Applications of Medium Range Probabilistic Flood Forecast for Societal Benefi ts: Lessons Learnt from Bangladesh - Reducing Disaster: Early Warning Systems for Climate Change - page 163

For a fl ood-prone country like Bangladesh, fl ood forecast technology plays an
extremely crucial role in saving lives and properties. The Flood Forecasting and
Warning Centre (FFWC) of Bangladesh Water Development Board (BWDB) is
responsible for fl ood forecasting within Bangladesh. The fl ood forecasting models
used by the FFWC are based on MIKE 11, a one-dimensional modelling software
used for the simulation of water levels and discharges in the rivers for up to 48-72 h
deterministic forecasts. The experimental model produces 1-10 day probabilistic dis-
charge forecasts. Due to its probabilistic nature, it has many limitations at the user
level for interpretation, translation and understanding of the early warning message.
Although there are many disaster risk management initiatives implemented by differ-
ent INGOs and NGOs, there are limited research and capacity building activities on
the application of medium range probabilistic fl ood forecasts information. For exam-
ple, the Center for Environmental and Geographical Information Service (CEGIS)
is piloting fl ood early warning system at the community level, using existing 48 h
deterministic forecasts (Riaz et al. 2010 ). However, 2 days' lead time could only be
used for emergency evacuation and household preparation work.

Box 9.1 Forecasting

• Short range forecast: Beyond 12 h and up to 72 h.
• Medium range forecast: Beyond 72 h and up to 240 h.
• Extended range forecast: Beyond 10 days and up to 30 days.
• Long range forecast: From 30 days up to 2 years.
• Ensembles forecasts: Ensemble forecasts are forecasts that contain a num-
ber of alternative predictions for the same forecast period. One such pre-
diction is called an ensemble member.
• Probabilistic forecasting: A technique for forecasting that relies on differ-
ent methods to establish an event occurrence/magnitude probability.

Source: WMO- GDPS , WMO ( 2012 )

To model this annual fl ooding occurrence for improved prediction, research on the
generation of ensemble medium range fl ood forecast, started in 2001, is developing a
series of forecasting schemes to increase lead time of fl ood forecasting in Bangladesh.
The forecasts system was tested in real time since 2007. In Bangladesh, a recent major
fl ood was in 2007.

9.2 Institutional Arrangement for Probabilistic Forecasts

Strong institutional networking and commitments have facilitated the development
of fl ood forecasting schemes and their application. At the international level,
Climate Forecast Application Network (CFAN) of Georgia Institute of Technology