Flood management Abstract Flood engineering has a long history. For example, evidence of engineering dates back 4,000 years in the Middle East and China. The aim of flood defence is to reduce the risk of flooding. Two broad categories of flood defence exist: Hard engineering structures, such as dams, levées and channel modification. Soft engineering techniques, such as afforestation, terracing of slopes and the re-creation of floodplains. Hard engineering structures are often point-specific, whereas soft structures are more extensive in area. Introduction As early as 1864 George Perkins Marsh highlighted the problem of flash floods and accelerated soil erosion in the Alps caused by deforestation. Deforestation in lower Provence, France, in the fifteenth to seventeenth century had resulted in erosion of good farmland and the deposition of coarse load on top of it. Today, there is growing awareness that methods of flood engineering should have an environmentally sensitive approach, linking the channel and the drainage basin/floodplain, and not just dealing with the channel. Hard engineering options Channelising has taken place on 25% of the main rivers in England and Wales. Channel widening increases the channel capacity but may lead to increased siltation. In urban areas, land may not be available for widening. Channel deepening can be achieved through dredging or narrowing the channel. It may, however, lead to increased bank instability. Channel straightening removes meanders in rivers, steepening the gradient and increasing velocity, and therefore reducing the flood stage. In contrast, sinuous channels have low channel efficiency. The most famous straightening was that of the Mississippi, where 16 cut-offs reduced the length of the river by 240km. Flood relief channels are used to by-pass existing urban areas where straightening is not an option (Figure 1), and are used as flood overflow channels. Large concrete channels, such as at the Thames in London, are expensive to build, damage the river ecology and are visually unattractive, especially at low flow. Bank protection may be provided by concrete wooden piles or groynes, or even by vegetation. Willow trees are particularly fast growing and take in large volumes of water. Figure 1 – Relief channel. Levées increase the carrying capacity of the channel but may increase the risk of flooding downstream. They may also increase sedimentation in the channel after the flood peaks have passed. In extreme conditions the level of the riverbed may rise so much that the riverbed level is higher than the floodplain. Levées are expensive, so are used to protect high-value property and vulnerable communities. Levée position is important. If placed on the edge of the channel, it is more likely to be breached. If placed in the floodplain, the amount of water that can be stored is greatly increased. Levées can fail for a number of reasons: Overtopping Piping Saturation flow between the base of the levée and the ground Under seepage. Research in Scotland has shown that levées on the outside bend of the river and overlying old river courses are most vulnerable to erosion. In addition, small-scale scour around posts at the top of the levée can destroy sections of levée. Moreover, the construction of new levées upstream may have impacts on ones downstream and destabilise them. Levées need to be maintained, but in countries experiencing rapid social change, such as LEDCs, such maintenance might not happen. In Vietnam, however, people used to be required to devote 30 days of labour per year to flood mitigation in their local area. That requirement is now down to 10 days. Various types of dam can be used to control and regulate water flow: Retarding dams are simple structures that have an outlet at the base, allowing water to drain out at a set rate. They cannot hold back water permanently. Detention basins are used to store water in times of floods. They are used for farming (pastoral) and other low-intensity purposes, such as recreation, but not for buildings or other valuable purposes. Storage dams have gates or valves to store and regulate water flow. These dams may be part of a multi-purpose scheme, such as the Three Gorges Dam on the Yangtze. Such schemes may need to keep the dam quite full for a variety of uses, such as HEP or irrigation. Single-purpose dams, used for flood control, can empty the dam of water following a flood and thereby be fully prepared to receive floodwater from the next flood. The location of a dam is important – it needs to be located near to the area to be protected. Some of the disadvantages of dams include dam failure and flooding of upstream locations. In addition, the storage capacity of a dam decreases with time, as sedimentation fills the area behind it. With a reducing capacity rate of 0.5% p.a., 25% of the storage is lost in 50 years. In central India, the Nizamsager Reservoir lost over 60% of its capacity in 40 years, and has been predicted to lose the rest of its storage capacity within a further 30 years. Environmental and ecological impacts of hard engineering Natural rivers have highly variable flow regimes with a variety of pools, riffles, meanders, innercliffs and slip-off slopes (Figure 2). River engineering produces a more uniform condition and a consequent reduction in biodiversity (Figure 3). Figure 2 – Meandering section of the River Cole. Figure 3 – Straight section of the River Cole. A historical pattern has been observed in the engineering of European and Asian rivers. Levées are built at first, increasing downstream floods and reducing the number of low flows. Later the construction of dams leads to decreased floods and more low flows. This can lead to channel aggradation (deposition) and degradation (erosion). Aggradation occurs as a result of erosion in the drainage basin. The riverbed, trapped between the levées, rises above the surrounding floodplain. In contrast, with increased sedimentation behind the dam following its construction, downstream ‘clearwater’ erosion occurs. In recent decades there has been increased awareness of allowing a river to interact with its floodplain. ‘River restoration’ refers to the return of rivers to a natural state, complete with meanders and floodplains. ‘Rehabilitation’, in contrast, refers to a partial improvement in river quality. Remediation leads to an improvement in the ecology of a stream, although the end result may not necessarily resemble the original stream. Soft engineering options Flood abatement tends to be focused on headwater streams (upper course) where a combination of high-intensity storms and farming practices increases the risk of flooding. High rainfall and snow melt also lead to increased soil erosion, thus any attempt at flood abatement also has an impact on reducing soil erosion. Flood abatement aims to tackle the flood problem as near to its source as possible. This means that a large area needs to be managed. In general, there is limited flood abatement in urban areas. Topographic manipulation Topographic manipulation involves altering the length of the slopes. A long slope produces more runoff and erosion than a shorter one, since water velocities increase downslope. Topographic manipulation techniques include: Terraces – artificial vertical layers on a slope designed to reduce water movement. They increase infiltration and sediment deposition. Contour ploughing – this involves the ploughing of surfaces along the contours to reduce overland run-off. Strip cropping – alternate strips of different crops are grown to produce variations in interception and run-off. Surface and underground water storage Much of the surface and underground water storage is natural, such as in small lakes and swamps, but some may be artificial such as ponds. Water spreading occurs mainly in arid and semi-arid areas. This is a process whereby surface flows from flash floods are diverted by dykes to increase infiltration and protect areas from flooding. Gully control In many steep mountainous areas dams are built to reduce the flow of water. These dams are cheap to build, often being made of loose rock. In flatter areas they can help to spread water. Vegetation cover management In general, a complete vegetation cover helps to reduce flooding through increased interception, increased evaporation and evapotranspiration, increased infiltration and reduced run-off. In addition to providing above-ground protection from rain-drop impact, vegetation adds organic matter to the soil and binds the soil particles together. Changes in flood characteristics occur after forestry. Following forest destruction by fire in the Yarrangobilly basin of the Australian Alps, peak flows increased from 60–80m3 sec-1 to 370m3 sec-1, whilst land cleared for tea cultivation in Kenya registered an increase in flows from 0.6m3 sec-1 to 27m3 sec-1. Afforestation should reduce peak flows through increased interception, evaporation, evapotranspiration and infiltration, and reduce overland run-off. However, in early years, the density of vegetation cover is low and so the impacts are not high. As the forest grows, there is increased interception, evaporation, evapotranspiration and infiltration, and so flood peaks are lower. Evidence from New York State has shown that reforestation of 58% of a watershed reduced winter and spring peak flows by between 16 and 66 %. Flood proofing Flood proofing can be achieved through a variety of structural and non-structural adjustments to the design, use or contribution of buildings. Raising properties above the flood level – known as elevation – is a common response, especially when the properties are detached. Examples include traditional long-houses in the Malaysian rainforest, and new properties in Maidenhead, England where the level of the door is above the 1947 flood limit. Dry flood proofing involves the sealing of a property so that flood waters cannot enter. It is only feasible for areas of low-level floods (<1m). Wet flood proofing allows flood water to enter buildings but cause minimal damage. Ground-level floors may be tiled or concreted and hooks in the ceiling hoist small items up when a flood is imminent. This tends not to be used widely in MEDCs. In a survey of 1500 homeowners in Louisiana, Illinois and Wisconsin (Figure 4), the majority of homeowners adopted some form of dry flood proofing. Most homeowners only used a method once they were flooded – preventative methods were not common. In addition, most used specialist contractors to do the work, suggesting that greater technical expertise would lead to better results. Figure 4 – Types of flood proofing retrofitted in Louisiana, Illinois and Wisconsin (survey of 1500 homes). Type Wet flood proofing Raised furnace/heater/appliances Raised wiring/fuse box Stopped using basement Sub-total Percentage 10 4 7 21 Dry flood proofing of a basement Glass-bricked windows Protected basement openings Installed sump pump Water-proofed basement walls Added dirt fill next to house Sub-total 4 6 16 11 11 50 Prevention of sewer back-up Installed back-up valve Installed standpipe or plug Sub-total 8 8 16 Protection from surface water Built wall around house Built levée or berm Improved drainage next to house Sub-total 2 2 9 13 Total percentage Number of responses 100 1538 Land use management Increased physical protection offered by structural defence has led to increased floodplain development. This is known as the ‘levée’ or ‘escalator’ effect. Studies of floodplains in England and Wales show that increased floodplain development following flood defence projects is widespread (Figure 5). Figure 5 – Profile of floodplain development trends in six urban locations in England and Wales. Estimated number of properties in the 1:100 floodplain Location River Thames, Maidenhead River Twyi, Carmarthen Date of completion of flood defence project 1960–66 1984 Nature of flood defence project Channel improvements to Maidenhead Ditch/Cut Flood wall improvements; new flood walls Flood defence design standard 1:10/20 Pre-project start (date) 1560 (1947) Post-project end (date) 3303 (1989) 1:100 0 476 Black Brook, Loughborough 1968 1979 River Tone, Taunton River Stort, Bishop’s Stortford 1969 1979 Channel realignment; embankments; embankment/drainage improvement Channel improvement; new weirs Channel improvements; new weirs 1:100 476 (1968) 672 (1988) 1:70/100 795 (1964) 912 (1989) 1:70 129 (1976) 210 (1989) Living with floods In the USA they believed, formerly, that hard engineering could deliver all of the answers. Comprehensive flood management recognises that this is not the case and that nature has a role to play. Even in the Mississippi, comprehensive planning is being developed (Figure 6) using a combination of levées, upstream reservoirs, land use planning, relocation of vulnerable premises, and turning frequently flooded areas into parks. It is proposed that key facilities and infrastructure should be located on higher ground and farmers in low-lying areas should convert to pastoral farming. All this would require a redirection of existing subsidies away from purely hard engineering to forms of soft engineering such as afforestation, wetlands restoration and wetland re-creation. Dwellings built on stilts represent another form of protection. Figure 6 – An artist’s impression of future comprehensive flood management around the Mississippi. Floods in Bangladesh The risk of flooding in Bangladesh is well known. Some 80% of the country is floodplain and 700 rivers make up a total river length of over 22,000km. The rivers carry about 2.5 x 10 9 tonnes of sediment, and most of the coastal region is at risk of flooding in the monsoon season. Moreover, the Bangladeshi population is very poor, with an average GNP of less than $500, the country has high rates of population growth (2.3% pa) and low levels of literacy (40–50%). By 1990 Bangladesh had over 8,000 hydraulic structures and 7500km of levées protecting approximately 2.5% of the land. These included: Levées among the main rivers, such as the 217km-long Bramaputra Right Embankment Freshwater polders – multi-purpose schemes designed for flood protection and farming Submersible levées – small levées to project the paddy rice in the pre-monsoon flood period Levées along small rivers – mainly in upland areas to protect against the floods Coastal-zone polders to prevent seawater intrusion and to protect against storm surges. In the floods of 1987 and 1988 these defences proved inadequate. For example, in 1987 over 1200km of levées failed, leading to losses of US$0.5 billion; in 1988 the figures were almost 2000km and US$1.3 billion. River engineering in Bangladesh will always be difficult since 90% of the water originates outside the country. The levées prevent the river from depositing in the floodplain, therefore the river deposits its load in the river bed. In addition, as the floodplain is not being replenished with sediment, and as sea levels rise, they will flood more of the floodplain with the result of the submergence of large areas of the country. Levées will also increase the flood risk downstream. Nevertheless, the scope for non-engineering schemes is limited, owing to the scarcity of flood-free land, and the widespread poverty and illiteracy. Land zoning is also difficult owing to the very high population densities. However, some alternative measures exist, such as: More raised flood shelters with health services, water and sanitation Improved storage facilities to store cooking equipment and tools Post-disaster employment for flood victims repairing levées More boats for use in emergencies Flood-resistant infrastructure Flood-resistant housing. Protection against flooding in Bangladesh is bound up with development. Poverty and illiteracy are major problems to be overcome. Only the literate population appear in favour of flood adjustment – the poor tend to be more fatalistic. In addition, it is the richer (less vulnerable) who are able to move to safer areas – the poor have limited choice but to remain. The notion of ‘living with floods’ seeks to gain benefits from the annual floods whilst reducing the risks during the hazardous events. Managing the flood risk in Oxford Flooding occurs in Oxford for a number of reasons: The rivers drain a large area. It is relatively flat and low lying. The bedrock is impermeable clay. Two main rivers converge in Oxford. Many bridges hold back water. Floodplain development has increased the amount of impermeable surface area. Agricultural practices have led to a reduction in tree cover in the catchment area and an increase in the amount of water flowing over the surface and into the rivers. The main ways in which the flood risk at Oxford has been managed has been through a mix of land-use zoning and flood-relief channels. In addition, levées, channel scour and straightening have all been used. However, according to the Environment Agency (EA) they might not be enough and there are major plans to overhaul and improve Oxford’s flood defences. Managing flooding in Oxford In December 2000 the Thames overspilled its banks and damaged 160 houses built on low-lying land. A similar flood in January 2003 damaged 250 homes (Figure 7). After years of study, the Environment Agency believes that the most effective solution is an 8km-long, 25m wide, flood relief channel. The £100m flood relief scheme could see Oxford bypassed by a water channel of similar size to the River Thames. A 25m-wide channel running from the Godstow Lock near Wolvercote to Sandford Lock south of the city is the centrepiece of a new flood defence strategy. The proposed channel would make full use of all the weirs, tributaries and streams as escape routes for floodwater. The scheme would also see the widening of the Bulstake, Hinksey, Seacourt and Weirs Mill streams (Figure 8). Figure 7 – Impacts of flooding on St George’s Meadow, which is part of Oxford’s floodplain. Figure 8 – Flood relief channels in Oxford. However, any scheme incorporating a flood relief channel through the western corridor has the potential for impacts upon the Oxford Meadows Special Area of Conservation and Iffley Meadows Site of Special Scientific Interest. The potential impacts tend to be greater to the north, where valuable instream and marginal habitats could be lost because of the widening works. However, a similar scheme at Maidenhead is said to have saved at least 1300 homes from flooding during periods of high risk. Construction is likely to begin in 2010. In addition, new pumps have been installed in a well at Osney Island. The pumps automatically switch on when the water level reaches a pre-set level, then water drains away from the well through three new gullies in Earl Street and a path between Earl Street and Duke Street. There is also a new filter drain along the path to help control the high water in times of flooding. The EA has considered other possible solutions, for example building above-ground reservoirs upstream of Oxford, but dismissed the idea as expensive and unworkable. Another option is to hold water back, not in reservoirs, but in wetlands and floodplains. For decades, the EA and its predecessors have been energetic drainers of farmland, with the channels of rivers and streams having been deepened and straightened. But, in this era of surplus farm produce, critics believe that instead of moving water downstream as fast as possible, the EA should try to hold it back in wetlands, and restore rivers to a more natural state, with their former meanders, rapids, riffles and gravel beds. Changes in land use to reduce run-off would offer major potential benefits that would be catchmentwide. However, the cost would be very high, and benefits would be achieved only over a long period of time. But this action ignores EU agricultural reforms that are looking to remove production subsidies and providing extra funds for agri-environmental programmes. Yet a small number of pioneering projects backed by the EA are showing the way forward. At Sherborne, Gloucestershire, the National Trust has restored extensive riverside water meadows, with the help of agri-environment payments from the Environmentally Sensitive Area scheme, while the EA has reinstated bends and spawning beds on adjacent stretches of the River Windrush. Likewise, on Otmoor, near Oxford, the RSPB has raised water levels on 267 hectares of formerly arable land purchased in 1997, and also created a 22-hectare reedbed reservoir with the help of the EA. The reserve now has nationally important numbers of teal, pintail and shoveler ducks, and dragonfly, and holds back upwards of half a million cubic metres of winter rain. Unfortunately, such initiatives have always been under-funded relative to hard-engineered drainage and flood relief schemes. Managing the Lavant In 1991, the government sanctioned the diversion of the River Lavant to prevent Chichester from being swamped by floods. The River Lavant, so insignificant in summer that it dries up, becomes a torrent in winter and devastates Chichester, West Sussex. A £4.7 million scheme to divert the river on a new 8km overflow route was approved by the government and the Environment Agency. It is part of £51 million worth of anti-flooding schemes. The Lavant caused the flooding of 750 homes, shops and factories and cut off the A27, the east-west main road, in 1993–94 (Figure 9). In 2001, the EA realised it was about to happen again and called in the army. They pumped water away and built emergency flood channels. Figure 9 – Map showing area flooded by the River Lavant in 1994. A combination of climate change and changed farming practices in the hills above Chichester, which makes the volume of water greater and the run-off faster, means that the EA estimates the flooding will occur once every 12 years if nothing is done. Although the scheme makes the river 9km longer, it will return it to the ancient riverbed that medieval people diverted it from. Flooding is now expected to occur only once in 75 years. Whether the original diversion was to beautify the city, flush the sewage away or provide a drinking water supply is not known. Either way, the river was prevented from following the river valleys down to Pagham Harbour and made to turn west through the city to Fishborne Creek. The problem of the current River Lavant in Chichester is that for part of its length the river is channelled into two culverts, which at times of peak flow cannot cope with the water. Therefore a grant of £3 million towards work was made by the government. Most of the money will be spent on building tunnels and bridges to take cars and pedestrians above the new flood river, which will operate as soon as the normal River Lavant flow is exceeded. Between Christmas Day 1993 and 10 January 1994 over 30cm of rain fell on Chichester. The result was a flow on the River Lavant four times larger than normal. Floods can last for many days because the river is fed by natural chalk aquifers, which overflow following heavy rain. Apart from the risk of the culverts not being able to take the volume of water, there is the potential for debris washed down in the flood to block them, leaving the water no escape, except through the crowded city centre. The new channel should overcome these problems. Conclusion Rivers are a vital part of modern society and there are increasing risks of flooding as more people live in vulnerable locations. Methods of river management are varied and may even increase the risk of flooding in some areas. The risks are not confined to LEDCs such as Bangladesh, but are just as important in local situations in MEDCs. The examples of Oxford and Chichester illustrate the local nature of the flood risk problem in the UK, compared with Bangladesh for example, and also just how expensive it is to manage rivers. Moreover, there are increasing calls for river engineering to adopt a holistic floodplain approach (comprehensive planning), which integrates methods of soft engineering with some of the more traditional hard engineering approaches.