Theor. Appl. Climatol. 74, 93–103 (2003)
DOI 10.1007/s00704-002-0714-4
1 Department of Atmospheric Sciences, University of Buenos Aires, Ciudad Universitaria, Buenos Aires, Argentina
2 Department of Atmospheric Sciences=Centro de Investigaciones del Mar y la Atm
osfera (CIMA),
University of Buenos Aires – CONICET, Ciudad Universitaria, Buenos Aires, Argentina
Objective method for classifying air masses: an application
to the analysis of Buenos Aires’ (Argentina) urban
heat island intensity
R. A. Bejara´n1 and I. A. Camilloni2
With 3 Figures
Received October 24, 2001; revised June 12, 2002; accepted October 10, 2002
Published online January 17, 2003 # Springer-Verlag 2003
Summary
During recent years, numerous studies have examined the
Buenos Aires urban climate, but the relationship between
large-scale weather conditions and the Buenos Aires urban
heat island (UHI) intensity has not been studied. The goal
of this paper is to apply an objective synoptic climato-
logical method to identify homogeneous air masses or
weather types affecting Buenos Aires during winter, and
to relate the results to the UHI intensity. A K-means
clustering method was used to define six different air
masses considering the 03:00, 09:00, 15:00 and 21:00 LT
surface observations of dry bulb temperature, dew point,
cloud cover, atmospheric pressure and wind direction and
velocity at Ezeiza, the most rural meteorological station of
the Buenos Aires metropolitan area (Fig. 1). Results show
that the mean UHI intensity is at its maximum (2.8
C) a
few hours before sunrise when conditions are dominated by
cold air masses associated with cold-core anticyclones,
weak winds and low cloud cover. Inverse heat islands are
found during the afternoon for all air masses indicating that
surface processes are not dominant at that time. The rela-
tively infrequent and warmest air mass is the only one that
presents a mean negative urban-rural temperature differ-
ence ( 0.1
C) during the afternoon with the smallest di-
urnal cycle of the UHI intensity probably due to the
prevailing high humidity and cloudy sky conditions. The
paper provides an insight into the Buenos Aires urban–rural
temperature difference under a variety of winter weather
types and results could be useful to improve local daily
temperature forecasts for the metropolitan area of Buenos
Aires on the basis of the routine forecasts of weather types.
1. Introduction
Urbanization processes produce many changes in
the surface nature and in the properties of the
local atmosphere. These processes involve an
alteration of the local climate as a result of the
transformations in radiative, thermal, moisture
and aerodynamic characteristics of the surface
and changes in the turbulent fluxes of momen-
tum, heat and water vapor. Nevertheless the
urban climate at any locality is generally gov-
erned by the large-scale weather conditions. The
interaction between the synoptic scale and the
local scale is a continuous seesaw: sometimes
the large-scale weather conditions are dominant
and at others the local conditions are prevalent
(Landsberg, 1981). For example, Lowry (1977)
pointed out the fact that the weather type situa-
tion should be considered in conducting statisti-
cal studies of the urban effect on temperature.
One of the most noticeable consequences of
urban growth is the consistent rise in the urban

temperature (i.e. Bornstein, 1968; Oke, 1979;
Colacino and Rovelli, 1983; Karl et al., 1988;
Barros and Camilloni, 1994; Camilloni and
Barros, 1997; Figuerola and Mazzeo, 1998).
The different surface structure between the urban
and surrounding rural environment produces the
well-known urban heat island (UHI) phenom-
enon. Oke (1982) lists a number of factors con-
tributing to the UHI, including altered energy
balance terms leading to positive thermal anom-
aly, anthropogenic heat sources, increased sensi-
ble heat storage, decreased evapotranspiration
due to construction materials and decreased total
turbulent heat transport due to wind speed reduc-
tion caused by canyon geometry. In addition, the
dependency of the UHI effect on wind speed,
cloud cover and the near-surface lapse rate has
been studied by many authors (i.e. Sundborg,
1950; Chandler, 1965; Oke, 1973; Lee, 1975;
Godowitch et al., 1985; Moreno-Garcı´a, 1994;
Kidder and Essenwanger, 1995; Camilloni,
1999; Runnalls and Oke, 2000; Morris et al.,
2001). These studies show that wind speed and
the amount of cloud cover are the most signifi-
cant meteorological parameters controlling the
intensity and development of the UHI. This is
because these variables define the ventilation
and insolation in and around the urban region.
Oke (1998) showed that by holding the effects
of clouds constant, by only using cloudless data,
Fig. 1. Map of Buenos Aires metropolitan area ( ) and location of meteorological stations considered in this study: Buenos
Aires Central Observatory (BACO) and Ezeiza
94 R. A. Bejara´n and I. A. Camilloni