Flood Forecasting and Early Warning: An Example from the UK Environment Agency - Reducing Disaster: Early Warning Systems for Climate Change - page 188

Table 10.1 Methods used for forecasting fl ooding
(i) Levels, records and trends techniques
(a) Historic fl ood records A comprehensive recording of historical fl ooding on a
catchment basis, with detailed information on river-moni-
toring sites. Provides an indication of how a site may react
to different weather conditions, based on its previous
behaviour. This is used as backup to more technical
methods, or where no other methods are available
(b) Alarm levels Based on known information, alarm levels will be set up on
river-monitoring sites, at appropriate levels, to alert of high
and rising river levels
(c) Linear trend It is used to record the rise in level at a river level gauge over
the previous 1 or 2 h and calculates the same rate of rise
into the future
(ii) Modelling and correlation techniques
(a) Rainfall to river level

correlation

A relationship between amount of rainfall, which can be used
to predict the peak river level. Used for river gauges which
are furthest upstream without the availability of any other
information

(b) Upstream and

downstream river
level correlation

Correlation method used as a forecasting technique by
monitoring the river level at an upstream river gauge to
predict a level at another gauge downstream. Based on a
relationship between two sites that is easily repeated

(c) Rainfall-runoff

modelling

Different versions exist. For example, Probability Distributed
Model (PDM) and Physically Realisable Transfer Function
(PRTF) Model. Real-time rainfall and river data is fed into
the models to predict timing and magnitude of maximum
river levels or fl ows. The model can use rainfall forecast
taken from radar prediction. The model depends on good
quality data to calibrate accurately and good quality rainfall
forecast to produce extra lead time with accurate results

(d) Routing modelling Routing models route the fl ow from upstream to downstream

river-monitoring stations, taking account of infl ow from
other tributary systems

(e) Hydrodynamic

modelling

Different versions exist. For example, ISIS and Mike 11 model
software. Originally developed for fl ood mapping purpose
but now adapted for use as a real-time forecasting tool.
These model river systems on detailed topographic survey
data rely on detailed topographic survey data to replicate
the 'real-world' river systems in a computer, to forecast
levels and fl ows at required locations

10.5.5 Delft-FEWS

Delft-FEWS is a confi gurable shell with an open interface data handling platform.
It provides tried-and-tested code to drive a wide range of existing forecasting models
with real-time data from hydrological and meteorological observations and permits
data assimilation, and the dissemination of prediction results through appropriate
products to the warning process (Werner et al. 2013 ) (Fig. 10.7 ).