Memorandum submitted by Richard Long (FL
145)
SHORT TERM URBAN FLOOD FORECASTING
Urban flood forecasting would help water companies,
local authorities, the Environment Agency, the emergency services
and highway authorities be better prepared when adverse weather
conditions occur.
Urban flood forecasting works by reading Met
Office short-term weather forecast data in the 1 to 6 hour in
the future range from the Met Office FTP site. This data is available
to water companies by subscription. Depending on the location,
the weather forecast data is at either 1x1 km or 2x2 km resolution
as predictions of rainfall intensity at five minute intervals.
The forecast rainfall data is used as input to a fast drainage
network simulation system. The network simulations are automatically
re-run every time updated weather forecasts are downloaded, producing
predictions of flooding from the sewer system. A flood routeing
system uses mapping and digital ground model data combined with
local knowledge where available to determine the likely impacts
of the predicted flooding. At impact thresholds that can be default
values or preset for specific areas, the urban flood warning system
will go into alert mode and can be configured to text, call, email
or page duty staff with details of the predicted incidents, including
location, predicted timing and severity. Warnings would be updated
as revised forecasts are processed.
The urban flood forecasting system would be
configured to run continuously without the need for operator intervention.
The flow chart below illustrates the process.
The benefits of urban flood forecasting could
be:
1. Mobilisation of call centre and operations
staff before the flood emergency happens.
2. Time to put contingency plans into operation.
3. The opportunity to check standby pumps
and other equipment are operating.
4. Setting up traffic diversions in advance
of gridlock occurring.
5. The opportunity to get emergency services
and operational crews to critical locations in advance of the
emergency.
6. Improved service to the public.
7. Warning of likely pollution events in
rivers resulting from overloading of the drainage system during
rainfall.
An urban flood forecasting system would include
a number of components, many of which already exist having been
developed for other purposes. Some limited further development
could now make such a system a reality.
|
| Component | Availability
| Provider |
|
| Weather forecasts 1 to 6 hours ahead in the form of predictions of rainfall intensity over 1 or 2 km squares
| Available now in many areas. Other areas would require installation of weather radar stations. Data is available to download from ftp site.
| Met Office |
| Software to convert radar forecasts into input data for drainage network simulator
| Available now | Black & Veatch/Met Office
|
| Interface to take rainfall forecast data and covert it into runoff for input to drainage network model
| Development required | |
| High-speed drainage network simulator to convert runoff into sewer flow and predictions of flooding.
| Available now | FastNett fast sewer network simulator from Mouchel Group
|
| High-speed two-dimensional flood flow surface routing simulator to take predictions of flooding from drainage systems and route them over land
| Available now | Flood Risk Mapper from Mouchel Group
InfoWorks from Wallingford Software
|
| Flood impact assessment tool to identify which properties are likely to flood
| Available now | Part of Flood Risk Mapper from Mouchel Group
|
| Generation of alerts | Development required
| |
| Software environment to control automatic downloads, simulation and warning cycles
| Development required. | |
|
Summarising the above table, the "difficult" parts
of the system, ie the weather forecasts and hydraulic simulation
engines already exist. They simply need to be connected and wrapped
in an overall control system.
Perhaps the biggest area of uncertainty is whether the Met
office forecasts are sufficiently accurate to avoid many false
alarms being generated while failing to predict things that subsequently
happened. Analysing archive data of forecasts and actuals that
the Met Office hold and are willing to make available could test
that.
Richard Long
November 2007
| 
