10.5 11.1 17.3 13.4 15.4 23.3
26.6 29.9 23.7 25.3 31.1 37.0 37.1 40.8 33.8 32.0 27.6 30.3
Decade totals
20.7
32.9
TABLE IV: Percent reduction sums of squares for linear and quadratic surfaces for the years 1960-1969. Month
% Var. Linear
% Var. Quadratic
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
26.7 23.9 28.9 19.1 8.2 1.8 6.4 2.2 8.8 4.5 13.7 24.1
29.7 32.3 32.7 31.9 43.2 52.4 49.7 54.1 46.5 44.3 35.1 30.2
8.7
44.0
Decade totals
the characteristics o f the decadal averages and as can be seen, the percent variance explained by the quadratic surface increases from 32.9% to 44.0% from the 1 9 5 0 - 1 9 5 9 period to the 1960-1969 period. The explained variance of the quadratic surface further increases to 51.4% for the 1970-1979 period. The nonm o n o t o n i c change in the linear surfaces is inexplicable at this time and deserves further study. Figures 3 - 5 illustrate the quadratic surfaces for each o f the three periods and they show i m p o r t a n t characteristics o f the changing thermal topography. The surface for the 1950-1959 period can be described as containing high
95
Changing Thermal Topography of the Baltimore- Washington Corridor: 1950-1979 TABLE V: Percent reduction sums of squares for linear and quadratic surfaces for the years 1970-1979. Month
% Var. Linear
% Var. Quadratic
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
24.6 9.9 13.1 21.1 5.4 18.7 25.9 30.0 30.3 25.8 31.5 22.9
48.0 59.0 55.4 50.3 63.5 50.7 45.8 40.7 43.6 49.8 44.7 51.8
Decade totals
21.5
51.4
points centered on Washington, D.C. and Baltimore, MD with an intervening trough of lower temperatures. The decadal trend for the 1960-1969 period is similar to the 1950-1959 surface but with two important differences. First is the strengthening of the Baltimore heat island and the attendant increase in its areal extent. This is undoubtedly due to population increases in the Baltimore
0 I
Fig. 3. Quadratic surface of temperature for the period 1950-1959 (in deg. F).
5 mi I
96
Arthur Viterito
o I
5 mi I
Fig. 4. Quadratic surface of temperature for the period 1960-1969 (in deg. F).
o I
Fig. 5. Quadratic surface of temperature for the period 1970-1979 (in deg. F).
5 mi I
Changing Thermal Topography of the Baltimore- Washington Corridor: 1950-1979
97
metropolitan area during that time. The second difference can be seen in the relatively lower temperatures in the northwestern and southeastern sections of the corridor. It is to be kept in mind that the temporal comparison of trend surfaces illustrates changes in thermal topography. Therefore, the increase in the Baltimore heat island intensity is in relation to the rest of the temperature field and cannot be explained in absolute terms (e.g. a regional warming trend for the period in question). The decadal trend surface for the 1970-1979 period is markedly different from the two previous decades. The resultant surface, in addition to explaining a very large percentage of the variance (and thus representing a good approximation to the actual surface), displays strengthened heat island intensities in the Baltimore and Washington metropolitan areas. The Baltimore island also increased its areal extent from the preceding decades. The largest change though is the emergence of 'saddled' topography running along the axis of the corridor. That is, the intervening trough is in the process of being elevated. In a sense we can talk of inverted topography of the temperature field. This result is somewhat expected in that as population increases in the corridor the thermal characteristics of the intervening areas will begin to resemble those of the terminal conurbations. Figures 6 through 11 show the January and July quadratic surfaces for the 30-year period. As can be expected, the strengthening of the heat corridor is
o I
5 mi I
Fig. 6. Quadratic surface of temperature for January for the period 1950-1959 (in deg. F).
98
Arthur Viterito
0 1
5 mi I
Fig. 7. Quadratic surface of temperature for January for the period 1960-1969 (in deg. F).
0 I
5 mi I
Fig. 8. Quadratic surface of temperature for January for the period 1970-1979 (in deg. F).
99
Changing Thermal Topography of the Baltimore- Washington Corridor: 1950-1979
0 I
5 mi 1
~///////A////////////A Fig. 9. Quadratic surface of temperature for July for the period 1950-1959 (in deg. 17).
0 I
5 mi |
Fig. 10. Quadratic surface of temperature for July for the period 1960-1969 (in deg. F).
100
Arthur Viterito
0 I
,5 mi I
Fig. 11. Quadratic surfaceof temperature for July for the period 1970-1979 (in deg. F).
evident in both seasons and is best expressed in the July surface. This indicates that the thermal characteristics of the Baltimore-Washington heat corridor most closely resemble heat island behavior of low-latitude areas. That is, increased long-wave radiative flux is the dominant causal mechanism and rejected anthropogenic heat is of secondary importance. Landsberg (1981) suggests that higher wind speeds in the winter months will diminish or even eliminate urban temperature effects. Conversely, higher solar radiation receipt and lowered wind speeds will enhance the urban impact in the summer months. However, for all seasons the trend clearly conforms to the decadal trend. 4. Discussion
At a time when there is growing concern over the impacts of human activity on climate it appears that inadvertant climate change on regional scales has received little attention. Much of the current interest centers on large scale change due to such factors as increased anthropogenic loadings of carbon dioxide to the atmosphere and of a more drastic nature, the effects of nuclear conflict on climate. However, as the results of this study show, regional climate modification is occurring and at significant levels. Pankrath (1981) and Fortak (1981) have addressed this problem qualitatively and have prognosticated that the anthropogenic heat load to the atmosphere in the upper Rhine valley will
Changing Thermal Topography of the Baltimore- Washington Corridor."1950-1979
101
increase steadily through time. They further recommend that the impact on temperatures in the region should be monitored. The picture which should emerge from these types of studies is one in which regional temperature fields will continue to change in the direction of increased warming. As of now, assessments of regional modification to precipitation have yet to be made and should be a focus of future research efforts. In conjunction, studies of the changes in temperature related and temperature dependent phenomena need to be made. For example, widespread urban growth in the northeastern U.S. will impact growing season characteristics on a large scale and thus pose new constraints and/or benefits to agriculture in this region. Heating and cooling requirements are sure to change and thus regional energy consumption trends will be affected. Air pollution, snowfall characteristics, hydrology and even health related phenomena are certain to be affected over increasingly larger areas. A natural outgrowth of regional temperature modeling is the prospect of long term climate prediction. That is, population projections can be utilized as useful, albeit crude, predictors of future temperature regimes for urbanized regions. It would appear that the empirical relationships of temperature and population can be modified to accommodate regional scales. Also of great importance is the impact of regional temperature change on the representativeness of climate data. Longitudinal studies of climate change and dynamics need to consider this 'urban contamination' and adjustments must be made accordingly. Kukla et al. (1986) and Karl and Quayle (1987) address this issue but the results offered in this research indicate that the number of stations affected by urbanization is probably larger than previously suspected. In short, this all important feature of atmospheric geography deserves further study. References Bornstein, R.D.: i968, 'Observations of the Urban Heat Island Effect in New York City', J. of Applied Meteorology 7, 575-582. Fortak, H.G.: 1981, 'Local and Regional Impacts of Heat Emission', in W. Bach et al. (eds.), Interactions of Energy and Climate, D. Reidel, Dordrecht, Holland, 569 pp. Johnston, R. J.: 1978, Multivariate StatisticalMethods in Geography, Longman, 280 pp. Karl, T. R.: 1985, 'Perspective on Climate Change in North America During the Twentieth Century', Physical Geography 6, 207-229. Karl, T. R. and Quayle, R. G.: 1987, 'Climatic Change In Fact and In Theory: Are We Collecting the Facts?', The State Climatologist 11, 8-18. Krumbein, W.C. and Graybill, F.A.: 1965, An Introduction to Statistical Models in Geology, McGraw-Hill, 475 pp. Kukla, G. and Gavin, J.: 1986, 'Urban Warming', J. Climate andAppl. Met. 25, 1265-1270. Landsberg, H.E.: 1974, 'Inadvertent Atmospheric Modification Through Urbanization', in W.N. Hess (ed.), Climatic and WeatherModification, John Wiley, 612 pp. Landsberg, H.E.: 1975, 'Atmospheric Changes in a Growing Community', Institute of Fluid Dynamics and Applied Mathematics Technical Note, No. BN 823, University of Maryland, College Park, MD.
102
Arthur Viterito
Landsberg, H. E.: 1981, The Urban Climate, Academic Press, 235 pp. Mitchell, J.M. Jr.: 1961, 'The Thermal Climate of Cities', Symposium on Air over Cities, U.S. Public Health Service Publication A 62-5, 131-143. Moffitt, B.J.: 1972, 'The Effects of Urbanization on Mean Temperatures at Kew Observatory', Weather 27, 127-129. Oke, T. R.: 1973, 'City Size and the Urban Heat Island', Atmospheric Environment 7, 769-779. Pankrath, J.: 198 I, 'Impact of Waste Heat Emissions in the Upper-Rhine Region' in W. Bach et al. (eds.), Interactions of Energy and Climate, D. Reidel, Dordrecht, Holland, 569 pp. Unwin, D.: 1981, Introductory Spatial Analysis, Methuen, 212 pp. (Received 30 June, 1986; in revised form 1 August, 1988)