11 February 2014

Floods happen.   Rivers have always flooded their flood plains or, in the case of the Somerset levels, the ancient marshland of Sedgemoor and the Levels, once an inlet of the sea that has been reclaimed by drainage.  The very word Somerset derives from the Old English for land of the summer people: the land was too wet to use in winter.   In 1607 floods associated with a major storm surge killed about 2000 people in Somerset and the Bristol Channel, probably the greatest loss of life from a weather event in UK history.  In the Great flood of 1872 over 100 square miles of land were flooded from October to March the next year.  In 1919, another major storm surge contaminated a similar area with sea water.   This month the figures suggest that parts of England have had the wettest January in the historical record, but in Somerset and the South-West it is only the fifth wettest.    This followed a wet December, and a previous winter in 2012-2103 when parts of the Levels were under water for more than 100 days between October and December.


Thus the current spell of flooding is not unprecedented but raises two obvious questions.   One is whether the impacts of the 2012/13 event could have been reduced; the second is whether we should expect more frequent flooding in the future.  The answer to the first question is undoubtedly yes.   But at a cost.    We most certainly can spend more on dredging channels, we most certainly can raise flood defences, given the political will to do so.   This is evident in the Netherlands, where standards of both coastal and river flood defences are much higher, and much much more costly.   The reasons for this difference in willingness to spend are readily apparent.   Much of the population of the Netherlands lives on land below sea level, and when the defences failed in the great storm surge of 1953 over 1200 people died.   Defences failed in East Anglia in 1953 too, and some 350 people died.   One of the biggest storm surges since 1953 occurred in the December event in 2013.   There were 26 breaches of the sea defences in East Anglia, but warning systems were in place, communities at risk were evacuated, 800000 properties were protected (according to the Environment Secretary), the Thames Barrier was raised to counter the highest water level since it was built, and nobody died.

So how much are we prepared to spend on flood defences?    And if the frequency of flooding is increasing in the future, as seems to be the general consensus amongst climate modellers, then how much do we need to spend on improving the current flood defences?  It does appear as if increased spending is easier to justify during or after an event.   New dredging of the Somerset rivers has recently announced by the current Government; while after the Carlisle flood in 2005, the flood defences were redesigned to a higher level.

The Carlisle case is particularly interesting.   When the flood occurred, plans for new flood defences were being displayed by the Environment Agency for public consultation.   These were intended to protect against the flood that has a 1% risk of being exceeded in any one year (sometimes called the event with a return period of 100 years on average.  The flood however was a bigger event.   This  is not unexpected, floods happen and bigger floods should be expected to occur, albeit with a smaller probability of being exceeded.   The Carlisle event was estimated to have a flood peak that might occur on in 170 years on average.  Thus, even if the new defences had been build the flood would have been overtopped and, in any case, part of the flooding was due to rainfall on the urban area, not due to the river flooding its flood plain.   Two of the fatalities occurred when trying to unblock a rainfall storm drain, not because of the river flooding.   So after the event, the new defences were redesigned to cope with a larger event with a return period of once in 200 years, with a correspondingly larger cost.    As it happened, these new defences were finished just one week before the November 2009 event that affected Keswick, Cockermouth, Workington and other Cumbrian communities.  But not Carlisle.   The water came within a few cm of overtopping the new defences but did not cause flooding, so Carlisle did not make the news.   The older design would not have prevented flooding in the 2009 event.   Does this suggest that it would be worth raising the defences everywhere to offer greater protection?

Perhaps, if the consequences of a flood event might be sufficiently catastrophic.   This would be the case in London, for example, where the levels of protection are much higher, justified by the huge costs that would be incurred if central London did get flooded.   But even there, protection is not complete.   A larger event that combined a flood event in the Thames, a very low atmospheric pressure causing a storm surge and a high spring tide could come along, though the estimation of the joint probability of such an event would be very small (and highly uncertain).  Elsewhere, the effects of flooding can be devastating on those directly affected, even if there are no fatalities, but the damages might not warrant a huge investment in defences, particularly if the losses that might be occurred can be offset or shared by insurance.  This is the balance that must be made. It depends on our estimates of the risk, as a combination of the probability of an event and of the damages such an event might cause.   Both are uncertain since generally we have less than 100 years of data to estimate the frequency of the larger and rarer events.  

It may also be the case that the frequencies are changing.    An increased frequency of flooding would be consistent with projections of climate change and expectations of the impacts of more urbanisation and intensive use of the soil.   But it is vary difficult to provide robust evidence for such impacts.  The data series are not long enough or not good enough.   There have been modelling studies that have suggested that the flooding in 2007 would not have been as intense without the increase in global greenhouse gases, but climate models do not, as yet, do that good a job of predicting rainfalls so it is again difficult to decide how much weight to give to such evidence.  The models certainly cannot yet provide robust estimates of how flood frequencies might change.

And in a sense it does not matter.   Floods will happen anyway.  There were major events on the Thames and Severn in 1924, 1947 and 1960, as well as in 2000, 2007 and 2013.  The decision that needs to be made, without good information about the frequencies of events that are relatively rare in any one location, is how far should people living in flood risk areas expect to be protected.   In the past, including in Somerset, low lying flood risk areas were avoided for permanent habitation.   Flood plains were left as water meadows, and crops were more likely to be planted in spring time, and were therefore less likely to be impacted by winter flooding.   

With nothing to be damaged, there was little impact.  In 1954, the UK’s record highest annual rainfall was recorded at Sprinkling Tarn in the Lake District (close to Seathwaite where a new 24 hour record of 314 mm was set in the Cockermouth event of 2009).  The total that year was over 6500 mm, the equivalent therefore of 21 Cockermouth floods.   In one year.   A trawl through the Cumberland and Westmoreland Gazette for 1954 revealed some discussion of the fact that it did not stop raining, but little in reports of damages.  There was little on the flood plains to be damaged, even if the rainfall was clearly less intense since the centre of Cockermouth did not get flooded to the same extent as in 2009.

Defences change that attitude.   It was somewhat depressing to hear a radio interview with a member of a local flood group a few months ago who said, when a flood defence scheme in the area had been completed, how good it was to see the smiles on people’s faces ”now that they were protected from flooding”.   They are not.   They are less likely to be flooded by moderate flood events but a flood that exceeds the capacity of the flood defences might come along this year, or next year, with perhaps a 1% chance in any one year, depending on the design.  So the aim should be to reduce the impact of a flood when it does happen, by not putting more at risk behind the defences.   Unfortunately, defences do not eliminate development on flood plains, they more often act to encourage it, thereby increasing the risk when a more extreme event happens.   In this way, while floods may (perhaps) not be becoming more frequent, the damages from floods continues to rise rapidly even with the major investment made in flood defences. 

It does not then seem to make any sense to invest more in defences if in the longer term that might lead to increased damages.   The most fundamental question therefore is why society as a whole should pay to protect the 1 in 6 households in areas assessed as being at risk of flooding, many of which have been built in the last decade, and some of which have been flooded within months of people moving in (at Kendal and Ruthin, for example).   Under a new agreement between the Insurance Industry and the Government as from 2015 everyone with a household insurance policy will do so through a levy.   In addition to this, hundreds of millions of pounds are allocated every year to flood management through general taxation, but even then proposals for new defences outweigh the money available every year, with only those considered the most cost effective going ahead.   

In fact, since the 2000 floods there has been a change within the flood risk management community towards policies aimed at mitigating the risk rather than trying to build hard defences aimed at protection.   Strategies that provide storage for flood waters in areas where damage will be minimised, land management to reduce runoff generation, restoring more natural river channels, flood proofing of properties at risk, and sustainable urban drainage schemes are being implemented as ways of mitigating a risk that cannot be entirely eliminated.    These are a more “natural” ways of dealing with an excess of water, but might leave some properties and land less well protected or dependent on paying for flood proofing.    Perhaps if people who wish to live in flood risk areas were required to install of flood proofing measures before moving in then it might focus attention on the fact that floods happen and protection can only be partial.

But could it be the case that such policies might increase the risk in some cases?   It has been suggested that in Somerset, the lack of dredging of the rivers in Somerset has exacerbated the problems of flooding in 2012/2013 and 2013/2014 by a gradual reduction in the capacity of the channels to move water to the sea.   These are not natural channels in this case.  They have been straightened and embanked as drainage channels by man. Some, such as the North and South Drain do not drain directly to the sea, but are dependent on being emptied by pumping, while gates to the tidal rivers, such as the Huntspill Sluice, only allow water out during low tide periods.   Getting water away from low lying areas is difficult and protecting those areas is expensive.  Once river embankments have been overtopped, of course, then the water will not drain away; it has to be pumped back into the channels together with all the rain that has fallen on the protected area.  At the time of writing the Environment Agency has 62 pumps in Somerset pumping some 15 m3/s of water off the land behind the flood embankments. 

It has been claimed that the Agency has neglected the Levels for more than 20 years (although this is actually longer than it has been in existence, having been formed from the old regional Water Authorities only in 1995) in lack of maintenance of the channels, pumps and control structures.  It seems clear that this rural area has not been a priority for expenditure relative to other areas.    This will have been exacerbated by the current cuts in the budget from the Government.   Unless there is a sudden change in priorities this Government will have cut spend on channel maintenance by 45% by 2015.   Groups such as the Floods on the Levels Action Group and the National Farmers Union are suggesting that this is a false economy when the costs of emergency pumping, disruption to the local population and need to produce food are taken into account (although one cost benefit analysis suggests that dredging would perhaps be a break even scheme at best).

Results of modelling studies commissioned by the Environment Agency after the 2012 winter floods has suggested that dredging would have significantly reduced the period of flooding in parts of the levels from months to days (although having seen only some output graphs I cannot comment on the assumptions made about the boundary conditions).   Dredging is not always a good solution.  In some places it might increase the risk to bridges, or to areas downstream, and in such a low lying tidal areas dredging can allow the tidal signal to propagate further and faster up the channels, causing water to back up at high tide and increase the variability of water levels.   This is one reason why flood defences in London have had to be raised in the 20th Century following dredging in the estuary.  

The Government has, in any case, now promised that the Somerset rivers will be dredged as soon as water levels fall to make it feasible.  This, as with redesigning the Carlisle flood defences after the Carlisle flood, must be considered a political decision that takes account of the misery of the people affected and the media coverage that has resulted.   Successes in flood risk management when areas do not flood are not nearly so newsworthy.  However, such decisions beg the question as to whether, once an area appears to be better protected, there will be more investment and more potential for damage in future events.  It does not appear as if we are managing the potential for future flooding that well, particularly in communicating to people in flood risk areas that they cannot be completely protected and that, even if flood defences are in place, they should prepare for a flood to happen.

Keith Beven

Lancaster University