Flash Floods, a UK perspective.
The flooding in Europe earlier this year was a tragedy, devastating for all those affected and the long term impacts they will have to deal with.
Shortly after the flooding, questions were raised over the flood warnings, in particular for Germany, their efficacy and why so many lives have sadly been lost. As Professor Hannah Cloke from the University of Reading stated;
“We should not be seeing this number of deaths from floods in 2021. It’s just unacceptable. There’s something going wrong with the system.”
What if we saw a destructive event of this type the UK? How would our warning systems fare?
A case of when….not if.
The destructive flooding experienced in July in Northern Europe could certainly happen in the UK. There have been a number of examples over the years which have resulted in damage and in the majority of cases, no lives lost. The Boscastle flood of 2004 is perhaps the most well known in recent times causing £15m of damage and sweeping cars out into the sea whilst the coastguard rescued those trapped from rooftops. There are many less well known ‘flash floods’, short in duration, destructive and localised. Here’s just a few examples from Northern England.
19 May 1989 Luddenden Valley, West Yorkshire
Luddenden village is situated in a steep sided valley in which Luddenden Brook drains down to the River Calder. On the 19 May 1989, 193mm of rainfall was recorded in two hours at Walshaw Dean reservoir which resulted in the brook level rising 3.5 metres in 20mins. Partly culverted the brook rapidly exceed the capacity removing, the gardens, sheds, vehicles, caravans and leaving a hole 20m wide, 80m long and 5m deep. Thankfully no lives were lost.
The residents believed a reservoir had failed however none were damaged and the reservoir storage could well have reduced the flood peak. Despite evidence of substantial rainfall on the moors above the valley, the recorded rainfall was discounted by the Met Office.
To put the rainfall (193mm in 2 hours) into perspective, large swaths of western Germany saw 24-hour rainfall totals of between 100 -150mm an hour. The Köln-Stammheim station, received 154mm of rain over a 24 hours period.
29 May 1944 Holmfirth, West Yorkshire
Holmfirth is situated some 6 miles from the town of Huddersfield in West Yorkshire. The flood of May 1944 is documented in a paper by Robert K Doe and Paul R Brown ‘A sea on the Moors’ which describes how the late afternoon storm falling west and southwest of Holmfirth with a 24-hour total of 114.3 mm caused significant damage and several people lost the lives in Holmfirth and Glossop which also suffered in the same event.
Doe and Brown describe the River Holme rising rapidly to four to five feet in a few minutes to become a torrent 80 feet wide and over 15 feet deep. Unconfirmed figures of 125–150 mm were reported locally. As a result, hundreds of homes were flooded, the area was covered in stone, clay, and other debris for months.
19 June 2005 Helmsley, North Yorkshire
The summer of 2005 saw a substantial rainfall event in the Helmsley area of North Yorks with 64mm recorded in one hour and nearby rainfall radar showing 125mm in 3 hours. Paul Wass, David Lindsay and Duncan Faulkner, wrote a paper Flash Flood! A lucky escape for 10,000 bikers. A very apt title as should the storm have arrived on the Saturday, rather than the Sunday, the temporary campsite a Duncome Park near Helmsley would have been occupied by over 10,000 people. The authors surmise that lives would have certainly have been lost in that scenario.
Michael Riley, who attended the event said “It’s a relief that this didn’t happen 24 hours earlier, as there were hundreds of bikers camping next to the river. One can only imagine what could have happened”.
What makes the Duncome Park example such a hazard is that flood warning services have been, to date, developed for property flooding. With no actual properties at risk in that area, there would have been no reason for the Environment Agency to create and manage a flood warning service for that area. Establishing a warning area would also require some form of detection to act as a trigger and ideally forecasting to provide the lead time needed. Even assuming all that was in place, the likelihood of a visitor such as Michael Riley travelling from Gloucester consciously subscribing to a flood warning service in advance of the festival is virtually nil.
Opt-in vs Opt out warning systems
Despite yearly flood awareness campaigns, apathy in signing up to warning systems is hard to overcome, even in areas which frequently experience flooding. In England and Wales, sign up rates to the flood warning service has averaged a 30% success rate. By comparison in Germany 20million Euros was spent on warning Apps with a result of just 8 million installs and only 10% coverage.
In contrast opt out systems where users are automatically subscribed have an exceptionally low exit rate. By working with BT for access to the emergency services database and mobile operators, the Environment Agency added over a million contacts to the flood warning system with an opt-out rate of just 0.6%. Natural Resources Wales whom currently share a warning system with the Environment Agency also followed this model.
Perhaps surprising the Scottish Environment Protection Agency has not adopted this approach meaning the number of people signed up to warning services in Scotland remains in the region of 25,000. Whilst subscriber numbers on a warning system might look great on the management KPI scorecard, the reality is that successful receipt and importantly action is something entirely different.
The reality of the numbers
Whilst the rate of change with peoples communications preferences and behaviour moves quickly, the wheels of change at a national level moves far slower. The result is flood risk management authorities sticking steadfast with providing telephone calls as a primary channel despite evidence that vast majority of flood warnings that are sent out via this channel are not received by the citizen.
With this being the case for both opt-in and opt out services. How successful are these at reaching recipients? In short, not very.
In the case of the Environment Agency, a total of 16,054 messages were sent between 8th February 2020 and 1st March 2020 inclusive. These include 25 individual severe flood warning messages and 1,121 individual flood warning messages. A total of 3,150,858 calls were sent — a call is an attempt by the flood warning system to contact a customer using a method or methods chosen by that customer. Each Customer may receive calls via a number of different methods.
The success rate of these telephone calls, is around 17%.
How do people want to be alerted?
When asked, the vast majority of those at risk of flooding want messages pushed to them as notifications without the need for self-registration. In 2020, as part of research into the neXt generation Warning System, the Environment Agency posted this question.
There is also an expectation of more intelligence from the warning service to alert them based on their location and only where the threat is severe and certain.
The adoption of cell broadcast technology via the national Emergency Alerts service is more in line with users expectations from a warning service. Reaching people based on their location and arguably targeting those at highest risk is more likely to reduce damage and risk to life. This technology also removes the constraints of only providing a flood warning service within the recognised flood plain and where there is property at risk.
Since the European Floods earlier this year Germany has announced it will be adopting cell broadcast technology and the UK Emergency Alert service is currently scheduled for launch in early 2022 (delayed from Summer 2021).
But what about detection and forecasting?
Traditional approaches to setting up a flood warning service would typically involve the installation of river level monitoring equipment and the development of forecasting models to provide advance notice or lead time. Of the three previous examples, only Holmfirth has a nearby river level monitoring gauge. It is of course highly unlikely that every catchment in all parts of the country would have the benefit of such infrastructure. Nevertheless, there are advancements in forecasting flash flooding with a degree of accuracy. The private sector is advancing this capability moving quicker than Government organisations in this space.
Aside from forecasting, local communities are often an effective route to raise the alarm, especially where individuals and groups already have established links to authorities and emergency planning departments. Many of these groups are are very pro-active in sharing on the ground intelligence via social media platforms.
Even a low/no lead time warnings have value especially when coupled with cell broadcast technology as alerts can be kept ‘playing’ to warn those who may be mobile and reduce the likelihood of individuals travelling into an area at risk.
What about direct alerting from weather warnings?
Routing rainfall forecasts through river level flood forecasts, human decision makers and separate alerting systems all shortens the critical notice period in emergencies. In other countries weather alerts are sent direct to individuals at risk.
Research into emergency alerts by Defra and Environment Agency in October 2019 found that close to 70% of respondents expected to receive emergency alerts from the Met Office. So why is this not the case in the UK?
The answer to this question is somewhat political. Direct alerting for flooding has been the domain of the Environment Agency (England), Scottish Environment Protection Agency (Scotland) and Natural Resources Wales. The Met Office stepping into this space would no doubt cause concerns in some quarters. It would highlight the inconsistency of alerting codes, colours and terminology that exists presently between these authorities.
The Met Office have also been rather cautious to truly embrace open data and standards in part due to a fear of weather warnings being re-used inappropriately by others. This was the same concern the Environment Agency had with flood warnings becoming open data and none of these fears materialised in reality.
The UK Emergency Alert service operates with the Common Alerting Protocol (CAP). With CAP now confirmed by the GDS open standards board as the UK standard for the exchange of emergency alert messages this should encourage authorities to adopt this to enable further re-use and exchange of alerts.
The Met Office are named together with the UK flood risk management authorities in the National Protocol for the Emergency Alerts service. This opens the door for direct alerting of weather warnings to citizens via smart phones.
Whilst the ability to send emergency alerts direct to citizens based on their location needs to be used appropriately, the sooner authorities adopt this technology the better. It is only a matter of time before the next disastrous flash flood in the UK.
