England & Wales, Global, Ireland, Scotland Climate action and sustainable development

Predicting the flood


Image by Roman Grac from Pixabay

Red sky at night shepherd’s delight. Red sky in the morning shepherd’s warning.

Predicting the weather let alone where and when flooding may occur continues to be complex science.  Radar, satellites, supercomputers, artificial intelligence and rain gauges have greatly improved our ability to forecast the weather, but even the technology still gets it wrong and we are still learning to get better at more accurate predictions.

The floods of 2007, affecting Gloucestershire and Worcestershire, led the Government of the day to carry out the Pitt Review on Flooding, this, in turn, led to the Floods & Water Management Act 2010.  This required, among many other related issues, the formation of Local Lead Flood Authorities (LLFAs) – essentially county councils and unitary authorities, which lead on managing local flood risks (i.e. risks of flooding from surface water, groundwater and ordinary (smaller) watercourses). This includes ensuring co-operation between the Risk Management Authorities in their area.

Most LLFAs now have flood models they use to help understand where flooding may occur be it pluvial (surface water), fluvial (river) or groundwater. However, these can be slow and costly to run and often don’t provide a high degree of certainty.  Frequently, local knowledge and other factors may be used to help determine when to believe the flood model or not or whether an area will be flooded.

The LGiU with Imperial College London has been involved in a three year EU project (funded by Urban Europe) led by Vrije Universiteit Brussel in Belgium with partners from the Netherlands and the UK to develop an early warning flood alert service called FloodCitiSense. This is being created for and by citizens and city authorities, in built-up urban areas where pluvial (surface water) flooding is increasingly a growing threat affecting local government and communities.

FloodCitiSense uses the urban living lab methodology that involves citizens as co-creators in developing solutions. In the case of FloodCitiSense this has involved citizens, local authorities and agencies in three pilot cities: Birmingham (UK), Brussels (Belgium) and Rotterdam (Netherlands). The three cities have been applying the same methodology to understand differences and any lessons learned that could apply to other urban areas. In the UK the pilot has looked at a specific area in Birmingham called Selly Park South that has experienced flooding first hand. The City Council has encouraged the formation of Flood Action Groups (FAGs) across the City, which include a dedicated group of citizens working with the City Council and contractors to help in flood alert and prevention. The relationship between the FAGs helps the City Council know when to issue sandbags and flood warnings to an area and informs the Council how severe potential flooding may be.

With active citizens, the city council, water company and other agencies, including the Environment Agency, they have been developing a mobile phone app, a cheap rain sensor (approx €125 each) and a website where the rain sensors data can be viewed and analysed. The project has gone through many stages in developing a prototype rain sensor and mobile app that is available to download from the app stores for both iPhones and Android smartphones.

A rain sensor on the LGiU’s own roof.

Two years in, we are now testing a second-generation rain sensor no bigger than a small medicine bottle, which has its own solar panel to keep it charged up to record rainfall and transmit its data every 5 minutes using a new wireless technology called LoRa, which is a technology that is used for the Internet of Things (IoTs) eg washing machines and fridges. Linking with Birmingham City University that has a department dedicated to LoRa, we are connecting the FloodCitiSense rain sensors to LoRaWAN Gateways in the city, as part of a global open source forum – The Things Network – that encourages the development of a LoRaWAN across the world.

At a recent workshop participants assembled their own rain sensors to install at home or work to form a local network of sensors that aim to provide early warning of potential flooding in the community to help the City Council gauge where flooding may occur. With the Council’s own flood model, more than 20 years of historical data and the use of A.I to improve predictions of flooding, the FloodCitiSense rain sensor is being tested to see if it could form part of a more reliable early warning service.

In parallel to the rain sensor, local citizens and agencies involved in FloodCitiSense have also been co-creating a mobile phone app that allows users to report a flood by taking a photo and submitting a simple report using icons to make the process quick and easy. There is also the ability to complete a more advanced report afterwards. The initial intention for the app is to allow people to send pictures and reports to the local authority. It is also envisaged that other agencies and active citizens will be able to view data from rain sensors on the app. Currently, the rain sensor data is available to view on an open-source data website developed by FloodCitiSense scientists involved in the project.

Although FloodCitiSense is in no way a silver bullet, this social science project involving new technologies could form part of a solution that helps local authorities and citizens understand where flooding may occur. Involving communities, academia, local authorities and other agencies in this project has also shown how good partnership working could help tackle some of the big local issues affecting us all.

The impact of the floods in Yorkshire, although fluvial (river flooding) in origin, brings home how more advanced warning may help local government, agencies and communities react more effectively to the threat of flooding in the future.

Barry O’Brien is LGiU’s Learning and Development Co-ordinator.


Leave a Reply

Your email address will not be published. Required fields are marked *