Ensuring accuracy throughout all components of a flood forecasting system

This week our team has been looking into the Sendai Framework for Disaster Risk Reduction. This is an international framework that spans fifteen years between 2015-2030 and presents a roadmap to allow communities to become safer and more resilient to natural disasters. It was adopted in 2015 by members of the United Nations, including Australia. It is of relevance to our work in natural disasters due to the growing emphasis on early warning systems and emergency management.

Whilst the traditional management approach to extreme weather has included hard defences and infrastructure, there is now recognition that earlier warnings can play a significant role in disaster management. It is a goal of the Sendai Framework to substantially increase access to multi-hazard early warning systems and disaster risk information and assessments over its 15 year implementation period.

But creating a system to accurately issue warnings is complex, particularly when they are weather related. Our own experience developing flood warning systems shows us that they are becoming increasingly complex, and consists of a growing number of interdependent components. Inaccuracy in any single component can compound through the system, leading to either a missed event or false alarms – both reducing the performance and trust within the system. Whilst some inaccuracy cannot be avoided, uncertainty must be quantified and where possible designed out of the system before implementation.

The development and implementation of a flood forecasting system must be fit-for-purpose, readily understandable and provide the necessary intelligence to enable effective disaster response and recovery. There is a spectrum of forecasting architecture ranging from manual to fully automated and simple to complex with the architecture needing to consider regional needs and individual catchment characteristics and response time requirements.

Scale is another important consideration with some systems supporting emergency management at a community level and other flood early warning systems spanning entire countries. Regardless of the architecture or scale all forecasting systems typically use three components - meteorological inputs, hydrologic models and outputs (flood levels or flood maps).

Figure 1. Considerations in determining the fit-for-purpose architecture requirements for a flood forecasting system

Meteorologic forecasts: These drive the system, and include the intensity, volume and spatial location of predicted rainfall. Inaccuracy in any of these elements increases the further into the future the system is predicting.

Hydrologic Model: This model analyses the rate of rainfall falling over a catchment, it interacts with estimated soil moisture, predicts infiltration losses, and estimates the amount of runoff that can be expected during a storm.

Flood levels and depths: Runoff needs to be converted into a water level or depth to understand its impact to people, property, infrastructure and the environment. This intelligence can be gathered once the impacts are calculated, either through pre-established flow vs water level relationships (rating tables), pre-computer flood maps for different flows, or live hydraulic modelling. Each approach has its strengths and weaknesses in terms of computational speed and accuracy, but they are all influenced by the quality of the input information and assumptions made by the modeller, e.g. if bridges are blocked with debris or not.

Warnings: Predicting if a flood is going to happen is only the first step in the process of developing an effective flood forecasting system. They next step is to ensure alerts and warnings make their way to the intended recipients and are issued and understood by stakeholders. A ‘spaghetti plot’ (yes it’s a thing) may be good for a hydrologist to understand statistical likelihoods but is not useful to a Local Disaster Coordinator to make decisions around evacuation.

Featured Posts
Recent Posts
Search By Tags
No tags yet.
Follow Us
  • Facebook Basic Square
  • Twitter Basic Square
  • Google+ Basic Square