Satellite analysis of radiation and heat fluxes during the Basel urban boundary layer experiment (BUBBLE)

Because urban areas show a different climate when compared to their surroundings and alter local
radiation and energy balances they are an important topic in climatology. Urban areas are also highly
heterogeneous areas when compared to rural ones, little research with satellite data has been
conducted until now. It is the goal of this thesis to model and illustrate, with the use of remotely
sensed data, urban alterations of the radiation and energy flux densities in the spatial domain. Besides
very detailed field measurements of the structure and dynamics of the urban boundary layer with a
large in situ data network from the Basel Urban Boundary Layer Experiment (BUBBLE), remote
sensing data were analyzed and validated from several satellites. The BUBBLE experiment was a joint
European research project under the umbrella of COST (Coopération Européenne dans la domaine de
la recherche Scientifique et Technique, COST 715: Meteorology applied to urban pollution problems).
For this purpose data from AVHRR, MODIS, ASTER, LANDSAT ETM+ and Quickbird were
acquired and processed. After each step of the modelling the results from the remotely sensed data
were compared and validated with the in situ data.
The first step was the validation of the thermal infrared (TIR) satellite data and an accuracy
assessment of six different Split-Window algorithms for the AVHRR. The results for the different
sensors showed an average accuracy of less than ±5 % even in urban environments for the different
sensors.
Afterwards the net all-wave radiation (Q*) was modelled with shortwave inputs derived from the
Short Wave Irradiance Model (SWIM). The modelled broadband albedo was also derived from
satellite data. The results of Q* showed a good mean absolute difference (MAD) of 26 Wm-2 over
rural and urban surfaces. The spatial distribution of Q* also agreed fully with the in situ results
showing a lower Q* for the urban areas than for the countryside. For a very high resolution modelling
of Q* in the city an experimental approach with thermal imagery from a helicopter overflight together
with data from Quickbird was used and showed the extent to which Q* in a city is influenced by the
albedo of the vegetation.
From the available Q* the storage (or ground) heat flux ΔQS was modelled using three different
models: the complete aspect ratio model (CAR), the Normalized Difference Vegetation Index (NDVI)
and the Objective Hysteresis Model (OHM). The most useful results were achieved with the OHM,
which was applied and validated with satellite data over an urban surface for the first time. The MAD
was 17 Wm-2 with an RMSE of also of 17 Wm-2.
After the successful modelling of the ground heat flux density, the latent QE and sensible QH heat flux
densities were modelled with a combined Bowen-Ratio /NDVI approach resulting in a MAD of 28
Wm-2 and 18 Wm-2 respectively.
All the results of this thesis provided quite accurate representations of the distribution of the radiation
and heat flux densities, as well as of the differences between rural and urban surfaces; therefore, the
model was applied and validated using datasets acquired from 2003 for the same research area,
showing similar results as for the BUBBLE campaign. This shows the possible transferability of the
model to other times and dates.
With the model described in this thesis the radiation and energy flux densities can be modelled
accurately in the spatial domain over urban (and rural) surfaces and used both for further urban
climatology studies and for urban planning.