Environmental Engineering Reference
In-Depth Information
Coupling Meteorological and
Hydrological Models for Real-Time
Flood Forecasting
10
GEOFF AUSTIN, BARNEY AUSTIN,
LUKE SUTHERLAND-STACEY AND
PAUL SHUCKSMITH
Overview
Hydrological models themselves have im-
proved significantly in recent years due to im-
proved computer performance and their ability
to capture the physical processes on a higher spa-
tial and temporal resolution. Models are now
much better at taking advantage of the data pro-
ducts offered by nowcasting and NWP models.
Computers and telemetry technology are also
much quicker at processing information.
Despite recent advances, it is clear that QPF
remains a difficult problem. This is in part because
precipitation is not one of the primary variables of
themodels and is estimated indirectly, often using
some sort of parameterization scheme. The appro-
priateness of the scheme depends on the ability of
the model to diagnose correctly the dominant
rainfall-producing process. However, to the extent
that NWP is an initial value problem, further
improvements should bemade bymore accurately
specifying the initial atmospheric state at higher
resolution and in more detail.
It is well understood that the Earth's atmo-
sphere is a chaotic non-linear system, resulting
in predictions of behaviour that are sensitive to
initial conditions (Lorenz 1963). Small perturba-
tions in initial conditions can lead to significant
differences later. Indeed, the growth of these dif-
ferences is what limits the time ahead for which
the forecasts are useful. More recently it has been
demonstrated that some indications about the
Radar and satellite data have been used success-
fully for years in rainfall and flood forecasting. In
fact, nowcasting (i.e. short-range forecasts based
largely on the extrapolation of current informa-
tion) forms the backbone of many real-time flood
warning systems, particularly for small urban
catchments where the time between rainfall and
serious flooding can be short. The use of meteo-
rological Numerical Weather Prediction (NWP) is
less common, but recently these models have
undergone major improvements, mainly due to
increases in spatial resolution and more sophisti-
cated representations of important processes, par-
ticularly those associated with rainfall production
and convection. These improvements have in-
creased forecast accuracy. Combining the advan-
tages of nowcasting for short lead times with the
benefits of NWP models for longer range forecasts
results in the best of both worlds in terms of
Quantitative Precipitation Forecasting (QPF) for
hydrological applications, particularly real-time
flood forecasting.
