Mobile-based Early Warning of Floods
Design a system to use data collected by mobile phone companies to measure rainfall, drive models to predict flood and warn people by SMS
EXPLAIN YOUR IDEA
Rainfall observations are very sparse for many parts of the world, with a poor distribution of rain gauges, a lack of surface-based rain radars and delays in processing satellite based rainfall data. The impact of these shortcomings are a lack of timely and accurate early warnings of floods at local scales. This impact is particularly acute for urban informal settlements that are often located in locations that were not built on by the original urban inhabitants because of their exposure to flooding.
Commercial cellular communication networks have a high density, particularly in urban and peri-urban areas. The electro-magnetic signals transmitted between antennas (base stations) within these network can be used to measure rainfall. Antennas within the cellular network are located in varying patterns with each base station having a range from less than 1km in cities to 8km in rural areas. Data routinely collected by mobile companies can therefore be used to measure rainfall virtually in real time. In turn these data can be used to drive hydrological models to predict flash flooding and warning messages distributed by SMS to those people at risk. Key to the success of such a scheme is community participation on delineating flood risk areas, on what to do when receiving a warning message and how to act appropriately with inherently uncertain forecasts.
Inhabitants of informal settlements and those charged with protecting communities and households at risk
Initially the project was conceived as applicable in regions throughout Africa where surface-based rain radar does not exist. A number of relationships exist with organisations in Kenya (Kenyan Red Cross) and national meteorological services, which means it might be quick to develop a trial there.
HOW DOES YOUR IDEA TAKE INTO ACCOUNT THE CONTEXT OF URBAN SLUMS AND CLIMATE CHANGE?
It takes into account climate change as current projections from the Intergovernmental Panel on Climate Change show an increase in intense rainfall events, and so flooding, with anthropogenic climate change. It is particularly focused on urban slums as these are often located in flood exposed locations (where other people have been reluctant to build), have unregulated building which can increase the risk of floods by blocking drainage systems and are populated often by people with little or no historical knowledge of local flooding risks. Lastly the informal nature of these settlements renders the issuing and acting upon of early warnings particularly problematic.
Yes, between one and two years
I’ve worked in a sector related to my idea for at least two years
TELL US A BIT ABOUT YOURSELF
We are an inter-disciplinary group of faculty and students at the University of Sussex, interested in applying our knowledge as anthropologists, sustainable development researchers, designers, engineers and information systems specialists to challenge inequality, poverty and marginalization.
IS THIS A NEW OR RECENT IDEA FOR YOU OR YOUR ORGANIZATION? HOW DOES IT DIFFER FROM WHAT YOU ARE ALREADY DOING?
Currently humanitarian organisations, governments, communities and individuals respond to urban flooding after the event rather than proactively. This is primarily because flood forecast information is not available to them at the resolution needed. In this sense it is a total change in disaster management.
HOW IS YOUR IDEA DIFFERENT FROM OTHER SIMILAR INITIATIVES? WHAT ARE YOU DOING DIFFERENTLY? WHAT UNIQUE ADVANTAGES DO YOU HAVE?
One other flood early warning system exists in Kenya, for the Nzoia river. This system relies on discharge measurements up river being used to forecast discharge and flooding further down river. Much of the catchment for this river is rurally situated and is notably larger than those affecting many urban areas. Flooding by contrast in Nairobi is much more ‘flashy’ responding quickly to high rainfall events. These flooding events have much shorter lead times and so require real time monitoring of the main precursor of the floods, rainfall. Measuring rainfall is notorious difficult in urban and peri-urban areas due to its high spatial and temporal variability. Our idea aims to address this shortcoming with a novel application of recent advances in rainfall measuring using the backhaul network, while the linkage with mobiles to disseminate information allows rapid and accessible warning.
WHAT ARE SOME OF YOUR UNANSWERED QUESTIONS ABOUT YOUR IDEA?
How does the proposed system integrate with the longer-term plans of the Kenyan government to move to proactive disaster management?
How willing would telecom companies be to allow long-term free access to the necessary data?
The feasibility of easily locating phone locations for receiving SMS flood alerts. This is not a problem with backhaul data which is fixed in location.
How can we extend early warning to those without access to mobiles?
WHY DO YOU THINK THE PROBLEM YOUR IDEA SOLVES FOR HASN'T BEEN SOLVED YET?
Until now a lack of rainfall data at high spatial and temporal resolution has prevented early warning systems for urban areas from being developed.
HOW HAS YOUR IDEA CHANGED BASED ON FEEDBACK FROM YOUR COMMUNITY?
We consulted with several humanitarian agencies in Kenya.
One of the fears they expressed was regarding the impact of false alarms on the credibility of the agencies. While the proposed system is monitoring based rather than forecast based, inevitably there will be some errors in rainfall measurements and routing of flood waters so issues of uncertainty and hence accountability of the system need to be communicated clearly.
They also expressed a desire for assistance to be tied to the warnings. At the moment this is not part of our idea, but we will be working on a user experience map incorporating this element. In addition, they felt that guidance for people on where to go if notified of an imminent flood. This would add significant complexity to our idea but is important from a human-centered design point.