How do you feel about this article? Help us to provide better content for you.
Thank you! Your feedback has been received.
There was a problem submitting your feedback, please try again later.
What do you think of this article?
The Problem: “I am a recipient of almost monthly ‘false alarms’ from the Environmental Agency that ‘flood is possible’ near my area. Can we do better with the help of citizen science?”
A Possible Solution: Flood Alert - An exploration of ways to combine global meteorological data, with individual street data to give more accurate, ‘hyper-local’ warnings of impending flood risk.
Flood Alert V1.0 - Pilot Recruitment Video
Aim:
This project explores developing an Internet of Things (IoT) device to provide localised or “hyper-local” flood warnings. Creative Technologist Jude Pullen, and electronics expert Pete Milne were commissioned by RS DesignSpark, to understand the causes of urban flooding, the limitations of existing warning systems, and the potential of citizen science in improving flood risk management.
Urban Flooding: A Man-Made Problem:
Urbanisation, particularly the increase in impermeable surfaces like concrete and tarmac, significantly contributes to flooding by preventing rainwater from soaking into the ground.
"if we Concrete and Tarmac over green spaces - when rain falls it does not soak into the soil or become absorbed by trees at a sufficient rate to neutralise its effect."
Inadequate wastewater treatment infrastructure only exacerbates the problem.
Limitations of Existing Warning Systems:
Current Flood Warning/Alert API systems from Met Office / Environmental Agency are laudable but primarily focus on river levels (“fluvial” data), neglecting surface water flooding (“pluvial” data), a major issue in urban areas.
Existing data is often too broad, leading to inaccurate warnings and "alert fatigue."
"the Environmental Agency can email me [in Walthamstow, London], that my nearest flood risk - the Lee Valley River and tributaries are experiencing high flow due to heavy rainfall. However, from my and other local resident’s experience this is often a ‘false alarm’ as it does not cause my particular street any issue."
Navigating the responsibilities of different agencies (Councils vs. Water Authorities) is confusing and frustrating for residents:
"The first mistake you’ve made is calling it a ‘Drain’. It’s actually called a ‘Gully’... so like me, if you don’t know the difference, you’re going to go through call-centre hell numerous times."
Even small-scale flooding events ("micro flash floods") can have significant consequences for residents, highlighting the need for localised and timely warnings.
As described by David Harding, from Thames Water, "first 6 millimetres" of rainfall are critical in urban areas, emphasising the need to manage initial surges effectively through clear drainage systems and permeable surfaces.
"All these factors mean that much of the surface water devastation is often attributed to the first 6 millimetres of rain that fall. If this can be managed ... then the ‘surge’ which often suddenly floods homes, can be significantly reduced."
Public engagement with flood risk is often hindered by a lack of clear information, complex bureaucratic processes, and a lack of trust in existing systems.
Citizen Science and Hyperlocal Data:
It is hoped that the Flood Alert will in time allow for the gathering of "hyper-local" data on surface water levels, blockages, and problematic weather patterns (e.g. heavy rain after a very dry spell) - all of which can significantly improve the accuracy of flood warnings and inform proactive mitigation efforts.
The first iteration, the project thus far, of the Flood Alert device currently aims to establish the ‘benchmark’ with the Gov.uk API data, and aims to educate residents, whilst also quantifying the ‘alert fatigue’ issue, to measure any future AI improvements against.
The second iteration (about to be launched) aims to empower residents to collect this data and advocate for improved infrastructure and maintenance. This is where Artificial Intelligence - specifically Machine Learning - is hoped to be used to examine complex patterns from multiple data sources. It should be noted this is a ‘fine-turning’ of the Government data, by combining it with local user data to give street-by-street level accuracy.
"In short, we ideally needed a combination of 3 layers of data: a. Met Office’s Weather Forecast Data. Combined with b. Environmental Agency’s River (Fluvia) Data. Plus… c. ‘Hyper-Local’ Surface Water (Pluvial) Data."
Machine Learning (AI) and The Power of Open Source Collaboration:
The project embraces open-source principles, encouraging others to build upon and adapt the Flood Alert design and code. Collaboration with technically-minded individuals and flood-affected communities is crucial for refining the device and maximising its impact.
Future development will explore integrating machine learning to analyse environmental data and user feedback ("votes" on alert accuracy), further enhancing the predictive capabilities of the Flood Alert device.
"On a supervised level, we are telling the AI ‘yes’ or ‘no’ on its ‘guess’ derived from the Environmental Agency API. However, where it gets interesting is to consider how this ‘vote’ also collects some supplementary data, in the form of a snapshot of what the weather is like and has been like the past week for example."
Conclusion:
The Flood Alert project aims at using innovative solutions to address the growing challenge of urban flooding. By empowering citizens with localised data and fostering collaboration between communities and experts, this open-source initiative has the potential to improve flood preparedness and mitigate the devastating impacts of this often-overlooked threat.
TL;DR:
I’ve been enjoying using NotebookML lately. I have not used it to write this project up but have fed my 23,000-word write-up into it, to see how it fared when producing an ‘AI Generated Podcast” and the results are astounding - at least at the time or writing in Sept 2024! In short, if you need a quick overview, here is 2 years worth of work in just under 15 mins, according to NotebookML.
Above: AI Generated Podcast on Flood Alert.
Blog Series Contents:
Prologue - The Case for 'Hyper-Localisation' of Civic Data
Research & Development:
Part 1: Filling the Local Data Gap
Part 2: Civic Services & User Experience Research
Part 3: Ideation of Flood Alert Concept
Part 4: Prototyping Back-Story
Part 5: Citizen Science Learnings
Open Source Build Guide:
Part 6: Build Guide for 3D Printed Assembly
Part 7: DIY Decals for 3D Prints
Part 8: Code & Data Guide
Future Ambitions:
Part 9: Project Reboot with Machine Learning
Interviews:
- Heather Shepherd (Flood Forum) - on Trauma of Flooding - At End of Part 1. (Video)
- David Harding (Thames Water) - on Technical Details of Infrastructure - At End of Part 2. (Video Pt1 / Video Pt 2)
- Edward Flaherty (Flood Forum) - on User Experience Design - At end of Part 3. (Video)
Comments