Edmund C. Penning-Rowsell, Nick Haigh, Sarah Lavery & Loraine McFadden

Transcription

1 A threatened world city: the benefits of protecting London from the sea Edmund C. Penning-Rowsell, Nick Haigh, Sarah Lavery & Loraine McFadden Natural Hazards Journal of the International Society for the Prevention and Mitigation of Natural Hazards ISSN X Volume 66 Number 3 Nat Hazards (2013) 66: DOI /s

2 Your article is protected by copyright and all rights are held exclusively by Springer Science+Business Media B.V.. This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your article, please use the accepted manuscript version for posting on your own website. You may further deposit the accepted manuscript version in any repository, provided it is only made publicly available 12 months after official publication or later and provided acknowledgement is given to the original source of publication and a link is inserted to the published article on Springer's website. The link must be accompanied by the following text: "The final publication is available at link.springer.com. 1 23

3 Nat Hazards (2013) 66: DOI /s ORIGINAL PAPER A threatened world city: the benefits of protecting London from the sea Edmund C. Penning-Rowsell Nick Haigh Sarah Lavery Loraine McFadden Received: 27 October 2011 / Accepted: 19 December 2011 / Published online: 29 August 2012 Ó Springer Science+Business Media B.V Abstract This paper describes the options appraisal undertaken in the UK within the major TE2100 project to investigate the future of protecting London from flooding from the sea. An economic analysis, within a Benefit-Cost framework complemented by Multi- Criteria Analysis, shows that improving the existing flood defences and, in 2070, constructing a new Thames Barrier downstream from the existing one are the front runner options for tackling the increase in flood risk that is anticipated in the future. Both sensitivity and scenario analysis have little effect on option choice. Uncertainties inevitably remain, however, when looking so far ahead, but it is clear that continuing to protect this area from the sea is highly cost-beneficial. Also the very high standard of protection now, and the robustness of the existing flood defence assets, mean that major new interventions will not be needed for some time (i.e. until c. 2070). We therefore have time to monitor the situation, carefully plan measures to maintain and enhance the existing defences, and to seek to restrain the growth of risk in the Estuary and in London through carefully designed and implemented resilience-building flood plain management measures. Rather than having to rush to new engineering works, because we have not anticipated what is needed but are forced to respond hastily to a crisis situation, the adaptive approach that is now possible is a key legacy of the TE2100 project. Keywords Storm surge Economics Multi-criteria analysis Flood risk management London E. C. Penning-Rowsell (&) L. McFadden Flood Hazard Research Centre, Middlesex University, Bramley Road, London N14 4YZ, UK Edmund@Penningrowsell.com N. Haigh Department of Environment, Food and Rural Affairs, London, UK S. Lavery Environment Agency, London, UK

4 1384 Nat Hazards (2013) 66: Context, history and our analytical framework 1.1 The wider international scene and the Thames estuary in context Estuaries are areas where flooding and flood damage is common worldwide, as a range of human activities are focused here to take advantage of a variety of physical attributes that make their locations attractive. These are the loci of valuable land for industrial and other developments, and ports here serve important hinterlands. Many have very long histories of human occupance. They are often therefore the subject of intense physical and socioeconomic dynamism (Townend 2004; Townend and Pethick 2002). The estuaries of the Mississippi, Hudson, Rhine, Elbe, Yangtze, Ulhas and the Pearl River and indeed most other major rivers of the world are notable in this respect, and each faces the threat of flooding, often exacerbated by both rising sea levels induced by climatic change and falling land levels resulting from human occupation and the pumping of groundwater that has developed. Both factors are key drivers of increased flood risk and the need over many decades for flood defence investment that this brings. The Thames is no exception. In addition to the daily tides, the Thames Estuary is prone to a significant increase in water levels caused by a North Sea surge, when an area of low pressure moves across the Atlantic towards the British Isles and reaches the relatively shallow parts of the North Sea just outside the Thames Estuary. In addition, strong northerly winds can increase the height of that surge. A surge tide entering the Thames Estuary can therefore increase water levels by over 1.00 m and can create a major flood threat, especially during a spring tide when normal tide levels are already high. The Thames tidal floodplain forms a corridor which passes through London and eastwards through North Kent and South Essex towards the North Sea (Fig. 1). In addition to the large number of people who live and work here, there are vital institutional and business centres and heritage sites. These include the Houses of Parliament, central and local government buildings, the Canary Wharf business district, the Tower of London and the National Theatre. There are also major transport links and numerous schools, hospitals, power stations and other key sites (Table 1). All are at risk of flooding from a surge tide, were the existing defences in the Estuary to fail or be overtopped, and Fig. 1 London and the Thames Estuary, showing the TE2100 project reach boundaries and the location of the Thames Barrier and possible new barrier locations

5 Nat Hazards (2013) 66: Table 1 Assets at risk from flooding in London and the Thames Estuary 350 sq km land area 55 sq km designated habitat sites 1.25 million residents (plus commuters, tourists and other visitors) Over 500,000 homes 40,000 commercial and industrial properties 200 billion current property value Key government buildings 400 Schools 16 Hospitals 8 Power stations More than 1,000 electricity substations 4 World heritage sites Art galleries and historic buildings 167 km of railway 35 underground railway stations 51 Rail stations (25 mainline, 25 DLR, 1 international) Over 300 km of roads all are at increasing risk as sea levels rise. Indeed this area was identified in the UK s Foresight Future Flooding project as one where future flood risk is probably the most serious in Britain (Evans et al. 2004a, b). 1.2 The history of flood threat and interventions Managing floods in the Thames Estuary is not a recent activity. There were tidal defences on the Thames Estuary more than 1,500 years ago. These defences protected Anglo-Saxon settlements in Kent and Essex. In the first century AD, the Romans built their city Londinium on the high ground which is today the square mile of the City of London; well above the five-metre contour level which approximately defines the area of tidal flood risk today. But the successful expansion of London over the centuries saw it cover the adjoining marshlands, and as sea levels rose relative to land, owing to eustatic adjustment resulting from glacial rebound, the challenge of maintaining tidal defence for our capital city was created. In the 17th century, Dutch engineers reclaimed Canvey Island and turned its three islands into one. Further upstream, with the construction of the docks, wharves and other urban development, large areas of marshland on either side of the Estuary were also reclaimed for a variety of uses including grazing marsh and agriculture. By the late 19th century, there was very little of the Thames Estuary which had not been modified in some way by human intervention. The network of tidal defences required constant attention to keep pace with rising sea levels, and the first of the London Flood Acts was passed following a series of damaging floods in London during the 19th century (Fig. 2). There was a major tidal flood in 1928 but an even worse catastrophe in 1953 (as in the Netherlands); both caused serious loss of life. This was the catalyst for the construction of the Thames Barrier and the associated defence improvements in the 1980s. The decision to build the Thames Barrier was taken on the advice of Sir Herman Bondi, Government Chief Scientific Advisor during the 1960s. He concluded then that I have no

6 1386 Nat Hazards (2013) 66: Fig. 2 Response to past floods: the successive raising of the river wall at Greenwich doubt whatever in my mind that such a major surge flood in London would be a disaster of the singular and immense kind It would be indeed a knock-out blow to the nerve centre of the country (Bondi 1967). This followed the Waverley Committee, which reported in 1954 after the 1953 east coast floods, recommending a dual approach of engineering structures backed up with a considered approach to development in the floodplain (Waverley 1954). Sir Herman Bondi s conclusions remain as valid today as they were over 40 years ago, and the evidence from Thames Barrier closures (Fig. 3) shows that without this intervention, London would have experienced serious flooding on several occasions since its completion in Our approach: economic efficiency and the TE2100 decision support framework (DSF) This paper sets out the approach taken within the Thames Estuary 2100 project and its resulting TE2100 Plan (Environment Agency 2010a) to appraise the estuary-wide options for flood risk management (FRM) from the point of view of the money values which society places on the costs and benefits of different approaches. The theoretical focus here is unashamedly on the economic analysis of long-term options, and a search for efficiency in this respect, not least because very large sums of public money may well be involved in continuing to protect London from the sea. Although the Plan was the subject to widespread and prolonged public consultation, to identify what might broadly be termed social issues and environmental concerns, value for money from the Plan remains the imperative.

7 Nat Hazards (2013) 66: Fig. 3 Thames Barrier closures since its implementation in The peak fluvial closure rate in 2003 was the result of more than 10 consecutive closures at a time of high river levels; the absence of such closures between 2004 and 2008 is simply a factor of the natural variability of such phenomena Timing is also important here. In the course of our analysis, different results have emerged for two distinct periods: up to 2050 and beyond. The year 2050 emerged as the dividing line between actions needed to sustain the existing FRM system, which now protects London to a high standard, and making decisions about significant enhancements which appraisal reveals are justified later in the planning period. The latter interventions, we conclude, will need to be operational in around 2070, based on Defra s (2006) FCD- PAG3 guidance on climate change (Defra 2006). Our analysis presents the estuary-wide level results of economic appraisal. Option appraisal within TE2100 was undertaken using an appraisal method developed in the TE2100 project, using both benefit-cost analysis (BCA) and multi-criteria analysis (MCA). This integrated method is consistent with Defra s FCDPAG3 guidance on economic appraisal (MAFF 1999) and, in turn, with the HM Treasury Green Book (HM Treasury 2003). The key features of the approach for option appraisal are The estimation of capital, maintenance and other (e.g. opportunity) costs associated with FRM options; The estimation of direct potential property flood damages arising under options (and different states of the world, e.g. associated with climate and socio-economic change and other uncertain variables); The estimation of other economic, social and environmental impacts of options (and different states of the world) through Multi-Criteria Analysis (scoring and weighting); Comparing benefits and costs for each option, based on estimated values of costs, direct potential property damages, and the values of wider economic, social and environmental impacts which are implied by the scoring and weighting exercise; Making recommendations for option choice, based on the balance of benefits and costs, according to the method set out in FCDPAG3, Chap 6 (generally known as the PAG decision process ) (MAFF 1999).

8 1388 Nat Hazards (2013) 66: The TE2100 DSF is generally made up of tried, tested and widely accepted analysis for the appraisal of FRM interventions. That said, the use of Multi-Criteria Analysis is fairly novel, particularly for a large-scale plan such as TE2100. However, the method used is that piloted at scheme level within the Environment Agency in 2008 and adopted as part of new Flood and Coastal Risk Management Appraisal Guidance (FCRM-AG (Environment Agency, 2010b)) the publication of which our analysis anticipated. 2 Designated policies and the required options 2.1 Policy designation In looking to evaluate the benefits of different FRM options, we first identified five possible strategic levels of FRM available to us and designated each of 23 geographical areas within the Estuary as deserving such a target policy (Table 2). The policies indicate an initial assessment of the level and standard of FRM which is justified by the number of people and the value of property and the other assets being protected in that area, as a strategic direction of FRM in each part of the Estuary. The social, economic and environmental value of each policy unit (i.e. area) was assessed through a formal process to allocate one of the FRM policies. The same process and policies are used throughout England and Wales, in the Environment Agency s Catchment Flood Management Plans (Environment Agency 2004), to ensure a nationally consistent approach to tackling flood risk and a level playing field when it comes to the allocation of scarce resources for FRM measures. 2.2 Key aspects of the approach Our approach to the appraisal of options (as tangible ways of delivering the assigned policies referred to above) using the TE2100 Decision Support Framework was based on the investigation of options of different degrees of complexity and cost (Table 3). However, the four options are all estimated to have identical benefits, as they have the same design standard (in terms of the return period of the events against which each, protects); so the choice here between them is purely on cost grounds. Option 1.4 was assessed as being the most cost-effective out of this group, and this therefore becomes the best option 1 referred to below, so we do not consider further here options Table 2 The flood risk management policies used by the UK s Environment Agency and applied to London and the Thames Estuary P1: No active intervention (including flood warning and maintenance). Continue to monitor and advise P2: Reduce existing FRM actions (accepting that flood risk will increase over time) P3: Continue with existing or alternative actions to manage and maintain flood risk at the current level (accepting that the likelihood and/or consequences of a flood will increase over time from this baseline) P4: Take further action to sustain the current scale of flood risk into the future (responding to potential increases in flood risk from urban development, land use change and climate change) P5: Take further action to reduce the risk of flooding (now and/or in the future)

9 Nat Hazards (2013) 66: Table 3 Options appraised for continuing to protect London from future flooding DM Do minimum 1.1 Improve existing defences (minimum maintenance, no adaptation a ) 1.2 Improve existing defences (minimum maintenance, with adaptation) 1.3 Improve existing defences (optimised maintenance, no adaptation) 1.4 Improve existing defences (optimised maintenance, with adaptation) 2 Best of the Option 1 group, with Tidal flood storage from Best of the Option 1 group, with new barrier at Tilbury from Best of the Option 1 group, with new barrier in Long Reach from Best of the Option 1 group, with new Barrier With Locks (BWL) at Tilbury from Best of the Option 1 group, with new BWL in Long Reach from Best of the Option 1 group, but converting the existing Thames Barrier to BWL from 2070 a Also called frontloading in some TE2100 appraisal documents the practice of making provision for works which may be required at a later time period (e.g. making foundations suitable for future raised defences) In terms of a baseline, the options appraisal described in this paper has been conducted relative to a walk away estuary-wide Policy P1. An estuary-wide P3 Policy has been taken as a business-as-usual do minimum (DM) option, for comparison with the do something options inclusive. For the sake of brevity, this paper does not consider the estimation of costs for use in appraisal, but detail of the tasks undertaken here is provided in a Technical Report (Environment Agency (2009), Chap. 7). Our assessment inevitably had to allow for sea level rise as a result of climate change. As is well known, the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (IPCC, 2007) forecasts sea level rise of m by , with an additional metres from potential ice sheet melt. The formal UK view has been encapsulated in the DEFRA06 recommendations (Fig. 4), projecting a rise of just Fig. 4 Sea level rise projections used in the TE2100 assessment

10 1390 Nat Hazards (2013) 66: over 1.0 m by 2105 or slightly more than the pre-existing projections. We used these DEFRA06 recommendations for our assessments. Option appraisal as described here has been carried out over an extended period to the year This is because we included several long-lived major responses within the options to be implemented at around 2070 (e.g. flood storage, new barriers, enhancements to the Thames Barrier, etc.), and these were judged to have broadly century-long lives in economic terms. To allow the consistent appraisal of all options, the appraisal period was therefore extended to allow the whole life of these responses to be represented. 2.3 Estimation of direct property damage and risk to life The benefits of protecting London from the sea involve the avoidance of massive future flood damage and other losses, not least as sea levels rise and existing defences are thereby rendered inadequate. Our potential flood damage estimation used a hydraulic flood model of the Thames Estuary and its floodplains based on the ISIS-TUFLOW modelling framework (Environment Agency 2008). The model, known as ECO?, includes property damage data (derived from the Flood Hazard Research Centre s Multicoloured Manual (Penning- Rowsell et al. 2005)) to estimate economic losses associated with future flooding and thereby avoided with the range of interventions we examined. The modelling also includes an approach to uncertainty based on probability distributions for the main input variables, particularly extreme sea levels. This flood modelling has represented probabilities of overtopping, breaching of fixed defences and the failure of major barriers (existing and new) under the various scenarios considered. The central case has assumed the completion of the major Thames Gateway urban development to the east of London (Office of the Deputy Prime Minister 2004); other development futures were explored as part of separate socio-economic scenario analysis (see below). The flood model also incorporates a formal estimation of risk to life, as per the 2008 Defra FCDPAG supplement methodology (Defra 2008), and included within the appraisal as a monetary value using the Department for Transport s current value of a prevented fatality (Department of Transport, 1994; see also NERA (2011)). But London s assets at risk are different from most UK towns and cities. Property damage estimates from the flood model have therefore been supplemented by, first, a separate study of likely direct damage to some key non-residential property sites in the Thames Estuary, which are known not be reflected appropriately within the National Property Dataset (NPD) used by the ECO? model. These sites include the Houses of Parliament, Canary Wharf, the Coryton Oil Refinery, City Airport, Kew Gardens, Tate Modern, Tilbury Power Station, Tilbury Docks and the Beckton Sewage Treatment works. In addition, secondly, to be as realistic as possible, the basic property flood damage values and fatality projections have been reduced to take account of likely floodplain management efforts in the future such as warning systems, building regulations and emergency response. In addition, all direct property damages have been inflated to account for the economic cost of Greenhouse Gas (GHG) emissions arising from repair and replacement activity, valued at the Shadow Price of Carbon as per Defra guidance (Defra 2009). This was informed by a short overview study which suggested an uplift factor of around 2%. The inclusion of this uplift ensures consistency with the cost side of the appraisal, where the costs of GHG emissions were also included. As well as estimates of direct property damage, our flood modelling framework provided outputs via colour-coded maps of the probability of flooding to given flood depth

11 Nat Hazards (2013) 66: thresholds. These Depth-Probability Grids enabled information to be extracted on at-risk receptors within the floodplain (houses; shops; factories, etc.). In turn, this information, in the form of numerical indicators, was used to help populate Appraisal Summary Tables (ASTs) to determine the wider economic, social and environmental impacts of each of the options in each of the 23 discreet appraisal areas (see below). 2.4 Multi-Criteria analysis: monetary estimates of other economic, social and environmental impacts Table 4 gives the full range of impact categories used here in our options appraisal. Those categories marked with an asterisk (*) there have been assessed directly in monetary terms, for inclusion in the final Benefit-Cost Analysis. Property and Risk to life impacts have been estimated using the flood modelling framework (see above). Agricultural land use, Navigation (i.e. the impacts on Port of London Authority operators), indirect impacts on business and Technical risk (the extra property damages from options not performing as intended) were estimated using monetary methods outside of the confines of the available hydraulic/economic modelling framework because they are less closely related to flood extent and depth. The remaining impact categories have been assessed using Multi-Criteria Analysis (scoring and weighting), in the absence of tractable methods of direct monetary valuation. The full method is detailed elsewhere (RPA 2005), but in essence it involves the following steps: 1. Step 1 Calculating quantitative indicators of impacts from the depth-probability grids. This involved, for example, counting the numbers of infrastructure and transport assets (electricity substations, schools, hospitals, railway lines, underground railway stations, landfill sites, historic sites, etc.), which are potentially affected by flooding at a given probability threshold at three possible flood depths (0.1, 0.6 and 2 m) and for three different key flood probabilities (referred to as p(x), p(y) and p(z) which differed across receptors). These counts and associated probabilities are then used to estimate Table 4 Categories of economic, environmental and social impacts assessed in the options appraisal described in this paper Economic Environmental Social Technical Property* Physical habitats Recreation c Technical risk* Key infrastructure and biodiversity Risk to life* Agricultural land use* Water quality and quantity Sense of community Navigation* Natural processes a Transport Other environmental b Indirect impacts on business* Landscape Historical environment * Primary monetary categories see text a Other than physical habitat provision, which is included in Physical habitats and biodiversity b This category is designed to include environmental impacts likely to be covered by the Water Framework Directive which are not included in other categories c Tourism is partially covered under social to the extent it is picked up here. It will also be reflected, again partially, in the historic environment category

12 1392 Nat Hazards (2013) 66: Annual Average Impacts (AAIs) at different epochs (time periods) using receptorspecific impact-probability curves. 2. Step 2 Using the AAI data to populate Appraisal Summary Tables (ASTs), where the impacts of each option are quantified and described. ASTs were developed for each option at a number of time-slices or epochs (Present Day, 2030, 2080, 2115 and 2170) and also completed for each of the four TE2100 reaches (West London, City Reach, Regeneration Reach, and Lower Estuary North and Lower Estuary South). 3. Step 3 Adding qualitative assessments of impacts to the ASTs to supplement the quantitative indicator data. For some impact categories (Physical habitats and biodiversity, Water quality and quantity and Natural processes), little or no quantified indicator data were available, and in these cases, qualitative assessments based on professional judgements from specialists within the TE2100 team formed the core of the analysis. 4. Step 4 Scoring the options according to their impacts identified in the ASTs, determining the weight which should be applied to the score in each impact area, and using the products of scores and weights to determine implied economic values for inclusion in the benefit-cost analysis (BCA). This scoring and weighting was done within the project team, relating both scores and weights to a comparison with those impacts for which monetary estimates were available, thereby pegging the weighted scores as much as possible to known monetary value evidence. 5. Step 5 Aggregating the implied values thereby derived ( m) across the epochs and reaches to gain a present value for the Estuary as a whole, for inclusion in the wider BCA. A disaggregation of the estuary-wide results to local levels was undertaken but is not reported here (see Environment Agency 2010a). 3 Benefit assessment results: the central case 3.1 Option impacts and benefits: direct monetary valuations The potential benefits of direct property flood damage avoided by all options are very large (Fig. 5). All of the do-something options have total property benefits in the region of bn (present value (PV), to 2170) and there is very little to choose between any of the main options in terms of direct property protection. Although, strictly, the barrier-withlocks options (BWL) ( inclusive) have the best performance here, as they are marginally less likely to fail in comparison with the standard barriers, the premium over the worst-performing other option (flood storage) is only some 1.5% in PV terms. The similarity in performance of options here is a measure of their success in delivering the strategic management policies determined at the earlier stage, although local results from the option appraisal were used to refine these policies (for example, if some policies were found not be justified once potential defence responses were considered). The benefits of fatality prevention and preventing the indirect impacts on business are, respectively, of medium and large scale: the former is scaled at c. 11bn and the latter at between 53bn and 58bn. Estimated numbers of fatalities under different scenarios have been valued using the Department for Transport estimate of the benefit of preventing a fatality, which is c. 1.4 m (at 2005 prices (Department for Transport, 2007)). Values of indirect impacts on business are estimates of potential disinvestment by the financial services sector in the face of different flood risk scenarios. These are based on estimates of leakage of Gross Value Added away from London and the UK to other

13 Nat Hazards (2013) 66: ,000 m 165, ,000 65,000 15,000 Sense of community Risk to life Historic environment Landscape Recreation Natural processes Other environmental (WFD) Water quality Physical habitats Indir. impacts on business Other transport Navigation Land use (agriculture) Key infrastructure Properties Total Present Value Benefits shown as white line P1 DM ,000 Fig. 5 Summary of option benefits ( m, PV to 2170, P1 ( Walk away = 0)) international centres, once the FRM standard of protection (in terms of return period) falls below a certain level. This is taken as 200 years, given our understanding of the threshold for commercial insurance cover. This assessment is very broad-brush, but the result does not affect the performance of the main options relative to each other (only relative to P1 and DM). Our estimates of the impacts of the different options on navigation and agriculture are low. In the case of agricultural impacts, a high-level estimate of damage was made in terms of loss of annual net farm income over the area concerned, based on an average estimate for Environment Agency s Thames Region of 415/ha in 2006/07. Impacts on navigation have proved very difficult to estimate definitively due to a lack of evidence on disruption costs, so our figures here are also broad brush. But two separate types of costs to navigation were estimated: first, the disruption costs arising from flooding, which makes wharves and docks unusable, and secondly, disruption costs arising from reduced access to certain river reaches during flood barrier operation, or from having to negotiate locks (as part of potential new barrier structures). Overall, all of the options are estimated to have some net benefits to navigation compared with P1 ( walk away ). Although the benefits of potential new barrier structures in the Tilbury area are reduced (because of disruption to terminals between there and Long Reach, through which over 40% of total Port of London Authority tonnage is handled), our assessment shows that this still implies net benefit over a P1 scenario. In other words, the costs of disruption because of increased flood risk to terminals under P1 exceed those from disruption owing to new barrier structures. 3.2 Option impacts and benefits: implied monetary valuations The value of impacts other than those to which a monetary valuation can be applied directly has been derived through Multi-Criteria Analysis (scoring and weighting: see above). The discussion here is limited to reporting the calculation of the present value aggregate implied value estimates, along with associated key caveats.

14 1394 Nat Hazards (2013) 66: The net total of MCA-derived ( implied ) benefits here is equivalent to about 65% of directly monetised benefits, and the implied multiplier to convert directly monetised benefits to total benefits is therefore around 1.6. The implied economic benefit estimates for Key infrastructure and transport (other than navigation) are very large (Fig. 5). And unlike for direct property damage, there is a reasonably clear distinction between the options in terms of protection to key infrastructure and transport. In particular, the barrier options ( inclusive) bestow around 5-6bn more benefit than options 1 and 2, mostly in relation to protecting transport assets such as the underground railway system in central London. Of the implied benefit estimates for environmental categories, the most dominant differentiating issue between the options is water quality, with Do Minimum (DM) and the BWL options having strong disbenefits (up to around 20 bn in present value terms). Although the water quality assessment includes impacts through the flooding of contaminated land (e.g. landfill sites), the vast majority of the disbenefit value for DM and the BWL options comes from physical and chemical factors in the channel in the absence of flooding and the free natural flow of the river that the barriers and their closures inevitably create. The BWL options fare badly here because the barriers are forecast to be closing on every tide towards the end of the appraisal period, whereas the other barrier options (including options 1 and 2 which retain the existing Thames Barrier) are closing less 50 times or fewer per annum in the period before 2135, albeit rising rapidly to about 400 times per annum by Options 3.1 and 3.2 are slightly worse in water quality terms than options 1 and 2, because the new Tilbury or Long Reach barriers would be downstream of major sewage works and power stations which would have a polluting effect in cases of frequent barrier closures and the impounding of water that they create. This is partially offset, however, by the greater volume of water enclosed upstream of the Long Reach and Tilbury structures, compared with the existing Thames Barrier, which has a net benefit to the City Reach. However, it is recognised that there is uncertainty surrounding the water quality assessment, as no formal modelling was carried out. Water quality issues are therefore the subject of sensitivity analysis (see below). The implied option benefit estimates for social categories (not including risk to life covered earlier) show that differentiation between the do-something options on social grounds (other than safety and security which is included in the directly-monetised categories) is limited to around 200 m in total PV terms. The barrier options ( inclusive) tend to perform marginally better than options 1 and 2, principally through the greater protection afforded to public buildings (which is the primary indicator of the sense of community variable). In summary, the Property, Key Infrastructure, Other Transport and Indirect impacts on business categories account for the vast majority of the FRM benefits we have calculated (c. 87% of the total positive benefits) (Figs. 5 and 6). The categories where the most differences between the options occur are Transport and Water Quality. The options with the highest overall net benefits are options 3.1 and 3.2, followed by options 1 and 2. Options are significantly disadvantaged by their environmental disbenefits. 4 Benefit-cost analysis (BCA) results The estimates of option costs, and directly monetised and implied benefits, were brought together into an overall analysis of the benefits and costs of the TE2100 options (including

15 Nat Hazards (2013) 66: ,000 10,000 5,000 m PV 0-5,000-10,000-15,000-20,000 DM Historic environment Landscape Natural processes Other environmental (WFD) Water quality Physical habitats -25,000-30,000-35,000 Fig. 6 Implied environmental benefits ( m, PV to 2170): see text DM as a do minimum ). Costs include the capital, maintenance and GHG costs of implementing the options, and the benefits are the avoidance of direct property and indirect business flood damage, plus wider economic, environmental and social benefits (relative to policy P1). Benefit-cost results have been produced in relation to a appraisal period. The appraisal is principally concerned with long-run (post-2050) actions; however, the worth of continuing activity in the interim (up to 2050) was also examined. Given the strong economic case for maintaining the current system in the short term, the pre-2050 analysis has been concerned with finding the most cost-effective way of doing this until more complex decisions regarding long-term management are required. The recommended approach is option 1.4, which involves optimising the balance between maintenance, repair and replacement of existing defences. This basic asset management approach has, however, also been embodied in the proposed post-2050 actions. 4.1 Use of incremental benefit-cost ratios and indicative standards Our approach has followed that set out in Defra FCDPAG3 guidance (MAFF 1999). This recommends the use of incremental benefit-cost ratios (IBCRs) to explore whether it is economically optimal and efficient to move to a more expensive option than suggested by average benefit-cost ratios (ABCRs) alone. The procedure for using IBCRs has to date been linked with the notion of indicative (flood defence) standards for different land use types. The argument is that given certain types of land are typically observed to have flood defence standards in particular ranges (e.g years for urban areas), then for cases where the ABCR does not lead to the choice of a scheme within that indicative standard range, it is legitimate to explore whether it is worthwhile, at the margin, to spend more money moving closer to that standard in terms of the increased benefits that are generated. How big the IBCR needs to be to justify

16 1396 Nat Hazards (2013) 66: a change of scheme depends on whether the starting point is below, within, or above the indicative standard range. We have used the current indicative standard guidance as it applies nationally, without making any claim that London should have a higher indicative standard, or is otherwise a special case. Although funding for future FRM measures on the Thames is not yet clear, for the time being we are assuming that Thames scheme expenditures would be drawn from the national FRM pot, and so would be competing with other schemes elsewhere in the UK. This means they should be considered using the same indicative standard and IBCR rules. Pursuing the Do Minimum (DM) option would mean the annual probability of flooding in the Thames Estuary would rise to an estimated 1% upriver of the Thames Barrier (under Defra 2006 climate change assumptions) and to an even higher probability elsewhere. On average, this means that the Estuary (as a whole) falls below the indicative standard range for an urban area at tidal flood risk (this range is from 1 to 0.3%). As such, moving to one of the do-something options (which all generally deliver at least the indicative standard) would be justified as long as the IBCR is robustly greater than unity (MAFF 1999). 4.2 Central benefit-cost results The benefit-cost analysis under central assumptions (consistent with all the results presented so far in this paper) is given in Table 5. The following conclusions are pertinent. First, the do-minimum option (DM) has a very large ratio of overall benefits to its costs, with every 1 of cost returning about 88 to society in a variety of economic, social and environmental benefits. Given that DM is a business as usual scenario, this provides strong justification for current FRM activity and expenditure on the Thames through London. Secondly, whilst DM has the highest average benefit-cost ratio (at 87.8 in Table 5), there is a strong marginal (incremental) case for moving beyond DM to one of the more complex options: the incremental benefit-cost ratios (IBCRs) show how much extra benefit can be obtained for every extra 1 spent moving to one of the other, more expensive, options. The options with the highest IBCRs of about 25 are options 3.2 and option 1.4 (Long Reach Barrier and Improve existing system, respectively), with option 3.2 having a small advantage (at 24.9). Given that the IBCR only needs to be robustly greater than unity, a move to option 3.2 would appear to be strongly justified. Thirdly, the case for moving beyond option 3.2 to a still more expensive approach is weaker. The highest IBCR for such a move (see boxed line in Table 5) is for option 3.1, with a value of 3.1. Although robustly greater than unity, it is important to remember that the previous move to option 3.2 has delivered not just the indicative standard, but the TE2100 Policies, which in many areas implies sustaining a standard of 0.1% annual flood probability or lower well in excess of the usual urban indicative standard. As such, the IBCR for going beyond option 3.2 really needs to be exceptional, and it is difficult to argue that a ratio of 3.1 meets this criterion. Furthermore, one key uncertainty relevant to option 3.1 is the assessment of navigation benefits which, as hinted earlier, probably underplays the merit of a Long Reach versus a Tilbury barrier. This implies that the IBCR for option 3.1 may be overstated. As such, the case for moving beyond option 3.2 to option 3.1 is further weakened. On this central analysis, option 3.2 (a new standard barrier at Long Reach) emerges as the preferred option for the Estuary as a whole, followed closely by option 1.4 (improve the existing system). These options would therefore appear to be front runners. However, this conclusion is subject to uncertainty analysis, below.

17 Nat Hazards (2013) 66: Table 5 Benefit-Cost analysis: Central assumptions, Defra climate change scenario ( m, Present Value to 2170) Option P1 DM Do nothing (walk away) Do minimum (maintain) Improve existingenhanced maintenance with adaptation Flood storage Tilbury barrier Long reach barrier Tilbury barrier with locks Long Reach barrier with locks Convert Thames barrier to barrier with locks Property damage 58,730 11,838 1,557 1,646 1,257 1, Other monetary disbenefits 69,678 5, (agriculture, navigation, life, business) Property damage? other monetary disbenefits Property and other monetary benefits, PVb1 (=P1 - option) 128,408 17,172 1,726 1,814 1,520 1,769 1, , , , , , , , ,419 Other benefits (implied 30,268 80,845 81,236 86,344 86,071 70,176 74,416 65,996 value), PVb2 Total benefits, PVb 141, , , , , , , ,415 Total costs, PVc 0 1,612 4,285 4,725 4,642 4,474 5,440 4,858 4,517 Average analysis (incremental analysis over P1) Average benefit cost ratio (=q/v) Incremental analysis over DM Incremental benefit cost ratio Incremental analysis over 3.2 Incremental benefit cost ratio Incremental analysis over 3.1 Incremental benefit cost ratio

18 1398 Nat Hazards (2013) 66: Option ranking Another way of showing the different costs and value of the different possible interventions is to rank options, based on the incremental benefit-cost ratios (Table 6). Here the shading indicates the cost ranking of the alternatives, with the darker shadings being the more expensive options. What this reveals is that option DM is relatively cheap, but only secures modest benefits in relation to the other options. As such, there is always a good incremental case for spending more money on one of the other options and securing a disproportionately large increase in benefit. So DM falls to the bottom of the table. Option 4.1 (Tilbury BWL) is the most expensive option, but despite the fact it has amongst the best performance in protecting property, infrastructure, business and transport (along with the other barrier/bwl options), it does not quite secure as many total benefits as options 1.4, 2 and 3.1/3.2, principally because of poorer scores on water quality, natural processes and physical habitats. As such, option 4.1 is unambiguously inferior to those other options (as it is also more costly) and comes second from bottom, just ahead of DM. Similar arguments can be made against options 4.3 and 4.2; they are more expensive than other options, whilst not securing as many benefits. Option 2, whilst performing rather better in benefit terms than the 4.x options, is the third-most expensive option, does not provide very much higher benefits than option 1.4, and is not as beneficial as the 3.x options (which are also cheaper). This leaves the 3.x options and option 1.4. The latter is bettered by the 3.x options in terms of overall benefit, in terms of extra property, transport and key infrastructure protection, and is only slightly offset by slightly worse environmental performance. Out of options 3.1 and 3.2, the latter is cheaper, but has slightly lower overall benefits, particularly in the areas of property, transport and infrastructure protection. Option 3.1 has marginally better performance in these areas, but (in this central analysis) this is not quite sufficient to justify the extra cost, given that the ratio of extra benefit to cost would need to be exceptional (see above). 4.4 Uncertainty: confidence intervals around benefit-cost ratios A high-level assessment, without modelling, has been made of the potential upper and lower bounds on the key cost and benefit categories for the different options, and how these translate into broad-brush confidence intervals surrounding the IBCRs (in essence these can be seen as 95% confidence intervals ). Table 6 Option ranking under the central assumptions Central ranking Option 1st 3.2 2nd 1.4 3rd 3.1 4th 2 5th 4.2 6th 4.3 7th 4.1 8th DM

19 Nat Hazards (2013) 66: Table 7 Option preference and IBCRs under best and worst cases (final stage IBCRs) Worst case Preferred Option = 1.4 IBCRs over P1 Central case Preferred Option = 3.2 IBCRs over DM Best case Preferred Option = 4.2 IBCRs over Option 3.2 Option 3.2 (21.7) 24.9 Option (24.7) (-41.1) Option 3.1 (20.9) (23.7) (6.0) Option 2 (20.0) (21.3) (-58.1) Option 4.2 (15.6) (18.6) 21.1 Option 4.3 (14.2) (17.9) (16.1) Option 4.1 (12.5) (14.6) (4.7) The IBCR suggesting the preferred option is shown in bold; brackets indicate another option/ibcr is better than the one bracketed Here worst is the based on a collection of variables designed to form a broad lower bound on results, and best is the collection of variables giving the opposite effect In this analysis, upper and lower bounds on individual impact estimates have been used to derive best case and worst case benefit-cost analyses, and broad-brush confidence intervals surrounding the IBCRs for the options identified in the central analysis (Table 7). The results show that the most efficient option changes from option 3.2 (Long Reach Barrier) under the central appraisal, to option 1.4 (Improve existing system) under the worst case, and to option 4.2 (Long Reach BWL) under the best case. This tends to confirm the central appraisal conclusion that options 3.2 and 1.4 are front runners, although the emergence of a BWL option under the best case is interesting. In practice, TE2100 s adaptable approach to options development means that the option 3 standard barriers and the refurbished Thames Barrier under option 1.4 could be designed with eventual conversion to barriers with locks in mind, so if an option 4-type barrier becomes preferred in the future, this could be implemented without significant wasted resources. 5 Uncertainty: sensitivity analysis More specific testing of the sensitivity of option choice to changes in individual assumptions has also been undertaken. The list of key uncertainties tested is as follows: Cost estimates, particularly for non-standard FRM engineering such as flood storage, barriers, and barriers with locks Overall level of implied benefits (as derived through the MCA process) The degree of extra transport and infrastructure (and also historic environment) implied benefits attributable to options 3.1/3.2 The relative level of implied water quality impacts in City Reach for options 1.4 and 3.1/3.2 The weight (implied value) of water quality impacts in the Regeneration Reach downstream of the Thames Barrier, and the weight of landscape impacts, especially in the City Reach The level of indirect impacts on business The level of navigation impacts

20 1400 Nat Hazards (2013) 66: These uncertainties can be grouped into three categories: to do with costs, to do with implied benefits as derived through MCA, and to do with estimates of certain direct monetary benefits. The tests explored the robustness of appraisal conclusions to changes in assumptions here. Our conclusions from this sensitivity analysis (Environment Agency 2009, Chap. 8) are, in general, that the preference identified in the central appraisal results for either option 3.2 (Long Reach Barrier), or option 1.4 (Improve existing system), appears robust to key uncertainties in individual assumptions, in the sense that Incremental Benefit Cost Ratios (IBCRs) for these options remain robust. Indeed, in most cases, it is only extreme (perhaps outlier ) changes in data input values which prompt the IBCR to become less than favourable (an exceptional ratio was taken as one in excess of 10 for these purposes). A caveat surrounding the water quality assessment does again need to be noted. Whilst reasonable changes to the existing assessment of water quality impacts do not suggest option choice sensitivity, it should be borne in mind that our water quality assessment has had to be largely qualitative and did not involve any formal modelling. 6 Uncertainty: scenario analysis The purpose of scenario analysis (as distinct from sensitivity analysis) is to examine the impacts of different general states of the world, as opposed to differences in individual variables. As McFadden et al. (2009, 642) have commented in a mid-term review of its development, the TE2100 project in its strategy process has a clear focus on developing a plan in which current decisions are influenced by possible physical and socioeconomic futures. A number of general scenarios (composed of a number of component variables) were therefore constructed and used to change the baseline (reference case) and option impact assessments used in our appraisal. 6.1 Climate change scenarios The four TE2100 climate change scenarios, and their associated degree of extreme water level rise by 2100, were Low around 0.5 m of rise; Defra06 around 1.0 m of rise; Medium High around 1.5 m of rise High Plus around 2.7 m of rise The appraisal performance of the options has been tested under a High Plus scenario, as well as the central Defra06 one. The High Plus scenario analysis has been based on revised model estimates of property damage and flood impact indicators under the assumed scenario conditions. These data were then fed into the scoring and weighting, and the benefitcost analyses used for the Defra scenario, but using modified weights to reflect the typically larger swing in impacts across options. For impact categories where more qualitative assessments were made (mostly in the environmental area), the key driver of impact was typically the number of barrier closures, and inspection of model results showed that critical barrier closure rates under the Defra scenario were generally reached an epoch earlier under the High Plus scenario. This meant that the total (undiscounted) implied value estimates for the qualitative assessments could simply be brought forward an epoch within the benefit-cost analysis: the option choices remain the same, but their implementation is accelerated.