Geoscience Reference
In-Depth Information
and visualize present flow within an urban district of Lisbon and to predict flow
modifications, should the city continue to grow windward.
Empirical models are based on observation. The objective is to reproduce
different parameters using statistical relations derived from observation [MAS 06].
“The approach is based on the assumption that the physical behavior is already
contained on the observed data” [MAS 06]. Data from a great number of measuring
points must be used. Several empirical models will be described for Lisbon,
concerning the study of air temperature and “physiologically equivalent
temperature” spatial variations at the mesoscale and at the microscale. Sometimes
the procedure has been combined with other models (
ENVI-met
,
RayMan
). It is clear
from the literature that a combination of models is used in most cases. “Using
complementary approaches, there is reason to believe we can gain comprehensive
understanding of turbulent flow, and the radiation and energy balances of the urban
areas” [KAN 06].
5.7. Modeling Lisbon's urban climate at the mesoscale
5.7.1.
Empirical modeling and geographical information to estimate and
interpolate near ground air temperature
Detailed maps of different thermal patterns need to be constructed for various
purposes, including the application of urban climate results [ALC 05; ALC 09]
(http://pdm.cm-lisboa.pt/pdf/RPDMLisboa_avaliacao_climatica.pdf).
This has been carried out in Lisbon using a series of empirical models [AND 03]
and adapting procedures previously followed by Carrega [CAR 92], Alcoforado
[ALC 94], Joly and Fury [JOL 96], and Andrade [AND 98]. The factors influencing
spatial variation of temperature were first detected by the way of a stepwise multiple
regression [WIL 95]. Temperature data were then computed within a grid, and
finally, thermal maps were produced [AND 03]. This methodology is similar to that
described by Joly (see Chapter 2), but adapted to an urban area. The use of a GIS is
indispensable for carrying out these procedures. Moreover, the GIS has the
advantage to be easily updated [SVE 02].
The data used for this experiment were collected from 20 non-rainy days in
September 2001 and in February 2002. Thirty minutes averaged temperatures were
computed during night-time over the 20 investigated days. Air temperatures were
standardized for each measuring point for each 30-minute period [ALC 06a]. As the
parameters influencing air temperature vary over time, the first step involved
building up groups consisting of days with thermal patterns. This was achieved
using an automatic classification algorithm (Ward's minimum variance method)
[WIL 95]. The algorithm identified seven air temperature patterns related to the
same number of weather types. For each pattern, average temperature was first
calculated, together with standardized temperature (T
az
) for each measuring point.
