Flood forecasting systems

April 10, 2020

Forecast data is great, but only great if it can be used to its full potential and can be used to confidently provide insight. So how can we do this? For JBPacific, Delft-FEWS is the preferred approach. Delft-FEWS is the same ‘open data handling platform initially developed as a flood forecasting and warning system’ used by the Bureau of Meteorology. 

 

What data do we use? 

 

Our flood forecasting systems use Delft-FEWS to read in weather forecasts directly from the Bureau of Meteorology, including data on forecast rainfall, wind, temperature and tide, as well as the newer Rainfields product which provides an ensemble of rainfall forecast. In addition to forecast information, we also combine observed data including recorded rain, river height and soil moisture to establish a view of flood conditions from the now to up to 7 days into the future. 

 

FEWS Display of QLD ADFD Rainfall and QLD AWRA-L Soil Moisture.

 

What are we forecasting?

 

Now the forecast and observed data is in the Delft-FEWS system, catchment flows can be forecasted using a calibrated URBS (Unified River Basin Simulator) hydrology model. Although there are many hydrologic models, URBS is a widely chosen option within the industry and has the benefit of linking-to third party software such as Enviromon, FEWS and Tuflow.

By using forecast rainfall over the catchment, flows can be forecast at a sub-catchment scale which can provide great insight on how water is expected to pass through a catchment.

 

FEWS Display of URBS forecast flows at specific sub-catchment boundaries.

 

The end results.

 

Flow forecasts can provide great insight to how much water is expected to pass through a catchment, but it is not necessarily the best way to provide intelligence on the possible damages or problems that the catchment could experience in the next week. So, we use hydraulic modelling to understand how the catchment discharge relates to flood levels. These forecast levels can now be related to flood extents maps, or even bridge levels to indicate possible inundation and bridge closures.

 

As the forecast window increases, the accuracy of the forecast becomes more uncertain, but limited heads-up makes it hard for response or warning to occur. Therefore, flood forecasting is a balance between timeliness of information and the accuracy of the forecast. The accuracy of forecast rainfall datasets is a challenge, which is why JBPacific performs extensive performance testing and R&D on our forecasting systems to better understand the accuracy of our forecasting systems.

 

Without a multi-disciplinary team of hydrologists, hydraulic engineers, software engineers and technology specialists, understanding flood behaviour and realising the full potential of a flood forecasting system would not be possible.

 

 

 

 

 

 

 

Please reload

Featured Posts

Machine Learning for Modelling Cyclones

July 29, 2020

1/1
Please reload

Recent Posts
Please reload

Archive