Assessing urbanisation effects on rainfall-runoff using a remote sensing supported modelling strategy

This paper aims at developing a methodology for assessing urban dynamics in urban catchments and the
related impact on hydrology. Using a multi-temporal remote sensing supported hydrological modelling
approach an improved simulation of runoff for urban areas is targeted. A time-series of five medium
resolution urban masks and corresponding sub-pixel sealed surface proportions maps was generated
from Landsat and SPOT imagery. The consistency of the urban mask and sealed surface proportion timeseries
was imposed through an urban change trajectory analysis. The physically based rainfall-runoff
model WetSpa was successfully adapted for integration of remote sensing derived information of detailed
urban land use and sealed surface characteristics.
A first scenario compares the original land-use class based approach for hydrological parameterisation
with a remote sensing sub-pixel based approach. A second scenario assesses the impact of urban growth
on hydrology. Study area is the Tolka River basin in Dublin, Ireland.
The grid-based approach of WetSpa enables an optimal use of the spatially distributed properties of
remote sensing derived input.
Though change trajectory analysis remains little used in urban studies it is shown to be of utmost
importance in case of time series analysis. The analysis enabled to assign a rational trajectory to 99% of
all pixels. The study showed that consistent remote sensing derived land-use maps are preferred over
alternative sources (such as CORINE) to avoid over-estimation errors, interpretation inconsistencies and
assure enough spatial detail for urban studies. Scenario 1 reveals that both the class and remote sensing
sub-pixel based approaches are able to simulate discharges at the catchment outlet in an equally satisfactory
way, but the sub-pixel approach yields considerably higher peak discharges. The result confirms
the importance of detailed information on the sealed surface proportion for hydrological simulations
in urbanised catchments. In addition a major advantage with respect to hydrological parameterisation
using remote sensing is the fact that it is site- and period-specific. Regarding the assessment of the impact
of urbanisation (scenario 2) the hydrological simulations revealed that the steady urban growth in the
Tolka basin between 1988 and 2006 had a considerable impact on peak discharges. Additionally, the
hydrological response is quicker as a result of urbanisation. Spatially distributed surface runoff maps
identify the zones with high runoff production.
It is evident that this type of information is important for urban water management and decision
makers. The results of the remote sensing supported modelling approach do not only indicate increased
volumes due to urbanisation, but also identifies the locations where the most relevant impacts took place.