QUARTERLY JOURNAL

OF THE
ROYAL METEOROLOGICAL SOCIETY
Vol. 108 JANUARY 1982 No. 455

Quart. J. R. Met. Soc. (198Z), 108, pp. 1-24

551 S24551.588.7

The energetic basis of the urban heat island

By T. R. OKE
Department of Geography, The University of British Columbia, Vancouver
(Symons Memorial Lecture, delivered 20 May 1980)

1. INTRODUCTION

The development of scientific enquiry into a natural phenomenon may be expected

to follow a sequence such as the following:
0 0
Recognition and description of the -, Linkage (e.g. statistically) of the
phenomenon feature to other factors
0 @
-P Study of processes causing the -, Construction of process-response
phenomenon models to predict behaviour of the
phenomenon
In such a framework the field of urban meteorology may be judged to be at an early stage
and to be evolving in a rather unbalanced fashion. The literature of the past 150 years is
replete with studies of ’urban effects’ carried out at levels 1 and 2. Usually they are concerned
with simple description or statistical analysis based upon empirical evidence from a single
city. With the exception of a very few notable studies, attention to the processes (i.e. the
causes underlying the observed effects) and to physico-mathematical modelling has been
restricted to the past decade. Of course it is not expected, nor indeed may it be desirable,
that research in a field should progress in a simple manner through the sequence 1-4, but
two important points should be evident. First. as time progresses the bulk of research in a
field should move to higher levels of enquiry. Second, the predictive power of process-
response models is limited by the extent to which the processes are understood.
Some special difficulties have contributed to this unsatisfactory state of the field
including :
(1) the inherent complexity of the city-atmosphere system. The atmospheric state is a
response to exchanges of energy, mass and momentum covering a wide range of space
and time scales; in urban areas the sources and sinks for these exchanges are located
in an extremely heterogeneous fashion and involve significant anthropogenic as well
as natural factors;
(2) the lack of clear conceptual/theoretical frameworks for enquiry especially in the
light of the complications placed upon conventional theory by (1) ;
(3) the expense and difficulty of observation in cities. Commonly one must deal with
conditions within a relatively large volume of air (typically lo2 to lo3 km3) containing
significant spatial and temporal variability thereby creating sampling problems. More-
over there are restrictions on the use of observation systems (towers, aircraft, ball-
oons, acoustic radar) not normally encountered in uninhabited terrain.
Here we will use the example of the urban ‘heat island’ effect to illustrate the state of
urban meteorological research. This will include a condensed review of our understanding
1
2 T. R. OKE

of the qualitative features of heat islands and their relation to meteorological and urban
factors (levels 1 and 2); a summary of the results of recent research into urban energy and
mass exchanges (level 3); and a first attempt to relate these energetic causes to their thermal
effects (level 4), albeit under rather simple circumstances.

2. THENATURE OF URBAN HEAT ISLANDS

Luke Howard was the first to provide evidence that air temperatures are often higher
in a city than in its surrounding countryside (Howard, 1833). In the interim this fact has
been demonstrated beyond doubt and the characteristics of this urban heat island effect
have been documented for many villages, towns and cities (for a fairly complete listing see
Chandler 1970, 1976 and Oke 1974, 1979). Based on this body of information it is now
possible to draw some generalizations.
For the purposes of this paper such generalization will initially refer to the features
of the screen-level heat island in a large temperate-climate city (population 2 lOOOOO),
located on flat open terrain, during a fine (cloudless skies and calm or very light winds)
summer day. Later we will relax these constraints to show the vertical extent of the heat
islandandmention the effects of geographical location, city size, weather, season and climate.
This division conforms to the concept of two distinct layers (Oke 1976): one, called the
urban building or ‘canopy’ layer (UCL) extends from the ground up to about mean roof
level, rather like a vegetative canopy layer; the other, called the urban boundary layer
(UBL), is a mesoscale internal boundary layer whose characteristics are determined, at
least partially, by the presence of the city beneath. This conceptual division has been given
observational support recently (Taesler 198I).

(a) The near-surface heat island under ‘ideal’ conditions

Figure 1 depicts in idealized form the spatial and temporal features of urban and rural
screen-level air temperatures giving rise to an urban heat island under the conditions
specified above. The temperature cross-section and isotherm map in Figs. l(a) and I(b)
respectively illustrate a number of common features of heat island morphology. Assuming
the surrounding topography to be of minor significance the ‘island’ protrudes sharply
above the background rural temperature field. The ‘cliff’ to the ‘island’ is especially marked
on the windward urban/rural boundary and fairly closely follows the outline of the built-up
area for much of the city perimeter. Most of the rest of the urban area is characterized by a
much slacker horizontal temperature gradient but is interrupted by warm or cool spots
associated with localized areas of anomalously high or low building density. A park or
lake might be relatively cool whereas an industrial area, an apartment complex, a shopping
node or the central city area might be relatively warm. With a light wind these thermal
features may show evidence of being advected slightly downwind of their source areas.
Similarly, the warmth of thc city may be carried into rural areas to the leeward and the
‘cliff’ on that side may be less distinct.
Time-dependent aspects of the phenomenon are illustrated in Figs. I(c) to (e) where
‘rural’ air temperatures represent the background values of the countryside around the
city and ‘urban’ the highest sustained (usually city centre) values of the urban area. In a
causal sense urban/rural energy balance and stability differences produce different rates of
near-surface warming and cooling (Fig. l(d)), giving rise to distinctive diurnal air temperature
rkgimes (Fig. I(c)), whose differences at given times define the urban heat island intensity,
AT,,-, Fig. l(e).
The rural sequence is a standard one. At about sunset there is relatively rapid cooling
(typically 2 to 3 ”C h-’) as the surface experiences a net radiative energy drain and with-
draws heat from the shallow layer of stable air immediately above it. As the surface tempera-
ture drops and the rate of radiative emission decreases, the cooling rate also declines (to
about 0-5“Ch-’) as the night progresses. As a result rural air temperatures at night exhibit
 I-
s
W
n

DISTANCE 5I-
cra

U
f

I-
$
I

5;
r
Built-up area

Figure 1. Hypothetical representation of the spatial and temporal

features of the canopy layer urban heat island in a mid-latitude city 12 18 24 06 12
with 'ideal' (calm, clear) weather. Spatial pattern (a) along cross-
section AB and (b) in relation to the plan outline of the city. Temporal
variation of urban and rural (c) air temperature (d) heating/cooling TIME (h) W
rates and (e) the resulting heat island intensity. Vertical scale units are
approximately 2°C for (a), (c) and (e) and 2 "C h-I for (d).
4 T.R. OKE

a simple exponential decay curve until just after sunrise, when the pattern is abruptly
interrupted by warming. Solar heating of the surface generates a turbulent sensible heat
flux which converges in the surface layer, whose depth is limited by the remnant nocturnal
radiative inversion above. As the mixed layer grows the rate of warming declines into mid-
afternoon at which time the maximum temperature occurs.
Without at this point invoking any physical explanations it is clear from Figs. l(c)
and (d) that the urban rCgime is very different. Warming and cooling rates are generally
smaller (except in the latter half of the nocturnal period) and noticeably lack the sharp
peaks around sunrise and sunset thereby producing a damped diurnal temperature wave.
As a consequence of the above the heat island intensity undergoes a marked diurnal
variation (Fig. l(e)). Diverging rates of cooling between the urban and rural environments
around sunset produce a sharp increase in intensity to a maximum a few hours (typically
3 to 5 h) later. Thereafter slightly greater urban cooling reduces the intensity until the early
daytime rural heating virtually erases the heat island. In some cities there are even reports
of slightly lower temperatures in the central area in comparison with the countryside
(‘cool’ islands). Based on this account the spatial distribution of the heat island given in
Figs. l(a) and (b) obviously relates to the time of maximum intensity. The spatial pattern
by day tends to be less well defined.
In summary, the canopy heat island is largely a nocturnal phenomenon attributable
to urban/rural cooling (rather than heating) differences especially in the period around and
following sunset. Its spatial pattern conforms closely to the distribution of surface cover
characteristics. The very general features emphasized in this discussion are well illustrated
by studies in individual cities: for examples of spatial features see Sundborg (1951);
Duckworth and Sandberg (1954); Takahashi (1959); Chandler (1965; 1967) and Oke
and East (1971); for temporal features see Chandler (1965); Hage (1975); and Oke and
Maxwell (1975).

(b) Vertical structure of the heat island

The influence of the canopy heat island extends into the overlying atmosphere. If the
‘ideal’ constraints used in Section (a) apply, the thermal modification takes the form of a
self-contained urban heat ‘dome’. In reality, winds are unlikely to remain totally calm
throughout the lower atmosphere for very long and hence the more normal form of the
thermally-modified layer is that of an internal boundary layer, entrained in the direction
of the wind (Fig. 2(a) and (b)). By day (Fig. 2(a)) the UBL increases in depth with time in
accord with the growth of the rest of the mixed layer in the lower troposphere (e.g. Smith
and Hunt 1978). Thus by mid-afternoon the UBL may be 0.5-1.5 km in depth, depending
upon the strength of the surface sensible heat flux and the stability of the air mass. The
presence of the city and its UBL may be evident by a slight doming of the normal mixed
layer (by up to approximately 0.25 km, Fig. 2(a)) (Spangler and Dirks 1974). Downwind of
the city the development of a new rural internal boundary layer leaves the urban-affected
air aloft, as a heat ‘plume’, often extending for tens of kilometres.
At night, because of the usually well-developed heat island, the city retains a surface
mixed layer whilst the surrounding rural areas have none, due to the presence of a surface-
based radiative inversion (Fig. 2(b) and(c)). The stability of the rural air is eroded to a depth
of 100-300m (Fig. 2(c)) as it advects across the warm, rough city and again an elevated
heated ‘plume’ is found in the lee (Clarke 1969; Oke and East 1971). A plot of the urban
heat is!and intensity with height for the city centre (Fig. 2(d)) thus shows the heat island
intensity to decline rapidly with height from its value at screen-level often becoming negative
(rural warmer than urban) near the top of the UBL in the so-called ‘cross-over effect’
(Duckworth and Sandberg 1954; Bornstein 1968).

( c ) Other heat island controls

Relaxing the constraints originally imposed on our discussion of the heat island we
 THE ENERGETIC BASIS OF THE URBAN HEAT ISLAND 5

0.4 I I I 1

0.3

0.2

‘Heat Island’
0.1

1
0.0 A

Figure 2. Generalized form of the UBL thermal structure in a large mid-latitude city during fine summer
weather (a) by day, including schematic profiles of potential temperature (0) and the depths of the urban
and rural internal boundary layers (- -) and the daytime mixed layer (-.-) and (b) at night. Comparison of
(c) rural and urban vertical temperature profiles and (d) the resulting vertical profile of heat island intensity
in the city centre at night.

can note briefly the role of other controls. For example the geographical location of a city
imposes a wide range of possible effects especially related to the presence of water bodies,
topographic features, and the nature of the soils, vegetation and land use in the region. The
combinations possible are almost as great as the number of cities and to some extent this
explains the preoccupation with descriptive heat island studies described earlier.
Possibly more amenable to generalization than that for location is the relationship
between heat islacd intensity and city size. Such relationships have been shown to exist
(e.g. Fig. 3), but current interest lies in an elucidation of the critical physical factors implicit
in measures of ‘size’ such as population. Is the appropriate measure one of city diameter,
which might be seen to be related to distance of fetch and therefore the accumulation of
heat, or is it one of architecture, building materials or anthropogenic heat which might also
be correlated with size? Similarly we wish to know why the European and North American
P vs. dTu-r(maz) relationships are different.
Weather controls on the heat island have been studied in some detail. Statistical studies
6 T.R.OKE

12 -

10 -

8 -

6 -

4 -

2r

0 : 1o3 1o4
105 1 06 10'

POPU LAT I ON

Figure 3. Relation between maximum heat island intensity LIT^-^(,,,^^)) and population (P)for European
and North American settlements. The crosses are for the growing new town of Columbia, Maryland reported
by Landsberg (1979). Figure modified after Oke (1973).

agree that the most significant meteorological variable governing heat island intensity is
the wind speed and next is cloud cover (Sundborg 1951 ; Duckworth and Sandberg 1954;
Chandler 1965). Closer study reveals that the wind speed relationship is non-linear (approx-
imately inverse square root, Oke 1973) and that cloud type, as well as amount, is important
since low cloud is more effective than an equal quantity of high cloud in limiting the heat
island intensity. Wind and cloud are clearly surrogate variables related to the relative roles
of turbulent and radiative transfer in producing temperature change. These roles are com-
bined to some extent in the single variable of atmospheric stability, which has also been
shown to bear a good correlation to heat island intensity (Ludwig 1970; Lee 1975).
In temperate latitude cities there is often a seasonal variation of the heat island with
the greatest frequency of occurrence and greatest intensities being registered in the warmer
half of the year, especially summer and autumn (Chandler 1965; Lee 1979; Unwin 1980).
This is tied in part to the seasonality of the weather controls (wind, cloud and air mass
stability) just discussed, but it probably also reflects variations in surface cover, such as
snow and vegetation, and in solar influences, such as angle of incidence in relation to urban
canyon geometry and attenuation by aerosols. The fact that the heat island is best displayed
in summer, whereas peak heating requirements are in winter, may indicate that anthropog-
enic heat is not a primary cause.
Whilst there is a relative abundance of research on the nature of heat islands in
temperate climates, there is a dearth regarding those of equatorial, tropical, sub-polar
and polar settlements. In broad outline the results available from low latitude cities bear a
resemblance to those discussed previously (Tyson et al. 1972; Daniel and Krishnamurthy
1973; Jauregui 1973), but at high latitudes the picture is altered by the seasonally extreme
radiation climate. In winter there is little or no solar radiation input so the normal cycle of
energy receipt is absent. It is replaced by a daily cycle of anthropogenic heat releases,
related to human activities (especially space heating demand), to which the heat island is
tied directly (Nicol 1976; Bowling and Benson 1978). The converse situation, involving
continuous daylight in summer, has not been studied in detail but we may anticipate very
little diurnal variation in the heat island. Observations in the spring and autumn transition
periods show the heat island features to approximate the temperate case.
 THE ENERGETIC BASIS OF THE URBAN HEAT ISLAND 7

In summary, the urban heat island is a thermal anomaly having both horizontal,
vertical and temporal dimensions, which has been observed in virtually all settlements,
large and small, where it has been sought. In the mid-latitudes, where it has been most
studied, its characteristics are found to be related both to the intrinsic nature of the city
(e.g. its size, building density, land-use distribution) and to external influences (e.g. the
climate, prevailing weather and seasons).

3. URBAN/RURAL
ENERGY BALANCES

The urban heat island effect clearly must be the result of urban/rural energy balance
differences. This has been realized for many years yet we still await the first observational
study to demonstrate this fact and to pinpoint the critical term(s). Here we suggest that
if progress is to be made it is important not to confuse the two principal scales of urban
meteorological enquiry, viz: the micro-scale characterizing exchanges and climates in the
UCL, and the meso-scale of the UBL. However neither layer should be considered in
isolation, since conditions in the UCL are often modulated by those in the UBL which, in
turn, represents both an areal integration of UCL effects and its own intrinsic qualities.
We will deal first with the diurnal energy balance of a typical rural surface in the summer.
This will be brief since it is well known, but is important, because it forms the basis of
comparison for the surface energy balances appropriate to the urban canopy and boundary
layers which follow. Discussion is limited to the case of a fine summer day in a temperate
climate.

(a) Rural energy balance

The rural energy balance illustrated by Fig. 4 is well understood (e.g. Sellers 1965;
Oke 1978a) as is its relationship to planetary boundary layer characteristics (e.g. Smith
and Hunt 1978). The heat energetics of the system are driven by the surface net radiant
flux density (Q*) which is dominated by short-wave radiation exchange by day and solely
due to long-wave radiation at night. The surface radiant energy surplus (deficit) is dissipated
(supplied) by heat conduction to (from) the underlying soil ( Q c ) and by convection of
sensible and latent heat to (from) the air (QHand Q E respectively). The actual sharing of
heat between the soil and the air depends upon many factors including the nature of the
surface, the thermal properties of the soil and the state of the atmosphere (especially the
level of turbulence). A surface covered by short vegetation such as that in Fig. 4 apportions
about 80-90% of the daytime radiant surplus to the air but, at night, the radiative deficit is
balanced largely by conduction of heat from soil storage with only lO-SO% drawn from
the atmosphere due to the relatively suppressed state of turbulent activity.
The partitioning of the turbulent transport between the sensible and latent forms
(Bowen’s ratio, p) largely depends upon surface moisture availability. When the surface
is wet, evapotranspiration is at the potential rate and is dependent only on energy availability
and temperature (Priestley and Taylor 1972). When the surface is moist, rather than wet,
the rate drops to about 80% of the potential value, a condition known as equilibrium eva-
poration where typical rural values of fl (based on daily energy totals) are in the range
04-08. In the case of the irrigated crop in Fig. 4 Q E is close to potential and fl is about 0.3.
Following a period of drying it is quite possible to find rural fl values of 1-5 or greater in
the mid-latitudes (e.g. Bailey 1977).

(b) Urban canopy layer energy balances

The myriad of surfaces present in the urban canopy layer make it impossible to draw
generalizations. Every surface facet has a unique combination of both its intrinsic properties :
radiative, thermal, moisture and aerodynamic, and those contributed by the rest of the
surrounding site environment: radiation geometry, position in flow field, adjacency to
8 T.R. OKE

600
-
N
Sept. 26, 1957
'E
-t3 400
t
v)
Z
W
0
X 200
3
A
LL
t
U
K 0
W
z
w

-200 ' I I 1 I I 1 I , I 1

2- I 1 I I 1 1 1 I 1

4
0 1 -
I-
:
z
0
w
3 -1 -
I I 1 I
-2

LOCAL SOLAR TIME

Figure 4. The diurnal variation of the energy balance and the Bowen ratio (8) of an irrigated field of
grass at Hancock, Wisconsin (data from Tanner and Pelton 1960).

contrasting surfaces likely to contribute to advective interaction, etc. This produces an

almost limitless array of energy balances and therefore microclimates. Despite this apparent
chaos, there are studies concerning a few of the more dominant surface units including
lawn (Oke 1978b) courtyard (Noilhan 1980) street canyon (Nunez and Oke 1977, 1980)
parking lot (Landsberg and Maisel 1972) and road (Terjung et al. 1971) environments.
Any attempt to produce a climatic classification of these surface units would have to
recognize moisture status (water storage capacity and surface moisture availability) to
be the primary criterion. Storage capacity is important in that water retention provides a
buffering or moderating influence, because gradual moisture depletion allows for evapora-
tive cooling to extend over a period of days or more. On the other hand, impervious elements
of the urban landscape experience an almost dichotomous wet/dry behaviour, wherein
the climatic effects of rain may be present for only a few hours before run-off and evaporation
dry the surface.
A surprisingly large proportion of the surface area of European and North American
cities is occupied by vegetation (40-70%), especially in their suburbs. These surfaces are
likely to have water storage capacities equal to those of rural areas and many of them are
irrigated in summer (depending on climate, abundance of water supply and local custom),
making water easily available to the atmosphere. Thus irrigated lawns, gardens, parks,
golf courses, and cemeteries lie at one end of the wide spectrum of urban surface types.
 THE ENERGETIC BASIS OF THE URBAN HEAT ISLAND 9

I I I I I I 1 I I I I
1
Aug. 6,
400
p"g""
200

-200 I I 1 I I I 1 I 1 1 I J
0 04 08 12 16 20 24

LOCAL SOLAR TIME

Figure 5. Energy balance of an irrigated suburban lawn (160 m2)in Vancouver, B.C. (after Oke, 1978b)

An example of an energy balance for just such a surface (an irrigated suburban lawn)
is given in Fig. 5 . The pattern is somewhat similar to that of the moist rural surface (Fig. 4)
but there are significant differences. In particular Q E can be seen to assume an even more
dominant role in daytime energy use. Evapotranspirative losses exceed even the potential
rate by about 30% on a daily basis. Probably this is made possible by the advection of
warmer, drier air from surrounding impervious surfaces. They act as additional sources
of sensible heat capable of forcing the lawn evaporation at a rate in excess of that support-
able by (Q*- QG).Such.an 'oasis'-type condition can persist only if the natural water
budget is supplemented by irrigation.
As long as un-irrigated urban vegetation has available moisture it will also be in receipt
of advective energy. Following rain such surfaces may be expected to behave like the lawn
already discussed but their drying cycle is likely to be accelerated in comparison with
non-advective rural vegetation.
At the other end of the spectrum of urban surface types are impervious areas: paved
roads, parking lots and buildings. These systems are characterized by almost exclusively
partitioning their radiant energy into sensible heat. They are the sensible heat source regions
for the microscale advective interaction with moist areas. One very common arrangement
of these relatively dry surfaces is the street canyon formed by a roadway between two
adjacent buildings. This unit consisting of three active surfaces (the walls and floor) intro-
duces the influence of geometry. Geometry is significant in a number of ways: for example,
it increases the surface area exposed to exchange processes; it controls and complicates the
spatial distribution of direct-beam solar radiation and precipitation inputs ; it gives rise to
radiative interaction between surfaces; it influences the loss of long-wave radiation to the
sky and it controls the mean and turbulent flow structure. As a result, even for a single
canyon constructed of materials possessing relatively uniform radiative, thermal, moisture
and aerodynamic properties, the radiation and energy balance rkgimes of the component
surfaces exhibit a great deal of spatial and temporal complexity. This is amply demonstrated
by the results from an urban canyon given in Fig. 6. The canyon is oriented with its long-
itudinal axis in a north-south direction, has dry walls but a slightly moist floor and a height/
width ratio of 0.86. Measurements were taken during fine weather with cloudless skies and
light winds. In terms of geometric scaling and moisture availability, the canyon is probably
a fair analogue of a street in the centre of a moderate-size city with a few scattered moisture
10 T.R. OKE

ZOO

-
p:
-200
E 0 04 08 12 16 20 24
I I

X
3
A 200
LL
>
(3
a
W 0
z
W
W

2
LL -200 I I

cr2 0 04 08 12 16 20 24
tn I I , 1 I

-200 '
0 04 08 12 16 20
I
24

TIME (PST)
Figure 6. Energy balances for each surface of a north-south oriented canyon in Vancouver, B.C. (a) West
(east-facing) wall, (b) floor and (c) east wall of the canyon. Data are averages for a 3-day period in September
1973 (after Nunez and Oke 1977).

sources (e.g. trees). The spatially-averaged energy balances for the three canyon surfaces
in Fig. 6 show their primary peaks to occur at quite different times of day, due to differing
times of maximum solar irradiance. The walls also show secotidary peaks associated with
reflection from the other wall, but overall the floor is the most active energy exchange
surface, Note that, in absolute terms, the magnitudes of the energy flux densities are rather
small (e.g. by comparison with the rural values). This is especially the case at night. By
day this is primarily dictated by the lack of illumination of one or more surfaces due to
horizon obstruction and the rather unfavourable angle of solar incidence on east and west
 THE ENERGETIC BASIS OF THE URBAN HEAT ISLAND 11

-90 I I I I I I I I I

8
-80 - 8
-
o West wall -
-70 - 0 East wall
o Canyon floor 8
-60 - 8 Canyon top -
-
N
-
'E -50 -
-
3
-40 - 0
8
-
*
-I
0
-
-30 - D g q p
00
-20 - 0 -

-10 - -

I I I I I I I I I
0'
0 0.2 0.4 0.6 0.8 1 .o

Figure 7. Relationship between net long-wave radiation (L*) and sky view factor (!,ha) for different positions
around the perimeter of a canyon cross-section. Average results for two nights with cloudless skies and light
winds in September 1973. Values of !,hs < 1 at the canyon top are due to the walls being of different height
(data from Nunez 1974).

walls. At night the low rates of radiative emission are largely a consequence of the reduced
sky view factor ($,) inside the canyon. (The sky view factor of a point is proportional to
the area of the overlying hemisphere which is open to the sky). The relationship between
$, and net long-wave radiation (L*) for different locations in the same canyon is given in
Fig. 7.
When one considers that the complete UCL morphology is composed of such relatively
simple canyons, having complex internal workings, as well as many others having totally
different orientations to the Sun, wind and rain, different height/width geometries and
different material properties, the situation becomes rather bewildering. On the other hand,
some solace is gained from the evidence provided by the available canyon observations
(Nunez and Oke 1977, 1980) and models (Terjung and Louie 1974; Terjung and O'Rourke
1980) which show that total canyon systems experience a relatively smooth diurnal energy
rkgime. For example, the results in Fig. 8 show the integrated energy fluxes calculated to
pass through the top of the canyon considered in Figs. 6 and 7. For each heat flux (Q)
the conversion of fluxes from walls and floors to an equivalent flux through the top of a
symmetric canyon is accomplished with
Q t = 2 Q d H / W ) + Qr
where H / W is the canyon heightlwidth ratio and the subscripts t , w and f refer to the top,
walls and floor respectively. In most respects the daily pattern of energy sharing shown in
Fig. 8 is similar to that of a flat surface with a severely restricted water supply. One minor
discrepancy is the rather short day length brought about by horizon obstruction. The
daytime peak net radiation is very similar in magnitude, and the nocturnal values are only
12 T.R. OKE

600 I I I I I 1 1

-200
0 04 08 12 16 20 24

LOCAL SOLAR TIME

Figure 8. Energy balance of a complete urban canyon system. Exchanges are expressed as equivalent
flux densities passing through the canyon top using mean hourly data for a 3-day period in September 1973
(after Nunez and Oke 1977).

slightly less negative, than those for typical rural surfaces (e.g. Fig. 4). Energy partitioning
is predominantly into sensible heat. In the canyon studied, approximately 60% of the day-
time radiant surplus is removed by the convective sensible heat flux and a further 30:d is
conducted into storage ( A Q,) by the canyon system. This gives a canyon value of approx-
imately 6 by day. At night, when turbulent activity is weak, almost all (approximately
90%) of the net radiative (long-wave) drain is drawn from storage.
Presumably, when dry, paved surfaces, such as roads and parking lots, dissipate radiant
surpluses and make good radiant deficits purely by sensible heat transfer to and from the
air and ground. When wet, they may evaporate water at at least the potential rate, but
this is usually short-lived because of efficient run-off to sewers. Therefore, as mentioned
earlier, the energy rCgime is probably of a rather 'flashy' nature but it should be recognized
that there is the possibility that water storage in such materials has been underestimated
(Lacy 1977).
The UCL energy balances considered have not explicitly included anthropogenic
heat releases due to space heating, industrial operations or automobile use. If such heat
releases enter the UCL (e.g. walls, windows, doors, vehicle exhausts, etc.) they may form
an important, even dominant, component of the local energy balance but in general they
are small (see the following section).
In summary, the UCL consists of a wide range of energy balance systems. Moisture
availability is one of the key variables controlling local partitioning of daytime radiant
energy. The range even includes the possibility that some surfaces may show no evaporation
at a time when others are losing water at a rate in excess of potential. Site geometry is
another crucial parameter because of its importance in determining the receipt and loss of
radiation. A third must be the thermal properties of the surface and building materials,
but with the exception of Kawamura's pioneering study (Kawamura 1964) little information
is available. Locally anthropogenic heat may also become important.
 THE ENERGETIC BASIS OF THE URBAN HEAT ISLAND 13

Short- wave Long-wave

”
A tmos. A tmos.
cooling
heating
14 T. R. OKE

lent fluxes in the rural case. The amount received at the surface (direct-beam and diffuse,
Flux 2 plus net back-scattered, Flux 4) is typically 2-10% lower in the city (Peterson and
Stoffel 1980). On the other hand urban albedo values are typically 0.05 to 0.10 lower than
for the countryside in the mid-latitudes (Oke 1974). As a result urban/rural net short-wave
radiation differences are considered to be rather small. The sign of any differences will
depend upon the relative strengths of the pollution attenuation and albedo factors. A
similar off-setting of effects occurs in the long-wave radiation budget where exchanges are
not only perturbed by the increased pollutants and the probably lower surface emissivities
of cities but also by a feedback with the heat island warmth. The higher surface temperature
of the city appears to outweigh the emissivity change thereby producing an enhanced
emission (Flux 5). However, a relatively large part of this is absorbed by the polluted layer
and re-radiated back to the surface along with that portion of the incoming sky radiation
(Flux 6) transmitted to the surface (Flux 7), and that emitted by the warm heat island air
(Flux 8). At night these combined long-wave radiative inputs are slightly larger in the city
(Oke and Fuggle 1972), and by day the excess may be greater still, due to emission from
solar heated pollutants (Rouse et al. 1973). In summary both long-wave input and output
are increased by urbanization so that urbanlrural net long-wave differences are not large.
From the above it follows that urban/rural net all-wave radiation (Q*) differences are small,
probably less than 5% (Oke 1974; 1979; Probald 1975; White et a/. 1978).
Our knowledge of the radiative processes within the polluted layer is far from complete.
The mathematical treatment of scattering and absorption by pollutants is reasonably
well developed (Atwater 1975; Coakley and Chflek 1975), but there are few studies of
urban atmospheres which combine information on both the physical properties of the
aerosols and their radiative effects. The profile of atmospheric warming produced by
shortwave absorption given in Fig. 9 is probably a reasonable consensus based on both
observations and model calculations. The corresponding profile for the long-wave region
is less certain. Most models (e.g. Atwater 1971; Bergstrom and Viskanta 1973) and nocturnal
observations (Fuggle and Oke 1976) show cooling to predominate but some daytime
observations suggest warming (Berlyand et al. 1974). The net radiative impact of urban
pollutants probably is to warm and therefore increase stability in the UBL by day, but at
night they act to cool and destabilize the layer.
If we assume, as seems reasonable for most mid-latitude cities in summer, that urban/
rural radiative and anthropogenic heat flux differences are relatively small at the surface,
then most of any energy balance differences must stem from a different sharing of heat
between the turbulent and storage and/or between the sensible and latent terms. Here we
will consider these possibilities using data from studies in suburban and urban terrain.
The former may be considered to consist mainly of 1-2 storey houses with gardens in the
fairly uniform, but dispersed, pattern often characterizing the dominant land-use of modern
cities. The latter is a more dense agglomeration of taller (2-4 storey) buildings including
residential, commercial and light industrial uses with less vegetation.
Simple consideration of the impact of urbanization on the water budget suggests that
the city is subjected to partial waterproofing which leads to decreases in both subsurface
water storage and evapotranspiration (Mather 1978; Oke 1978a). The available data
clearly support this view but show that the city is by no means to be considered a ‘desert’.
Consider the example suburban energy balance shown in Fig. 10. These results are
judged to be fairly typical of summer conditions in Vancouver (Kalanda et al. 1980;
Oke 1978~;Yap and Oke 1974) and probably a number of other mid-latitude cities including
Adelaide, Australia (Coppin 1979) Los Angeles and St. Louis, USA (Carlson et al. 1981;
Ching et al. 1978) and Uppsala, Sweden (Oke 1978~).Note that the sensible and latent
heat flux densities are of similar magnitude during the daytime so that /3 is about 1.0.
The rate of evapotranspiration is approximately 80% of the equilibrium value and therefore
far from insignificant. Such relatively high rates (considering that about 34% of the surface
is impervious) probably are due to ‘micro-oases’, such as the lawn in Fig. 5, whose exception-
 THE ENERGETIC BASIS OF THE URBAN HEAT ISLAND 15

__-
6oo
500
tt- July 29, 1978 11
400

300

200

100

-Inr
00 04 oa 12 16 20 24

LOCAL SOLAR TIME

Figure 10. Energy balance of a suburban area in Vancouver, B.C.

TABLE 1. AVERAGE DAYTIME PARTITIONING OF HEAT BY BY RURAL, SUBURBAN AND URBAN LANDSCAPES.'
CONSENSUSOF OBSERVED VALUES OF BOWEN'S RATIO (p), ESTIMATES OF AQs/Q* AND RESULTING TYPICAL
VALUES OF THE RATIOS QH/Q* AND QE/Q*.

Land-use
s
Observed rangel Typical
AQs/Q*
Estimated range3 Typical
QdQ* &/Q*
Typical Typical
Rural 0.1-1.5 0.5 0.05-0.25 0.15 0.28 0.57
Suburban 0.25-2.5 1 .o 0.20-0.25 0.22 0.39 0.39
Urban 0'5-(>4*0) 1-5 0.25-0.30 0.27 0.44 0.29

Temperate climate, summer, non-advective conditions.

Sources: Bailey (1977), Carlson et al. (1981), Ching et al. (1978), Coppin (1979), Kalanda et al. (1980),
Oke (1978c), Sellers (1965) and Yap and Oke (1974).
Sources: Oke et al. (1981) and Sellers (1965).

ally high rates compensate for the removal of other transpiring surfaces. The range of
values for suburban terrain under summer conditions is however reasonably large, extend-
ing from about 0.25 following rain to greater than 2.5 at the end of a dry spell (Table 1).
Data from urban sites are more sparse (Carlson et al. 1981; Ching et al. 1978; Yap
and Oke 1974) but are consistent in that they show a further decrease in the role of latent
heat compared with rural terrain. Nevertheless QE is still far from negligible (Table 1).
Within the daily pattern two interesting features have been noted. First, the mid- to late
afternoon decrease of Qrroften tends to lag behind that of Q*; and second, QHmay remain
positive (directed into the atmosphere) at night (Yap and Oke 1974). These features can
result in p being relatively large in the late afternoon and positive at night. The timing of
these features corresponds to that of most rapid heat island growth (Section 2(a)) but any
direct linkage has yet to be shown.
Therefore as summarized in Table 1, suburban and urban surfaces are characterized
by rather variable energy partitioning. This is largely controlled by moisture availability.
When both the city and countryside are wet, differences will be small; in drier conditions
the city tends to become a relative source region for sensible heat, although urban irrigation
can mitigate (perhaps even reverse) this tendency.
It is usually assumed that heat storage (AQ,) in the city is significantly larger than in
16 T. R.OKE

its surrounding areas because of the greater thermal conductivity (k) and heat capacity (C)
of some building materials. The appropriate parameter which combines these properties
is the thermal admittance, ,u (viz. (kc)%),sometimes called the thermal inertia. It isa measure
of the thermal responsiveness of a surface for a given heat flux. Ceteris paribus surfaces
of large p, such as cities, should accept (release) heat to (from) storage with relative ease,
and register relatively small surface temperature changes. Especially at night, when storage
assumes a more important role in the total energy balance of both urban and rural environ-
ments, this factor may be important in maintaining warmer urban temperatures. The day-
time situation is more complex (Bryson and Ross 1972).
Confirmation of this increased heat storage is hard to obtain because direct measure-
ment of AQ, at the city scale is not possible at present. Probably the most practical, but
over-simplified, approach is to parametrize AQ, in terms of Q* (e.g. Oke et al. 1981). The
method relies on a series of empirical equations relating these two fluxes for a variety of
urban materials and surface units (including canyons). These are weighted according to
the fraction (ai) of the urban landscape covered by each of the i surface types:
n

~ Q =s C
i=l
ui(aiQ* + bi)
where a, and b i are statistically determined coefficients. Values derived in this manner are
included in Fig. 10 and Table 1 where it is seen that cities do appear to be better heat stores
than their environs.
The cause of such heat storage differences is not yet clear because urbanlrural thermal
admittances do not appear to show the contrasts expected (Carlson et al. 1981; Oke 1981).
The apparent discrepancy may be accounted for by considerins, in more detail, the roles of
such influencesas geometry (increased area for exchange in the city) ;and moisture availability
(greater daytime heating in the drier city).
In summary, our knowledge of meso-scale urbanlrural energy balance differences,
although less precise than is to be desired, tends to confirm intuition. The impact of urban-
ization is to favour partitioning of energy into sensible rather than latent heat and to increase
the importance of heat storage by the system.

4. URBAN
HEAT ISLAND GENESIS

(a) Urban canopy layer

Starting with the first study of a heat island by Howard, urban meteorologists have
continued to forward their favourite set of causative factors. Table 2 presents a list which
includes most of those suggested (it may be of interest to note that Howard proposed no
less than five of the seven). It should be realized that, until the past decade, the basis of
such lists rested almost entirely upon intuition and reasoning from indirect evidence. Now
it is fair to say that each of the seven mechanisms has been shown to be capable of operating
in the direction required (Section 3), but still we are unable to assign a definite rank ordering
of their contribution to heat island development except in a few special cases.
One extreme case is represented by an arctic settlement in mid-winter (i.e. during polar
night, Bowling and Benson 1978).The energy balance is greatly simplified due to the absence
of solar radiation, evapotranspiration and heat conduction between the buildings and the
ground. The balance i n calm conditions essentially reduces to that of a net long-wave
drain upon an anthropogenic heat source. Complications are introduced by increased wind
speeds (giving convective exchange) and if the air is polluted (especially with water in the
form of ice fog). Nevertheless in comparison with most other urban climates, the energetic
controls involved in this special case are defined fairly well.
The best opportunity to understand the causation of mid-latitude heat islands relates
 THE ENERGETIC BASIS OF THE URBAN HEAT ISLAND 17

TABLE 2. SUGGESTED
‘CAUSES’ OF THE URBAN HEAT ISLAND (Not rank ordered)

Altered energy balance terms leading to positive Features of urbanization underlying energy
thermal anomaly balance changes
A . Canopy layer
1. Increased absorption of short-wave radiation Canyon geometry - increased surface area and
multiple reflection
2. Increased long-wave radiation from the sky Air pollution - greater absorption and re-emission
3. Decreased long-wave radiation loss Canyon geometry - reduction of sky view factor
4. Anthropogenic heat source Building and traffic heat losses
5. Increased sensible heat storage Construction materials - increased thermal
admittance
6. Decreased evapotranspiration Construction materials - increased ‘water-
proofing’
7. Decreased total turbulent heat transport Canyon geometry - reduction of wind speed

B. Boundary layer
1. Increased absorption of short-wave radiation Air pollution - increased aerosol absorption
2. Anthropogenicheat source Chimney and stack heat losses
3. Increased sensible heat input-entrainment Canopy heat island - increased heat flux from
from below canopy layer and roofs
4. Increased sensible heat input-entrainment Heat island, roughness - increased turbulent
from above entrainment

to the ‘ideal’ case illustrated by Fig. 1. This is because it is a fortunate coincidence that the
phenomenon is best displayed (has its greatest magnitude) under conditions when the
energetic controls are simplest. Our task is to explain a phenomenon which finds its best
expression in large mid-latitude cities on calm, clear summer nights and whose morphology
shows a marked conformity with that of the city’s structure. We will therefore concentrate
initially on consideration of differences of urban/rural surface energy exchange at night
which may lead to the observed differences in cooling rates. The following simple conceptual
framework is offered as a starting point.
On the basis of observation, (as summarized in Section 3) or as a result of the con-
strained conditions, we may effectively eliminate ‘causes’ A l , A2, A4, A6 and A7 in Table 2
from further consideration. Thus urban/rural energy balance and cooling rate differences
will depend simply upon the relative strengths of the net long-wave radiative sink and the
sub-surface heat store in each environment (i.e. ‘causes’ A3 and A5). Ceteris paribus the
relevant and therefore critical properties governing these differences are the radiation geo-
metry (I/Js) and the surface thermal properties (given by the thermal admittance, p). Some
indication of the importance of differences is given by comparing L* values on the canyon
floor in Fig. 7 with those for locations near the canyon top where I/Js tends to 1 *O(the latter
being of similar magnitude to those typical of open rural sites e.g. Fig. 4). Recent work
using scale models, where I), is varied while holding p constant, shows that the influence
of radiation geometry is capable of simulating the temporal features of the ‘ideal’ nocturnal
heat island (Fig. l(c) to (e)) well (Oke 1981). The same study also shows that in the absence
of $, differences an increase of p is also able to replicate these features. Thus in the real
world both effects are probably in operation, their relative roles depending on the size of
urban/rural $, and p differences. Whereas the former are easily demonstrated it is difficult
to obtain truly representative values of the latter, which depend on the multitude ofmaterials
involved and are complicated by vegetation, snow-cover and soil moisture differences.
Further general support for this simplified view of ‘ideal’ heat island causation is
provided by Fig. 11. It shows the relationship between the maximum heat island intensity
(i.e. ‘ideal’ conditions) observed in a large number of cities and the sky view factor of their
central area. The good correlation achieved answers many of the questions posed in relation
to Fig. 3: the use of geometry alone attests to the importance of local site factors rather
18 T. R. OKE

Figure 1 I . Relationship between maximum heat island intensity (dTu-r(max)) and the sky view factor (t(lg)
in the centre of 31 settlements (after Oke 1981). All data refer to canopy layer measurements on calm, clear
summer nights.

than advective heat accumulation ; the restriction to summer data greatly reduces the pos-
sibility that anthropogenic heat is significant; and the merging of the European and North
American data suggests that geometry is the common underlying control (22 of the cities
used in Fig. 3 are included). The relationship of course does not exclude the possibility
that thermal admittance differences are also involved, but the similarity of building materials
used in constructing the centres of large cities may diminish their effect in Fig. 11.
These arguments refer to surface energetics and cooling and we are left to infer that
they are reflected in air temperature change. In detail, atmospheric cooling occurs as a
result of vertical and horizontal divergence in the transport of heat by radiation and/or
turbulence as given by the thermodynamic equation. Under perfect ‘ideal’ nocturnal
conditions when turbulence is suppressed and there are no phase changes in the air, the
cooling rate for an open rural site will be given by the vertical divergence of the radiative
flux :
aT aL*
= Z / P C ,

where p c , is the heat capacity of air. The equivalent rate for air within the urban canopy
layer is given by the volume divergence:
 THE ENERGETIC BASIS OF THE URBAN HEAT ISLAND 19

Vancouver (urban)
0 Canyon (meas.)
0 Canyon (talc.)
o Vancouver (rural)
-
-

- -
a
10 -
-
1 I I I I I I I I i d
12
0 2 4 6 8 10

TIME FROM SUNSET (h)

Figure 12. Cumulative cooling curves for canyon, city centre and rural environments on cloudless nights
with light winds. Canyon results from Nunez and Oke (1976), for explanation of calculations see text.

thereby accounting for the existence of net horizontal exchanges. In the real world turbu-
lence is rarely absent, but in the case of a canyon where the air may be almost stagnant,
due to greater frictional retardation, cooling may be approximated by this simple divergence
model. This is verified in Fig. 12 which shows the agreement between measured cooling
of the canyon air volume and that calculated from radiative divergence observations from a
three-dimensional array of net pyrgeometers (Nunez and Oke 1976). Although not strictly
comparable, the Vancouver urban cooling curve for ‘ideal’ conditions is also included in
Fig. 12 along with the corresponding rural curve. Similarity between the canyon and urban
cooling curves is evident.
In summary the observed features of the UCL nocturnal heat island under ‘ideal’
conditions are reasonably accounted for by the available energy balance information
involving only ‘causes’ A3 and A5 from Table 2.
There is very little information on which to base an explanation of the daytime features
of the ‘ideal’ heat island. In general terms, just as the evening heat island growth is produced
by the slower cooling rate of the city, so its early morning disappearance is the result of slug-
gish warming in the canopy layer. The absence of a sharp peak in warming, such as that
experienced in the countryside (Fig. I(d)), probably is due to the combined effects of canyon
shading at low Sun angles, higher thermal admittance and the lack of a capping inversion
within the lowest few hundred metres of the urban atmosphere. The existence of a ‘cool’
20 T. R. OKE

island in the middle of the day usually is attributed to canyon shading in the city centre
but full explanation awaits further detailed study.
In other than ‘ideal’ weather conditions the analysis becomes complicated especially
by the impact of increased turbulent transfer. The well established inverse relationships
between nocturnal heat island intensity and cloud and wind speed, and the direct relation
to increasing atmospheric stability, are clearly to be interpreted as reflecting a diminution
in the relative roles of radiation and heat storage in comparison with convection and
advection, the former tending to enhance differentiation of microclimates whereas the
latter tend to promote mixing and homogeneity.

(b) Urban boundary layer

Study of the mechanisms underlying the thermal excess in the boundary layer over
cities has been restricted largely to modelling efforts (both numerical and scaled flow)
with relatively little direct observation. The models are helpful in providing insight but
they require field verification not only in terms of their output but also their process under-
pinnings. Such work might reasonably start by investigating the four suggested ‘causes’
in Table 2, although full treatment of ‘cause’ B3 requires a canopy layer sub-model which,
as we have just discussed, is not yet available.
The increasing depth and temperature of the daytime mixed layer are the result of heat
flux convergence: both radiative due to aerosol effects and convective due to horizontal
and vertical transport and mixing. Both mechanisms are likely to be enhanced in the UBL
(Table 2). However whereas the net radiative impact (‘cause’ Bl) clearly seems to produce
increased warming (Section 3(c)), the convective impact is more complex. The greater
upward sensible heat flux from the city (‘causes’ B2 and B3, Section 3(c)) combined with
increased urban turbulence intensities (Counihan 1975) are likely to produce greater
bombardment of the base of the UBL capping inversion by convective plumes. This will
create an enhanced downward heat flux (‘cause’ B4). Therefore the UBL heat budget
benefits by increased convergence of heat both from above and below. On the other hand,
it should be mentioned that this process, plus the increased uplift experienced in the UBL
as a result of frictional retardation of airflow over the rougher city (Angel1 and Bernstein
1975; Auer 1975), is responsible for the doming of the layer over the city (Fig. 2(a)). This
tends to reduce the possible heat island intensity by increasing the volume of air affected.
Therefore all four UBL ‘causes’ listed in Table 2 are probably in operation by day, but
their relative roles remain to be evaluated.
At night all ‘causes’ except the radiative one are capable of explaining the UBL heat
island. However, if we take the summer fine weather case, ‘cause’ B3 would appear to be
the most important mechanism. Urban-rural surface energy balance comparisons point
to larger sensible heat input to the urban atmosphere in the late afternoon and a suppression
or even reversal of the return flux at night (Section 3(c)). This is due, at least in part, to the
development of the UCL heat island which is likely to generate instability between the
below- and above-roof levels. This certainly would fit the lower portion of the observed
temperature profile in the urban area at night (Fig. 2(b and c)). The upper neutral or stable
portions of that profile may be due to roughness rather than buoyancy effects resulting
in entrainment at the elevated inversion base. The ‘cross-over effect’ (Section 2(b)) might
then correspond to the layer from which heat has been removed, although it is possible
that it also might be due to long-wave radiative flux divergence at the top of the polluted
UBL.
One of the few reasonably clear examples of boundary layer heat island causation
concerns the high latitude settlement in winter. With calm and little or no solar heating,
the heat input to the UBL is dominated by anthropogenic releases (Bowling and Benson
1978). It can be shown also that even without a heat source the advection of stable air over
a rougher area will result in a positive thermal anomaly.
 THE ENERGETIC BASIS OF THE URBAN HEAT ISLAND 21

Perhaps the most important point regarding heat island causation is that the mechan-
isms responsible for the canopy layer anomaly are not the same as those in the boundary
layer. The former is largely the result of the immediate site character (especially building
geometry and materials), whereas the latter probably represents both an advective accumu-
lation of this warmer air with that entrained from above, and internal radiative effects.

5 . CONCLUSION
We conclude from the foregoing that in meteorological terms urban heat islands are
well described but rather poorly understood. Further descriptive studies of heat island
morphology in more towns and cities are likely to be of only local or applied interest,
though there are some exceptions. For example, present evidence is meagre, or even con-
flicting, for the cases of daytime heat and ‘cool’ islands; UBL structure and down-wind
urban ‘plumes’; and heat islands of tropical cities, the latter being of special practical,
and perhaps scientific, importance.
Understanding will be enhanced by greater attention to process. The present review
has shown that only the rudiments of the fundamental energy and water balances of cities
have been investigated, whereas those of most natural environments are far more advanced.
As a result, it is possible to provide little more than hand-waving arguments to explain
UCL and UBL heat islands under other than very idealized conditions. The major challenge
is to identify relevant processes and scales of activity in the urban atmosphere and to provide
consistent generalizations and syntheses as a basis from which to develop and test numerical
models. In the case of the heat island it would seem that this may require separate canopy
and boundary layer sub-systems with appropriate coupling.
Much of what has been said with respect to the heat island applies with equal or
greater force to research into urban wind and humidity effects and to the very complex
question of urban cloud and precipitation modification. If the challenge is not taken up
quickly urban meteorology will be a long time in approaching the goal of becoming a
predictive science.

ACKNOWLEDGEMENTS
My work in urban meteorology has been greatly helped by my colleagues and students
especially Drs R. Fuggle, J. Hay, M. Nunez, D. Steyn and D. Yap. I am also grateful to
the Natural Sciences and Engineering Research Council of Canada and the Atmospheric
Environment Service of Environment Canada who have provided financial support.

REFERENCES
Angell, J. K. and Bernstein, A. B. 1975 Flow across an urban area determined from double-
theodolite pilot balloon observations, J. App. Met.,
14, 1072-1079.
Atwater, M. A. 1971 The radiation budget for polluted layers of the urban
environment, J. App. Met., 10, 205-214.
1975 Thermal changes induced by urbanization and pollutants,
J. App. Met., 14, 1061-1071.
Auer, A. H. Jr. 1975 Calculations of divergence from the wind field over an
urban area, Proc. 2nd US Nat. Conf. Wind Eng. Res.,
WERC/NSF, Ft. Collins, Colo., pp. 1-20-1 to 1-20-3.
Bailey, W. G. 1977 Atmospheric and surface control on evapotranspiration
during soybean maturation, Ph.D. thesis, McMaster
Univ., Hamilton, Ont., 162 pp.
Bergstrom, R. W. Jr. and 1973 Modelling the effects of gaseous and particulate pollutants
R. Viskanta in the urban atmosphere, Part I: Thermal structure,
J . App. Met., 12, 901-912.
22 T.R. OKE

Berlyand, M. Ye., Kondrat’yev, K. 1974 Combined investigations of peculiarities of the meteoro-

Ya., logical rCgime of a big city in the example of Zaporozh’ye
Vasil’yev, 0. €3. (CAENEX-72), Met. i Gidrol., No. 1, pp. 14-23.
Vasil’chenko, 1. V.
Zhvalev, V. F.
Zrazhevskiy, 1. M. and
Ivlev, L. S.
Bornstein, R. D. I968 Observations of the urban heat island effect in New York
City, J . App. Met., 575-582.
Bowling, S. A. and Benson, C. S. 1978 Study of the subarctic heat island at Fairbanks, Alaska,
Envir. Monk Rep. No. EPA-600J4-78-027, US EPA,
Research Triangle Pk., N. C., 150 pp.
Bryson, R. A. and Ross, J. E. 1972 The climate of the city, Ch. 3 in Urbanization and Environ-
ment, Detwyler, T. R. and Marcus, M. G. (eds.)
Duxbury Press, Eelmont, 51-68.
Carlson, T. N., Dodd, J. K., 1981 Satellite estimation of the surface energy balance, moisture
Benjamin, S. G. and availability and thermal inertia, J . App. Met., 20,
Cooper, J. N. 67-87.
Chandler. T. J. 1965 The Climate of London, Hutchinson, London, 292 pp.
1967 Night-time temperatures in relation to Leicester’s urban
form, Met. Mug., 96, 244-250.
I970 Selected bibliography on urban climate, Tech. Note No.
155, WMO No. 276, World Met. Organiz., Geneva,
383 pp.
1976 The climate of towns, Ch. 14 in the ‘The Climate of the
British Isles’, Chandler, T. J. and Gregory, S. (eds.),
Longman, London, pp. 307-329.
Ching, J. K. S., Clarke, J. F. and 1978 The variability of the heat flux and mixed layer depth over
Godowitch, J. M. St. Louis, Missouri, WMO Symp. Bound. Layer
Physics App. Spec. Problems Air Poll., WMO No.
510, World Met. Organiz., Geneva, pp. 71-78.
Clarke, J. F. 1969 Nocturnal urban boundary layer over Cincinnati, Ohio,
Mo. Weath. Rev., 97,582-589.
Coakley, J. A. Jr. and Chplek, P. 1975 The two-stream approximation in radiative transfer:
including the angle of the incident radiation, J. Armos.
Sei.,32,409-418.
Coppin, P. A. 1979 Turbulent fluxes over a uniform urban surface, Ph.D.
thesis, Flinders Univ., Adelaide, 196 pp.
Counihan, J. 1975 Adiabatic atmospheric boundary layers: a review and
analysis of data from the period 1880-1972, Atmos.
Envir., 9, 871-905.
Daniel, C. E. J. and 1973 Urban temperature fields at Poona and Bombay, Indian J.
Krishnamurthy, K. Met. Geophys.. 24,407-41 2.
Duckworth, F. S. and 1954 The effect of cities upon horizontal and vertical temperature
Sandberg, J. S. gradients, Bull. Anier. Me!. SOC.,35, 198-207.
Fuggle, R. F. and Oke, T. R 1976 Long-wave radiative flux divergence and nocturnal cooling
of the urban atmosphere, I: Above roof-level, Bound.
Layer Met., 10, 1 13-1 20.
Hage, K. D. 1975 Urban-rural humidity differences, J . App. Met., 14, 1277-
1283.
Howard, L. 1833 The Climate of London, Vols. I-Ill, London.
Jauregui, E. 1973 Urban climate of Mexico City, Erdkunde, 27,298-307.
Kalanda, B. D., Oke, T. R. and 1980 Surburban energy balance estimates for Vancouver, B.C.
Spittlehouse, D. L. using the Bowen ratio-energy balance approach,
J. App. Met., 19, 791-802.
Kalma, J. D., Aston, A. R. and 1972 Energy use in the Sydney area, Proc. Eeol. Soc. Aust., 7,
Millington, R. J. 125-1 42.
Kawamura, T. 1964 Some considerations on the cause of city temperature at
Kumagaya City, Geog. Rev. Jupan., 37, 560-565.
Lacy, R. E. 1977 Climate aHd Building in Britain, Build. Res. Estab. Report,
HMSO, London, 185 pp.
Landsberg, H. E. 1979 Atmospheric changes in a growing community, Urban
Ecol., 4, 53-82.
Landsberg, H. E. and Maisel, T. N 1972 Microrneteorological observations in an area of urban
growth, Bound. Layer Met., 2, 365-370.
 THE ENERGETIC BASIS OF THE URBAN HEAT ISLAND 23

Lee, D. D. 1975 Rural atmospheric stability and the intensity of London’s

heat island, Weather, 30, 102-109.
1979 Contrasts in warming and cooling rates at an urban and a
rural site, Weather, 34, 60-66.
Ludwig, F. L. 1970 Urban air temperatures and their relation to extraurban
meteorological measurements, Amer. SOC. Heat.
Refrig. Air-Condit. Eng., Publ. No. SF-70-9, pp. 40-45.
Mather, J. R. 1978 The Climatic Water Budget in Environmental Analysis,
Lexington Books, Lexington, Mass.
Nicol, K. S. 1976 Anthropogenic heat and its relation to building and urban
climate in Inuvik, N.W.T., MSc. thesis, Univ. Brit.
Columbia, Vancouver, B.C., 130 pp.
Noilhan, J. 1980 Contribution a I’etude du microclimat au voisinage d‘un
bltiment, Ph.D. thesis, L’universite Paul Sabatier de
Toulouse, Toulouse, 160 pp.
Nunez, M. 1974 The energy balance of an urban canyon, Ph.D. thesis, Univ.
Brit. Columbia, Vancouver, 161 pp.
Nunez, M. and Oke, T.R. 1976 Long-wave radiative flux divergence and nocturnal cooling
of the urban atmosphere, 11: Within an urban canyon,
Bound. Layer Met., 10, 121-135.
1977 The energy balance of a n urban canyon, J . App. Met., 16,
11-19.
1980 Modelling the daytime urban surface energy balance,
Ceog. Analysis, 12, 373-386.
Oke, T. R. 1973 City size and the urban heat island, Atmos. Envir., 7, 769-
779.
1974 Review of urban climatology, 1968-1973, Tech. Note NO.
134, WMO No. 383, World Met. Organiz., Geneva,
132 pp.
1976 The distinction between canopy and boundary layer urban
heat islands, ATMOSPHERE, 14, 268-277.
1978a Boundary Layer Climates, Methuen, London, 372 pp.
1978b Advectively-assisted evapotranspiration from irrigated
urban vegetation, Bound. Layer Met., 17, 167-173.
1978c Surface heat fluxes and the urban boundary layer, WMO
Symp. Bound. Layer Phys. App. Spec. Problems Air
Poll., WMO No. 510, World Met. Organiz., Geneva,
pp. 63-69.
1979 Review of urban climatology, 1973-1976, Tech. Note No.
169, WMO No. 539, World Met. Organiz., Geneva,
100 PP.
1981 Canyon geometry and the nocturnal urban heat island:
Comparison of scale model and field observations, J.
Climatol., 1 (in press).
Oke, T. R. and East, C. 1971 The urban boundary layer in Montreal, Bound. Layer Met.,
1,411-437.
Oke, T. R. and Fuggle, R. F. 1972 Comparison of urban/rural counter and net radiation at
night, Bound. Layer Met., 2, 290-308.
Oke, T. R., Kalanda, B. D. and 1981 Parameterization of heat storage in urban areas, Urban
Steyn, D. G. ECOl., 5, 45-54.
Oke, T. R. and Maxwell, G.B. 1975 Urban heat island dynamics in MontrCal and Vancouver,
Atmos. Envir., 9, 191-200.
Peterson, J. T. and Stoffel, T. L. 1980 Analysis of urban-rural solar radiation data from St.
Louis, Missouri, J. App. Met., 19, 275-283.
Priestley, C. H. B. and 1972 On the assessment of surface heat flux and evaporation using
Taylor, R. J. large-scale parameters, Mo. Weath. Rev., 100, 81-92.
ProbBld, F. 1975 A v8rosklimatologia idoszerii feladatai, Zdojhrus, 79, 69-76.
Rouse, W. R., Noad, D. and I973 Radiation, temperature and atmospheric emissivities in a
McCutcheon, J. polluted urban atmosphere at Hamilton, Ontario,
J . App. Met., 12, 798-807.
Sellers, W. D. 1965 Physical Climatology, Univ. Chicago Press, Chicago, 272 pp.
Smith, F. B. and Hunt, R. D. 1978 Meteorological aspects of the transport of pollution over
long distances, Atmos. Envir., 12, 461-477.
Spangler, T. C. and Dirks, R. A. 1974 Meso-scale variations of the urban mixing height, Bound.
Layer. Met., 6, 423-441.
24 T.R. OKE

Sundborg, A. 1951 Climatological studies in Uppsala with special regard to the

temperature conditions in the urban area, Geographica,
Geog. Instit., Uppsala, No. 22, 111 pp.
Taesler, R. 1981 Urban meteorological studies in Uppsala, Acta Univ.
Uppsala, Abstr. Uppsala Dissert. Faculty of Sci.,
No. 592, Uppsala, 21 pp.
Takahashi, K. 1959 Relation between the air temperature distribution and the
density of houses in small cities of Japan, Geog. Rev.
Japan, 32, 305-313.
Tanner, C. B. and Pelton, W. L. 1960 Energy balance data Hancock, Wisconsin, Soils Bull. No. 2,
Dept. Soils, Univ. Wisconsin, Madison, Wisc., 110 pp.
Terjung, W. H. and Collaborators 1971 The effect of cyclonic storm on the energy fluxes at the
urban interface - a preliminary experiment, Archiv.
Met. Geophys. Bioklim., Ser. B, 19, 367-416.
Terjung, W. H. and Louie, S. S-F. 1974 A climatic model of urban energy budgets, Geog. Analysis,
6, 341-367.
Terjung, W. H. and 1980 Influences of physical structures on urban energy budgets,
O’Rourke, P.A. Round. Layer Met., 19,421-439.
Tyson, P. D., Du Toit, 1972 Temperature structure above cities : review and preliminary
W. J. F. and Fuggle, R. F. findings from the Johannesburg urban heat island
project, Atmos. Environ., 6, 533-542.
Unwin, D. J. 1980 The synoptic climatology of Birmingham’s urban heat
island, 1965-74, Weather, 35, 43-50.
White, J. M., Eaton, F. D. and 1978 The net radiation budget of the St. Louis metropolitan
Auer, A. H. Jr. area, J. App. Met., 17, 593-599.
Yap, D. and Oke, T. R. 1974 Sensible heat fluxes over an urban area - Vancouver, B.C.,
J. App. Met., 13, 880-890.