Doing flood risk science differently: an ... - Wiley Online Library

ransactions

of the Institute of British Geographers

Doing flood risk science differently: an experiment in radical scientific method S N Lane*, N Odoni*, C Landstro¨m**, S J Whatmore**, N Ward and S Bradleyà In this paper, we describe an experiment in which the position of scientists with respect to flood risk management is fundamentally changed. Building on a review of three very different approaches to engaging the public in science, we contrast the normal way in which science is used in flood risk management in England and Wales with an experiment in which knowledge regarding flooding was co-produced. This illustrates a way of working with experts, both certified (academic natural and social scientists) and noncertified (local people affected by flooding), for whom flooding is a matter of concern, and where the event, flooding, is given agency in the experiment. We reveal a deep and distributed understanding of flood hydrology across all experts, certified and uncertified, involved in the experiment. This did not map onto the conventional dichotomy between ‘universal’ scientific expertise and ‘local’ lay expertise. By working with the event we harnessed, produced and negotiated a new and collective sense of knowledge, sufficient in our experiment to make a public intervention in flood risk management in our case-study location. The manner in which the academic scientists involved in the practice of their science were repositioned was radical as compared with normal scientific method. It was also radical for a more fundamental reason: the purpose of our experiment became as much about creating a new public capable of making a political intervention in a situation of impasse, as it was about producing the solution itself. The practice of knowledge generation, the science undertaken, worked with the hybridisation of science and politics rather than trying to extract science from it. key words flood risk management co-production hybridisation

flooding

scientific method

participation

*Institute of Hazard, Risk and Resilience, Durham University, Durham DH1 3LE email: [email protected]

**Oxford University Centre for the Environment, University of Oxford, Oxford OX1 3QY Faculty of Social Sciences, University of East Anglia, Norwich NR4 7TJ àCentre for Rural Economy, Newcastle University, Newcastle-upon-Tyne NE1 7RU revised manuscript received 27 July 2010

Introduction On the 26 June 2007, Pickering, a small market town in Ryedale, North Yorkshire, UK, suffered a major flood event (Plate 1) with approximately 48 homes and businesses flooded.1 The town has experienced flooding in 2002 and before that in 2000 and 1999. Proposals for a flood alleviation scheme for the town of Pickering were developed after the 2000 event and considered by residents in July 2003.2 These proposals would have protected

51 homes and 13 businesses from floods of a similar magnitude to those experienced in June 2007 at a cost of approximately £6.7 million (in 2003 figures). Unfortunately, the cost-benefit ratio associated with this scheme was not favourable,3 has not become favourable4 and the town, as with many similar to it, was to remain undefended. Just 4 days before the event of the 26 June 2007, our advertisement for members of what was to become the Ryedale Flood Research Group (RFRG) had appeared in local newspapers in Ryedale. In

Trans Inst Br Geogr NS 36 15–36 2011 ISSN 0020-2754 2010 The Authors. Transactions of the Institute of British Geographers 2010 Royal Geographical Society (with the Institute of British Geographers)

S N Lane et al.

16

Plate 1

Pickering, North Yorkshire, 26 June 2007 Source: Mike Haigh

March 2007 we had begun a 3-year project to examine the production and circulation of environmental knowledge in relation to rural land management and the ways in which the creative

potential of ‘knowledge controversies’ can be positively harnessed in the practice of interdisciplinary public science. Our focus was on the science of flood risk management, and our first of two case


Doing flood risk science differently

studies was Ryedale. We wanted to try out what happens when local people and academics are involved in the knowledge production process from the outset through an experiment5 in doing (flood risk) science6 differently. This paper provides an account of that experiment. We begin by positioning our experiment within broader approaches to public engagement with and in science and consider the normal ways in which science informs decisionmaking in flood risk management. We then detail our experiment and use our experience to draw some wider lessons for public participation in science. The paper is sustained by evidence from government reports, interviews with flood risk scientists in consultancies and government organisations, and transcribed material from video footage of our own experiment.

Types of engagement between scientists and publics Callon (1999) outlines three categories for what he calls ‘the involvement of lay people’ in science and technology – the Public Education Model (PEM), the Public Debate Model (PDM) and the CoProduction of Knowledge Model (CKM) – and we use his framework to situate our own experiment. Callon’s PEM positions Scientific Knowledge as the opposite of lay knowledge, grounded in the supposed universality and objectivity of science as a practice, institutionalised as Science, and governed autonomously by its own norms and procedures (Callon 1999). It is these characteristics that are assumed to give Scientific Knowledge and Science authority over other sorts of knowledge, and we use ‘SK’ and ‘S’ to indicate this authority. Callon argues that intermediaries (e.g. public authorities) provide the necessary contact points in a linear relationship between Science and the public, one that requires trusting relationships between Scientists and lay people. According to Callon, it follows that when mistrust in Science arises, whether due to unintended consequences, realisation of scientific fallibility,7 the perception of failings in the intermediaries that sustain the position of Science, or the role played by the media in distorting what Scientists are saying, the assumed antidote is to restore the linear flow through public education. In other words, science is sufficient but the public are deficient (Sturgis and Allum 2002). The PEM leaves Science intact and focuses on education to overcome the deficit, assuming that more rational public behaviour in relation to decisionmak-

17

ing will result (Callon 1999). A Scientist’s public engagement must focus on communication, wellillustrated by the Royal Society of England and Wales’ view of the twin responsibilities of a scientist: The first is to attempt an accurate assessment of the potential implications for the public. The second is to ensure the timely and appropriate communication to the public of results if such communication is in the public interest. (Royal Society 2006, 5)

Callon recognises that most publics are differentiated and situated because they possess ‘specific, particular and concrete knowledge and competencies, the fruit of their experience and observations’ (1999, 85). As the knowledge that Science can deliver about the real world is inevitably incomplete (e.g. Wynne 1992), it is always at risk of becoming contested when it is exposed in particular situations. Further, the knowledge that Science generates should be opened up so as to enrich and ‘enhance the abstract and inhuman knowledge’ of scientists (Callon 1999, 85). Thus, the Public Debate Model accepts that Scientific Knowledge should be at least provisional until the point at which those who have a stake in the implications that follow from that knowledge, as well as a contribution to make in terms of debating that knowledge (cf. Collins and Evans 2002), have an opportunity to comment on it. Examples include focus groups, the ‘extended peer review’ of Funtowicz and Ravetz (1991 1993) and Yearley (2006), deliberative mapping exercises8 and citizen juries (Irwin 1995). These approaches are, in many ways, devices that address the challenge (Wynne 2003) of enabling the silent or silenced to speak, widening the knowledge made admissible. For Callon, this raises questions of just how different voices are reconciled, the representativeness of those voices allowed to be heard (see also Rayner 2003) and the difficulty that publics may be ‘no more pusillanimous than the researchers . . . no more prisoners of their beliefs than the experts’ (Callon 1999, 88). The PDM is at its most radical when debate is extended ‘upstream’ to consultation over how a problem is framed (e.g. Wynne 2003; Costanza and Ruth 1998; Cockerill et al. 2006; Johnson 2009). Even then, the PDM sees Science, Scientists and their underpinning working practices as things to be left intact. We add two additional concerns with the PDM. First, however upstream the PDM is positioned, some PDM type engagements may be little more than exercises in legitimation (Harrison and Mort


S N Lane et al.

18

1998; Rayner 2003; Stirling 2008), criteriacompliance (Stirling 2008), a process to be worked through to tick boxes rather than influence outcomes. The ultimate interpretation is that it reinforces deference (Yearley 2000) to Science and Scientific Knowledge. Second, there is an internal contradiction in the PDM between seeing knowledge generated through Science as provisional and uncertain, debateable; and the assumed notion that Science produces certainties on which the consensus will settle. Not only has the provisional nature of scientific knowledge been clearly demonstrated in a number of cases (e.g. Wynne 1992), but Science itself has evolved a series of practices (e.g. repetition in experiment, experimental control, statistical inference, peer review) as an internal (and partly externalised when published) means of strengthening trust in the knowledge it produces. If practising science develops the trust that scientists hold in their own work, then perhaps the only means by which others might develop that same trust may be through involvement in the same practices (Yearley 2005). Callon’s CKM attempts to answer this question. He argues that both the PEM and PDM deny the competence of publics to participate in the process of knowledge generation. In the CKM he recognises a ‘constantly renewed tension’ between the production of ‘standardised, universal knowledge . . . [and] . . . knowledge that takes into account the complexity of singular local situations’ (Callon 1999, 89).9 Callon’s CKM notes that the typical mapping of the PDM onto an expert–lay divide – in which experts possess universal knowledge and differentiated publics possess local knowledge that can challenge assumptions made by those applying universal knowledge to particular places – does not capture the capacity of publics to be involved in all elements of knowledge production. Nor does it challenge public understandings of a problem that may be as highly differentiated as those in scientific communities. In the same way that Scientists question their understandings through practice, so must publics. Thus, under CKM, knowledge is co-produced through a process of dynamic, collective learning involving those for whom an issue is of particular concern, whether as a result of their professional position (their ‘certification’ following Collins and Evans 2002), their personal position with respect to an issue or their personal experience of an issue. This explicitly recognises ‘more socially distributed, autonomous and diverse

forms of collective enterprise’ (Expert Group on Science and Governance [EGSG] 2007, 10), something that ‘would perhaps be the most effective single commitment in helping address legitimate public concerns about Europe as a democratic knowledge-society’ (EGSG 2007,10). Knowledge generation is no longer a property of Science, and the knowledge it produces is no longer accorded special privilege over other knowledge. Science becomes science. This process does not eliminate the need for the involvement of Science, rather it removes its privilege and emphasises that it is, on its own, insufficient (Collins and Evans 2002). Callon’s CKM captures a theme central to debates over expertise in decisionmaking (e.g. Collins and Evans 2002). Expertise is more widely distributed within society than many might imagine. The question becomes how to mobilise and to diversify that expertise and what happens to the expertise of those traditionally Scientists during this mobilisation and diversification. Whatmore (2009) describes the agency that ‘hot situations’ (Callon 1998), ‘matters of concern’ (Latour 2003) or ‘experimental events’ (Stengers 2005) can have in exposing those elements of the material fabric of our everyday lives that have become hardwired into the products and policies that shape those lives. In turn, by putting those events to work, new kinds of understanding may be generated that unsettle those products and policies (Whatmore 2009). Flood events have been recognised as catalysts for policy change in the UK (Johnson et al. 2005; Penning-Rowsell et al. 2006), especially in terms of translating ideas that were already the subject of discourse into policies. Here, we use the event of flooding in Pickering to bring together social and natural scientists and affected residents as an epistemological practice that obliges both the certified and the non-certified to work with the event, together forcing the redistribution of expertise to other people, things and places. To understand why this redistribution of expertise is different, the next section outlines the normal practice of flood risk decisionmaking in England.

The ‘normal’ practice of flood risk decisionmaking: Pickering before the 25 June 2007 Pickering and flooding Pickering is located at the boundary between the predominantly Jurassic sandstone and gritstone



escarpment of the North Yorkshire Moors National Park and the glacial-lacustrine deposits of the Vale of Pickering. The catchment of Pickering Beck has an area of approximately 70 km2. The town is located where Pickering Beck emerges from a confined valley, formed by late glacial meltwater erosion. The town itself has a long history of flooding, although there has been a marked increase in flood frequency in the instrumental record. Since 1974, floods led to the inundation of properties in September 1993, March 1999, October and November 2000, August 2002 and June 2007, and flows almost resulted in flooding in September 2008, November 2009 and January 2010. Documentary records provided by the British Hydrological Society (Chronology of Hydrological Events, http:// www.hydrology.org.uk), which cover the period to 1950, record major flood incidents in Pickering in May 1864, August 1866, June 1872, October 1880, August 1884, October 1892 and March 1947. Obtaining reliable hydrological data for the catchment is difficult because the two gauges in Pickering both have unreliable stage-discharge curves due to the effects of both structures and gauge bypassing. However, the Environment Agency was able to obtain a reliable flow estimate for the onset of inundation of properties in the November 2009 event (6 properties between 12 and 15 m3 s)1) and subsequent refinement of the hydraulic models for the town of Pickering suggests that at flows greater than 15 m3 s)1, the number of flooded properties rises rapidly to 50. The peak discharge of the June 2007 event has now been recalculated as 28.7 m3 s)1, and during this event, the flow exceeded 15 m3s)1 for approximately 23 hours and 12 m3 s)1 for approximately 18 hours. Inundation depths were > 1.5 m for some properties during these periods.

Flood risk management in England: the national context The determination of flood risk associated with rivers and the design of schemes to reduce that risk in particular places normally addresses five areas, although these are not necessarily undertaken sequentially. Four of these areas are concerned with risk estimation and possible methods to reduce risk: hydrological analysis; hydraulic analysis10; assessment of losses; and assessment of possible physical interventions (e.g. flood defences). The fifth is concerned with public engagement regarding estimated risks, which may include wider elements of flood risk management,

19

rather than just flood risk reduction through physical interventions. Central to risk estimation and intervention is the requirement imposed by central government to determine the costs of flood losses and hence the cost-benefit of interventions to reduce those losses (MAFF 2001), notwithstanding recent moves to introduce additional criteria into the prioritisation of interventions that meet cost-benefit requirements (Defra 2005). Hydrological analysis, including hydrological modelling, is used to determine the probability of events of different magnitudes, generating a set of river flows that can then be applied to hydraulic models. The geographical boundary between the hydrological treatment (generally upstream) and the hydraulic treatment (generally downstream) is blurred as the upstream boundary of the hydraulic model may be extended upstream so as to capture the effects of upstream processes (e.g. floodplain storage) upon downstream flows. Hydraulic models determine the spatial patterns of flood inundation associated with each flow to estimate consequent economic losses. There have been recent attempts to standardise this loss estimation in a GIS framework (the Model Decision Support Framework, http://www.mdsf. co.uk) by importing hydraulic results and using these to calculate economic and social impacts. MDSF allows for some basic simulation of options within the domain covered by hydraulic results, but the assessment of possible schemes normally requires further application of hydraulic models to capture the complex effects of structures like flood defences on flow hydraulics and hence water levels. In order to determine cost-benefit ratios, and to compare options, the model predictions for these schemes may also be analysed in an MDSF. Unfortunately, this practice is challenging. MAFF explains: There are general difficulties in working with rare hydrological events in the ‘tail of the frequency distribution’. This situation can apply even when relatively good data sets are available for analysis. Records of river flow seldom exceed 50 years at present and yet typical design standards of 100 and 200 years are required. (2001, 14)

Similarly, determining where water will extend to on a river floodplain is based on a series of numerical simplifications to, and approximations of, governing equations, which result in flood extents and depths becoming highly uncertain (e.g. Yu and Lane


S N Lane et al.

20

2006a 2006b; Hunter et al. 2007). Thus, and notably over the last decade, a series of devices, practices and institutions have co-evolved. In England and Wales, where a local flood issue is identified as meriting scientific investigation (whether through an event, or through national scale profiling of risk), the Environment Agency commissions one of a small number of certified consultant scientists11 to undertake a strategic flood risk assessment. The science practised by these consultants is driven by a series of constraints, including: advice on the best pieces of technology to adopt,12 and guidelines on how to apply the associated technical devices13 and the professional capabilities of the staff available to do the work. There is some opportunity for innovation, such as by applying new technologies. There is a predominance of use of certain sorts of hydrological models and associated procedures, notably those recommended in the Flood Estimation Handbook (FEH, Institute of Hydrology 1999) for hydrological analysis and one of two models (ISIS, jointly developed by two private companies, HR Wallingford and Halcrow; or Hec-RAS, developed by the US Army Corps of Engineers) for hydraulic analysis. Consultants take these procedures (i.e. technologies that contain some universalism) and make them work in particular places, something that inevitably requires data. Hydrological analyses may require estimates of the magnitude of previous extreme events in the catchment to inform a flood frequency analysis, event data ‘borrowed’ from similar nearby catchments where record lengths are short or rainfall or catchment data to drive rainfall-runoff models; hydraulic models require data to initialise the boundary conditions and to allow the model to be calibrated to match known historical events (what Oreskes et al. (2004) call forcing empirical adequacy). As with risk estimation and intervention methods, public engagement is also steered by national policy recommendations. Public engagement policy in relation to flood risk management is developed by the Department for Environment, Food and Rural Affairs (Defra) in England and Wales. Unlike risk estimation and design of schemes where consultants play a dominant role, such policies are implemented by local Environment Agency offices in relation to individual schemes, drawing on the consultants’ risk estimation. The last decade has seen two important changes. First, many flood defences were introduced when defence priorities were somewhat different, such as associated with

the need to protect agricultural land from flooding to sustain food production. For example, the Ministry of Agriculture and Fisheries’ (1948) account of the 1947 floods in England was entitled Harvest Home, describing the extraordinary efforts put into preventing leve´e failure so the harvest could be ‘brought home’. More recently, there has been a shift towards a broader approach to flood hazard management, culminating in Defra’s Making Space for Water initiative (Defra 2005), which includes alternatives to traditional engineered defences such as more careful land use planning, catchment-wide analysis of flood risks and greater use of rural floodplains as washlands. This shift has also recognised that the public need to be more engaged with flood risk, its geography and the wider set of measures that can be adopted to reduce exposure and vulnerability (Defra 2005). Thus, second, the last decade has seen substantial attention given to how to engage publics in both flood science and flood risk management. This has not just involved efforts at better flood risk communication (e.g. Defra 2004), but recognition that local people have a right for wider engagement in flood risk management decisionmaking. Defra commissioned an independent study of the extent to which the Environment Agency was providing such ‘opportunities to participate in deliberations and influence decisions’ (Defra 2008, 18), including the Environment Agency’s own implementation of public engagement. The study concluded ‘Stakeholder engagement tends to be focused on consultation and the provision of information; rather than participatory stakeholder engagement per se’ (2008, 173). However, the Environment Agency’s national policy team have developed a tool kit for use by regions and areas in their day-to-day implementation of public engagement responsibilities: Building Trust with Communities14 (Environment Agency 2004a 2004b). It is based on 12 principles (Table I), including filling knowledge deficits (principle 3); the mapping of trust onto communication (principle 4); and the importance of separating decisionmaking and communication of that decisionmaking (principle 5).

Flood risk management in Pickering Pickering Beck becomes a main river15 just upstream of the town and therefore the Environment Agency has powers to defend the town from flooding. Initial work by consultants made use of the rainfall-runoff modelling strategy


Doing flood risk science differently Table I

21

Building Trust with Communities’ 12 core principles (Environment Agency 2004a, 13–14)

1. Fair for all. Every person who has an interest in, or who could be affected by, the issues under discussion must be encouraged to take part. 2. Be clear at the start about what changes the Environment Agency can or cannot promise and be clear about the mechanisms of the decision-making processes. 3. Ready information. Be sure you give people as much information as possible and explain where information is missing or is uncertain. 4. Show respect for diverse views and cultures by making sure that minority views are taken on board. Respect interested parties and taxpayers by making sure that your work with local communities is seen as a priority and has widespread support from the community. This is your opportunity to build trust by being courteous, empathic and helpful. 5. Feed back. Use existing channels to make sure that you report back to all interested people as fully and as quickly as possible. 6. Take action. Put final decisions into action as soon as possible. This will strengthen participants’ belief that their involvement was worthwhile. 7. Each time there will be lessons to be learned for both the Environment Agency and the community groups, building mutual understanding, trust, respect and relationships. Some initiatives will fail but they should be seen as valuable contributions as they provide fresh insights. 8. Stand alone. The Environment Agency needs to remain independent throughout the exercise. 9. Common approach. The Environment Agency needs to convey that it is guided by principles that are based on objective professional standards and must be seen to apply these standards across different contexts. 10. No time wasters. Make effective use of time and funding resources for all. 11. Balancing act. The amount of time spent on a project should depend on how important it is. 12. The bigger picture. The aim of everything the Environment Agency does is to improve the environment.

that was the precursor to the FEH methodology, focusing on the large flood that hit the town in 1993. After floods in 1999 and 2000, this work was given greater priority, resulting in a hydraulic analysis and the recommendation for construction of a series of permanent flood defences along Pickering Beck in and near the town itself. The Environment Agency submitted an application (in 2001) for planning consent to construct these defences despite the fact that the cost-benefit analysis was not favourable. However, there was local resistance to the EA’s original scheme and the plan was withdrawn. Following the August 2002 flood, consultants Babtie, Brown and Root were commissioned to comment further on causes of the flooding and the number of properties at risk from future events. Babtie (2003) updated the hydrology, given publication of the new FEH in 1999, and used ISIS’s routing methodology as well as ISIS’s full hydrodynamic modelling capabilities to explore eight different options for Pickering: 1 2

3 4 5 6

do nothing do the minimum (keep channels clear of vegetation, also do some minimum dredging and channel maintenance) improve maintenance and warning develop upstream storage widen the existing channel regrade the existing channel (altering the gradient of the channel)

7

8

construct local flood defences (raising the height of flood defence walls, building some new walls) and construct a flood diversion tunnel.

With the exception of (4), all of the above measures related to flood prevention either in the town or in its immediate environs. The upstream storage solution was modelled using a combination of three methods (Babtie 2003): 1 a priori estimation of the required reduction in runoff volume to eliminate flooding at Pickering 2 modelling the effects of two reservoirs in the upper catchment using the FEH unit hydrograph rainfall-runoff methodology (Institute of Hydrology 1999) and 3 combining the modelled outflow hydrographs with and without reservoirs and routing them downstream to estimate flows at Pickering. Upstream of the town, in the catchment of Pickering Beck, there are only two rain gauges and one unreliable water-level recorder. Thus, it is not surprising that it was necessary to adopt a ‘lumped’ approach that treated all attributes upstream of their downstream boundary (i.e. where they interface with the hydraulic model) as a single spatial unit. This is not necessarily a problem for the estimation of flow probabilities and these approaches have generally served flood risk management very


22

S N Lane et al.

effectively. Indeed, Babtie (2003) followed industrystandard practice in the Pickering study. A problem occurs, however, if certain solutions are not permissible because standardised technologies restrict what can be assessed. In this case, the FEH analysis can only provide a restricted assessment of the effects of upstream storage measures, and cannot explore the sequential and interacting effects of a number of flood storage zones created by bunds.16 Babtie (2003) showed that the option of improving and constructing walls was the most cost effective solution of those that they were able to explore. Unfortunately, there was a major change in the funding arrangements for flood defence during the same period. While government funding for flood risk management doubled in real terms between the late 1990s and 2005–6 there was a reduction in the percentage of this funding that was routed from Central Government through local authorities for allocation by Regional Flood Defence Committees (from c. 70% in 2003–4 to c. 10% in 2004–517). This routing gave Regional Flood Defence Committees a substantial role in being able to prioritise investment. From 2004–5, a national points-based scheme was introduced, in the context of pressing needs for flood investment elsewhere. In the words of the local area Environment Agency Flood Risk Manager (March 2008):

Given the number of times that Pickering has been flooded in recent years, this may seem unfair. Indeed, it has been noted that the UK’s flood risk reduction strategies can appear to be technically and economically the most effective, but simultaneously unfair in terms of vulnerability and equity considerations (Johnson et al. 2007). Given the repeated occurrence of flood events in Pickering, the Environment Agency has actively been engaged in consultation with the local community, before, during and after events. The Environment Agency Flood Risk Manager for the area noted: ‘I have been in York at the Agency for 10 years, and in my 10 years, here I have stood in flood water in Pickering on, you know, more than one hand’s full of occasions.’ The emphasis of this engagement is, in the best of intentions, consultation, generally throughout the process of considering options:

During that period of re-appraising the scheme, DEFRA came up with . . . the ubiquitous long term plan, medium term plan, which is this national points scoring system, that we have now. We are soon to get the outcome measures scheme, but for now we have the point scoring scheme. And that came in during this time. And Pickering found its rightful place on that system. It has got a score of about four. At the minute, schemes of about 30 are progressing.

With somewhere like Pickering, and with somewhere where you have held flood surgeries, and wash-up meetings and all these things after all the floods, to be honest you have already got an idea of what people want and expect, because you have got this feedback from the people on the ground. So if . . . I was to develop an early meeting for Pickering, I perhaps might not go to that level of detail for that early meeting, because we have already got that information, we have seen them enough on the back of floods. (Environment Agency Flood Risk Manager, March 2008)

When the 2007 flood occurred, Pickering was still unprotected. The EA Flood Risk Manager went on to note: we took a quick look at the flooded properties in the 2007 flood versus the flooded properties that were within the appraisal of the original scheme. And they weren’t significantly different . . . We have got that scheme placed within the priority score and schedule, and it has got a score four. It is not destined to be happening, at the minute we have got a ten year priority scoring list, and it doesn’t appear on that ten year list. So at the minute, it is years 2018 onwards, and that is all it says for Pickering, you know, it is there on that list.

you have a project start-up meeting, which is an internal thing, and then either you open the early consultation up fully, and you get people in just to come and give their views on what the options should be, what it feels like to them, looking forward [to] what would they like to see for the town. Or you close [it] that little bit more and just get key stakeholders in to give their early views.

In a place with repeated flood events, but not much progress in terms of delivering schemes, this can be difficult:

A member of Environment Agency staff (March 2008) responsible for managing public engagement, including in Pickering, noted that, after flood events, drop-in sessions allow EA staff ‘to find out the nitty gritty of what happened’, but that they are much less good at knowing what to do with it: I think we are a bit bad at sort of saying, when somebody comes and says ‘I have lived by this river for the last 50 years’ . . . just dismissing it as folklore. And in actual fact there is an awful lot of knowledge there, and


Doing flood risk science differently I think . . . I sometimes don’t think we take that seriously enough. I mean we record it all, but whether we do anything with it . . . whether we compare it with what our models show. I think . . . the Environment Agency is very heavily staffed by engineers and techy people . . . there is that historic knowledge there, and people have lived in this area a long time. And I sometimes don’t think we take that seriously enough.

The evidence from interviews with local people reveals immense frustration with Environment Agency staff which, while not surprising in a situation where little progress with flood risk reduction has been achieved, also reveals a deeper frustration (Table II). The frustration includes the way meetings are managed to dampen controversy and the ways that knowledge is presented in ways that do not allow full scrutiny and which suppress debate (i.e. the lack of ‘nitty gritty’). There are claims that suggest there is a strong hierarchy of knowledge, with perceptions that local people were being Table II

23

‘made fun of’ behind their backs, ‘discounted in an offhand way’ and a general lack of trust. By June 2007, there had largely been a breakdown in trust in the Environment Agency, as the key intermediary between scientists and local people, in Callon’s (1999) terms. As the Environment Agency Flood Risk Manager notes, the 2007 event did not change the fact that there was little that could be done in Pickering in terms of conventional flood defence schemes. Not only does public engagement surrounding flood risk management in Pickering map largely onto a mixture of Callon’s PEM and PDM, there is a widespread perception amongst local people that public engagement is part of an active means of managing the politics of making difficult decisions in controversial situations. By sustaining a knowledge hierarchy, an institution can deliver and sustain its own logic while simultaneously delivering its commitment to public engagement.

Views of the Environment Agency’s engagement regarding flooding in Pickering

In a discussion (Competency Group [CG] meeting 2, November 2007) about how the project outputs would be used, as part of a concern to make sure that they would be of use, a local member was worried that the RFRG’s work would go unnoticed by the Environment Agency: ‘[in] all the dealings with the Environment Agency they just seem to blithely go along their own way, and take damn all notice of anybody. So you give them a good example like that, is anybody going to follow it up and say ‘‘Have you taken any notice?’’‘ In a discussion over whether or not river maintenance activities should continue to be withdrawn in Ryedale, a local member of the RFRG (CG4, March 2008) noted: ‘The main thing that concerns me is that every time this issue of bank clearance, silt clearance, whatever, comes up, the Environment Agency, and certain others say, ‘‘It is not worth doing it, because it has very little effect overall’’. And it is such a hard argument to overcome, when you get a sort of bland statement like that, with no real, definite argument behind it.’ In a discussion as to why one local member (CG4, March 2008) was in disagreement with the Environment Agency, they concluded: ‘When they will say things that are patently untrue, how can anyone discuss anything with them? I would like to cooperate, I would love to.’ In a discussion of previous experience of flood-related meetings in Pickering, a local member (CG6, July 2008) noted: ‘There is a tendency for a chairman to decide he is going to keep charge all the time. But he is not going to allow controversial questions, and he wants a quiet meeting and he wants to go home. And that, unfortunately, without knocking the Environment Agency, again, that is what happens all the time.’ In a follow-up interview, a local member (June 2009) contrasted their experience of the RFRG with previous flood-related meetings they had been to: ‘There was a big meeting of the EA and an inquest afterwards at their own head offices, [of] the regional flood defence committee, and I attended that meeting. And they made fun of me, did the EA executive, in front of the other . . . There’s the executive and then there’s the board people, the ones from the councils and various elected people like that – the representatives serving on that flood defence committee. And they made fun of what I had to say afterwards, because a councillor from ** told me so. [** ] didn’t know me before I went, but ** knew what I was saying was genuine.’ In an interview (March 2008) with a local councillor, the interviewer asked about the value of local knowledge about flooding. The councillor responded: ‘I think that it clearly should . . . it should fit in. At the moment it clearly doesn’t. It is not allowed to, and that I think is a serious error of judgement because it is like the . . . the hydrologists, the scientists are all going a very straight, linear route that is almost like purely theoretical. And that would be . . . that would be ok if they were 100% right. Now they may be significantly correct, in terms of the proportions of what works, and what doesn’t work, in terms of flood management. But I still feel, even though I understand where they are coming from, I still feel quite strongly that they . . . there is a lot of historical evidence, that they have just completely discounted in a very offhand way. And that there is a kind of myopia there that is again born of confidence, but confidence is moving into arrogance, I think.’ In an interview (March 2008) with a second local councillor, the interviewer asked about what they thought the Environment Agency’s views was of low cost solutions. The councillor responded: ‘Don’t trust the Environment Agency. The Environment Agency . . . if they get their experts on it, they will pay them a fortune, and they will come up with a massive system. The Environment Agency cannot be trusted.’


24

An experiment in doing flood risk science differently Our experiment in doing flood risk science differently explicitly addressed the fundamental purpose of scientific contributions to flood risk questions. If one sort of knowledge is being mobilised to sustain a particular sort of institutional logic, then one role for science, albeit a radical one because of the traditional assumption that science should be divorced from politics, is to provide a means by which other sorts of knowledge might be mobilised to sustain other logics. This is what we attempted to do in our experiment, through the kind of CKM envisaged by Callon (1999). Mirroring the distinction we make above between Science and science, it is not an Experiment as an act that has become formalised as a part of Scientific enquiry but an experiment where, using the origins of the term, we try something out through experience.18

An Environmental Competency Group: the Ryedale Flood Research Group Our experiment was based on a novel form of public engagement called an Environmental Competency Group which, in the case of the experiment reported here, chose to call itself the Ryedale Flood Research Group (RFRG). Our approach is distinctive for five reasons (Ryedale Flood Research Group 2008). First, there is a focus on the practice of knowledge-production as well as the knowledge produced itself. This makes the approach distinctive from approaches to public engagement like focus groups, where the focus is on what people think or believe about the products of knowledge. Second, we take the emphasis on knowledge production as being about more than talking about how knowledge is produced. Rather, we see research as a collaborative process in which participants work together and with other ‘things’.19 Third, we place an emphasis on the role that an event can play in bringing into sharp focus the prevailing framings associated with a problem, and the people and things bound to it, so mobilising and enabling those people and things otherwise excluded from the process. In our case, our focus was on academics and local people for whom flooding is a ‘matter of concern’ and who, by virtue of this concern, are able to make active contributions to the progress of, in this case, flood risk science. By bringing them together over a sustained

S N Lane et al.

period, we aimed to generate new collective knowledge and skills (competencies). Fourth, it is unusual for academics and local people to work together on projects associated with flood risk estimation. Local people, in particular, as we noted above in relation to Pickering, are engaged with flooding in particular ways, emphasising the end points of events (post-flood surgeries to explore options; option appraisal over ways forward) and not what happens between these end-points. There is a dominant and wider tendency for applied environmental scientific research to be driven by those with statutory responsibilities (e.g. the Environment Agency or local government), commonly intermediaries in Callon’s terms, rather than those who have to live with the consequences of those responsibilities, whether local people or academics. Fifth, and following from this, we did not set out to be representative of pre-existing stakeholder groups (e.g. land managers, flood victims, public bodies, local politicians and officials etc.). Explicitly, this recognises that there is no such thing as a ‘representative’ group because any attempt at representing the complex social-economic-political composition of communities is itself an act of framing around a preconceived notion of what that composition is.20 Ryedale was chosen as our focus before the floods of June 2007. We recruited eight local members through advertisements placed in shop windows, community notice boards, including in a supermarket and a doctors’ surgery, the local library, and museum networks as well as the local press. Five academics, a facilitator and a meeting manager and recorder completed the group. All local members were offered four things, but were not promised that the RFRG would actually make progress with reducing Ryedale’s flood risk: 1 an opportunity to ‘make something together’ – though we stressed that we did not then know what 2 an opportunity to be part of, and help shape, a national project about how scientists and local residents can work together more effectively 3 intellectual engagement in a collective endeavour, learning about different experiences and understandings of the local water environment and 4 an opportunity to produce some oral testimony (through life history recordings) of people living in Ryedale at the time of the 2007 floods.



In other words, the framing that we brought to the experiment was a wider context of trying out a different means of practising science, in which both academics and the public worked together to co-produce knowledge rather than starting out to address a particular flood risk problem using a particular sort of method. Once the eight local members were recruited, six meetings of the Group took place at two-monthly intervals between September 2007 and July 2008 (CG1 through CG6). Contact with Ryedale-based members included regular e-mails and telephone calls and working together to produce written resources for the Ryedale Flood Research Group’s website, which was password protected, with access restricted to members. The facilitator met individually with each local member between meetings. As the meetings progressed, local members became actively involved in generating materials, initially including records such as photographs, subsequently data such as estimates of Manning’s n values for river roughness from photographs for water-level determination. Most local members also attended a reading group on consultants’ reports. Meetings were video-recorded and transcribed with the permission of all members. All meetings were attended by all members except for two, each on one occasion due to family ⁄ personal commitments. Initially, meetings were structured by the facilitator, but as the collective competence of the RFRG developed, so the meetings started to struc-

Table III

25

ture and direct themselves. Table III summarises the broad activities covered in each meeting. Two elements of the work in Ryedale need particular mention in order to inform subsequent discussion. First, from CG3 onwards, the RFRG’s sense of collective competence grew to the point where it wanted to make an active intervention in the flood risk management in Ryedale. We had agreed to work confidentially throughout the project. But, after discussion, the Group decided to ‘go public’ through an Exhibition of our work. This was mounted in October 2008 and attended by approximately 200 people. Over the same period, Ryedale became the focus of a bid to Defra as a demonstration project in using rural land management to reduce flood risk, in which the solutions generated by the RFRG would be further explored alongside others. This bid was successful, and the project is now underway. Second, through the RFRG we developed a new model, the ‘bund model’. We make no claims to the innovation inherent in this model, but rather the process by which it was developed and used by the RFRG. When we embarked on the project, we had expected to take an existing, physically based, hydrological model (see Lane et al. 2009) to explore the effects of general upstream land management impacts on downstream flood risk. Before the first meeting of the RFRG we had focused on synthesising relevant data together, including digital elevation data, and coding some key data

Summary of RFRG activities

CG1. Starting the Narrative. Introductions; explanation of the wider project, discussion of ‘brought objects’ within break out groups (we were all asked to bring an ‘object’ to illustrate our connection to flooding); discussion of flooding as a matter of concern; co-production of a history of flooding in Ryedale. CG2. Working with Crossing Points 1. Discussion and explanation of the social science elements of the project, including research into how flood science is undertaken by academics and consultants; discussing Environment Agency flood inundation maps in break out groups and plenary; plenary discussion of how these are produced, introducing models; marking up of maps in terms of solutions for reducing flood risk; plenary discussion; discussion of what needs to go into models, hydrology and hydraulics. CG3. Trying Things Out. Round up of activities from all; plenary discussion of computer modelling; trying out the ‘bund model’ to look at upstream storage in break out groups; plenary discussion, including limitations; identification of research needs, including data, and plans for collecting it. CG4. Working with Crossing Points 2. Round up of activities from all; reports on data gathering; discussion of wider developments in flood risk policy; working with video imagery produced by local member; discussion of river maintenance and its relationship to flood risk; discussion of next phase of modelling; planning for what the RFRG might go on to ‘produce’. CG5. Trying Things Out. Round up of activities from all; plenary discussion of hydraulic models; working with a hydraulic model to explore maintenance impacts on water level in break out groups; plenary discussion of findings; reflection on flood risk policy in relation to maintenance; decisionmaking over what the RFRG would produce; allocation of tasks. CG6. Where next? Round up of activities from all; plenary discussion of how to intervene in Ryedale flood risk management more generally; finalisation of a ‘going public’ event; completion of ethical and data permission agreements.


S N Lane et al.

26

management routines needed for that model (algorithms to calculate catchment boundaries, extract relevant topographic data from national datasets, calculate flow routes over the surface of the landscape). Following from CGs 1 and 2, the modelling activities began to get a different steer from two important directions. First, frustrations were expressed among local members regarding the failure of previous modelling efforts to capture upstream storage processes, especially those associated with a large number of smaller storage areas. The physically based hydrological model that the modellers were working with focused on predicting changes in river flow allowing for changes in processes like infiltration but, as is standard in physically based hydrological models, it forced small-scale storage out of the model through use of a pit-filling algorithm that made sure that all water flows downhill. Second, a more detailed qualitative understanding of flood-generating processes in Ryedale began to develop, generated by both local members (e.g. through personal accounts of flooding) and academic members (e.g. analysis of discharge records, especially as the 2007 event data became available). By the end of CG2, it was clear that the group wanted a focus on upstream storage. Initial considerations were given to developing a detailed hydraulic model of the river network but, aside from the absence of necessary boundary condition data (notably inflow discharges from many tributaries), such models do not allow for feasible assessment of multiple small management activities nor for the active ‘trying out’ of solutions by members of the RFRG. In the most extreme flood events the Pickering Beck catchment is largely close to saturation, and losses to evaporation and soil storage appear to be negligible: all rainfall is translated into runoff. The shallow nature of the soils (discussed in CG2) in the Pickering Beck catchment partly allows this assumption, along with more detailed consideration of the characteristics of the 2007 flood: the amount of rainfall estimated to have fallen on the catchment was similar to the amount of runoff that left the catchment during the storm event, such that little rainfall was lost to soil storage. The primary focus of the model was then to route this rainfall across hillslopes and through the river channel network, while allowing the user to place small bunds within floodplains that can hold water back as long as the storage areas that they define are not full. The model was coded by an academic

member of the team between CGs 2 and 3 and a fuller description of the model is provided in Odoni and Lane (2010). We tried the model out in CG3, which prompted discussion about the wider nature of the modelling process as well as the decision in CG4 to look at the additional issue of sediment and vegetation management in rivers. In turn, we used a different model in CG5 to look at how river maintenance influences water levels and flood risk. The models were both produced by the RFRG but also in themselves became objects through which the RFRG established its own collective competence.

Reflections on our experiment The aim of this section is to draw upon our experience of the RFRG experiment with respect to the practice of flood risk science, notably flood risk modelling, in the context of Callon’s schema of types of public engagement.

Theory building and problem framing The process of choosing a mathematical model for use in flood risk analysis, for a Scientist, is a complex combination of at least three factors: 1 theoretical training 2 experience, which may itself include applications of particular sets of theory through mathematical models and 3 on the basis of our interviews with a sample of flood risk science consultants, the technical devices, or modelling software, that the Scientist has access to, those with which they are comfortable, and those approved by national policy prescriptions. The framing that we brought to the RFRG was not about producing solutions at first. In CG2 we explored the maps of flood risk, produced by the Environment Agency, for the Vale of Pickering. These are maps of what is expected to flood under flows with particular return periods and we wanted local members of the group to criticise them. The critique that resulted was not that the predicted inundation patterns were wrong, but that some of the predicted flooding should not be allowed to happen. When marking up areas where the Environment Agency maps were thought to be wrong, a local member commented: This bit is wrong down here because although it floods, they haven’t done anything about it to [ . . . ] They are


Doing flood risk science differently not maintaining the dykes, and they neglected to clear some of the branches and things. (CG2, November 2007)

A second local member responded: ‘How do you reflect that, though [i.e. draw it on the map]? Because I could say that about Sinnington’. Despite our own initial framing of the experiment, the experiment was forced, from early on, to try to solve this problem in ways that would reduce flood risk in Pickering, and with a focus on upstream storage. As our initial (academic) framing became replaced by that of the group, so we began to co-produce not just outcomes, but the very resources (models) that we would use to sustain those outcomes. As Table IV shows, there was a wealth of expertise distributed across local members that was available for this purpose and the RFRG expressed this bilingually, languages that existed in juxtaposition, unlike the imagined precursor to collective learning of ‘unilingual’ expression.

Experiential, ‘universal’ knowledge Table IV also shows the rich understanding of the fundamentals of hydrology and hydraulics, of

27

Newtonian physics, as possessed by local members of the RFRG. This knowledge was not simply local, as is often portrayed (e.g. Callon 1999), but universal. Local members possessed the expertise necessary not only to guide what the model should do but to contribute to the conceptual development of the model itself. This knowledge was partly experiential but was not acquired through a singular local place. As one of our local members reflected after the experiment, Well, if you go back, what? Thirty, thirty-five, forty, maybe, years? We spent a lot of time in Scotland and in Wales, walking in the hills and so on, and sightseeing. And [**], of course, being an engineer, was very interested in all the hydroelectric schemes and that kind of thing. And having gone with him, and taken the children along, as you do, [ . . . ] I had seen places where you had what I would call a bund, where you stop the water from going somewhere and interfering, so that you kept a regular run of water in one direction . . . And I thought, well, if they could stop the water like that and hold it back behind a small dam, and then let it run through gradually, you know, why not? That was what gave me the idea of the bunds in the first place.

Table IV Shared concepts: modellers’ and Modellers’ modellers’ (local members’) concepts CG1: ‘You think about a gutter and how much can go through it and if it fills up it comes over the top. So if you have got half the size of a gutter, it comes over the top more quickly . . . ’ CG1: ‘Because of course it depends where you are, because if you are closer to the main channel, you have got something rushing really fast. But the rest of it is spreading out . . . it doesn’t tend to be moving at great speeds, it is just sort of spreading out.’ CG2: ‘Logic says that you have got to work out the contours and work out which is the lowest lying land. I suppose there must be some sort of formula to work out exactly the volume of water you are expecting to come down and therefore to what volume it will fill that level.’ CG2: ‘But Pickering is a slower process I think, than Sinnington is. And it is not such a sudden thing. I mean you can see Sinnington rising. I don’t think you can here so much. [because] Pickering Beck goes much further north, and it is gathering more water.’ CG2: ‘The other thing that is important is that we have got heavy clay soil. But the soils vary in different parts, whereas of course the clay soil around Great Barugh means that other areas are more sandy perhaps and drain more easily.’ CG3: ‘So to protect Pickering, the nearer the dams are to Pickering the better?’ [Local member 1] ‘Well yes certainly’ (Local member 2]

Modellers’ (academic flood risk scientists’) concepts The principle of volume conservation for an incompressible fluid

The depth dependence of frictional resistance in riverfloodplain flows The assumption that is central to the diffusion wave approximation of the 2D shallow water equations Potential energy as a momentum source coupled to the principle of volume conservation for an incompressible fluid

Hydrograph attenuation

Infiltration and runoff generation as controlled by soil type

Design of flood storage schemes to remove flood wave peaks


S N Lane et al.

28

The way in which this local member had established their knowledge through an experiential search for the universal is in many ways no different to the way that a scientist assimilates knowledge through working in different places on different problems. Knowledge was also negotiated: both between academic and local members, such as over how high to build the bunds, and pointing to the need to have a large number of smaller ones; and amongst local members, such as over the need to give the bunds an outflow (Table V). It was not simply a case of translating local knowledge into a series of concepts and assumptions to inform model development. The supposed tension between the universal knowledge of Science and the local understandings generated through everyday life (Irwin 1995; Callon 1999) may be more a consequence of how we, as experts, classify expert and vernacular knowledge than the nature of that knowledge itself. Not only are there similarities in the processes by which knowledge is acquired, but also in the content itself. While recognising that the redistribution of expertise and negotiations over knowledge did not extend to the coding of the model in software, they did inform other elements of the modelling process (Table VI): including: 1 where to try thing out in terms of where to make a difference 2 where to try things in terms of the feasibility of delivering solutions (e.g. land management change) 3 how this feasibility might be grown, such as through linkage to wider elements of the rural economy and

4

awareness of the wider institutional context that might shape the eventual solution adopted.

The multitude of choices available in the modelling process was pared back to focus on specifically what might work where. The model was a critical object in identifying where bunds might be most effective, but it was not given the kind of authority normal in flood risk science in which deference to a model’s predictions is the means by which decisions are justified. The model became one part of a new knowledge of the flood system, co-produced through the practice of science.

Very particular models? The bunding model that emerged between CG2 and CG3 was a very particular sort of model. It came out of the RFRG’s wish to re-open the question of upstream storage, but its form was also influenced by a series of concepts and assumptions that help explain the form that the model took (e.g. Tables IV, VI and VII). In so doing, it radically challenged the traditional claims regarding the rationale for a model. Most modellers tend to see models as a form of covering law: general model + local conditions ¼ local predictions There is a supposed universality that allows a model to travel between places for application to local conditions in order to generate local predictions. This is the basic characteristic of both hydrological and hydraulic modelling in flood risk science as reflected in the suite of ‘off-the-shelf’ software used by flood risk consultants (e.g. models and procedures in the FEH; ISIS, Hec-RAS). The

Table V Negotiating knowledge An exchange between an academic member and a local member in CG3 while using the bunding model, the academic member noted: ‘So, you can see that the small dam hardly provides any storage time at all. When you get about 2.5 m it seems to be a threshold . . . if the dam is bigger you get more flood reduction.’ The local member responded: ‘You will get more problems, health and safety and all kinds of things to do with . . . the bigger you go like that, you are running into problems.’ The academic member asked: ‘So how big do you think we should go then?’ The local member responded: ‘2 m I think, but you want a lot of them. I think that is enough.’ In an exchange between two local members in CG3 while using the bunding model, the first noted: ‘The problem with lakes is that a lake will fill up . . . just naturally, so what you want is something that is empty basically, and then fills up when the water comes down’. And the second commented: ‘Yes, but again . . . if you are building a dam and it fills up, and say the rain stopped and it is alright for a week, and then we get another deluge of rain, it is already filled up. What happens then?’ And the first responded: ‘But you have got essentially a temporary dam that is letting water through all the time, so all you have got it holds back for a certain length of time, but you . . . don’t think of it like a reservoir . . . think of it like a reservoir with a big hole . . . so under normal conditions the water goes straight through, and it doesn’t fill up at all, the actual river flow, it can cope with.’



29

Table VI Local knowledge and the wider context In a discussion of what could be done to reduce flood risk in Pickering Beck, a local member (CG2, November 2007) noted: ‘I suggest some sort of storage, reservoir, possibly.’ A second local member responded by pointing to the sub-catchment that Babtie et al. had proposed for a storage reservoir and said: ‘Wasn’t one mooted up here somewhere?’ The first local member responded: ‘This is a very insignificant bit [pointing to the sub-catchment]. It is this one that is the one [pointing to the main stem of the river]. The one that goes up to Lockton is where all the water comes from. This, [pointing to the sub-catchment] it doesn’t matter . . . ’ In a discussion of where bunds might be sited, a local member (CG2, November 2007) noted: ‘And the other problem is that once you get down to Newtondale, you have got the railway there, which means you are causing problems at the railway. They are not going to be pleased about that.’ While using bunding, and trying out different bund locations, an academic member (Break Out Group 1, CG3, January 2008) pointed higher up the main catchment: ‘But maybe up here we can do something’. A local member responded: ‘Once you get right up into Farndale and something it is still agricultural, but it tends to be the sort of sheep farming, rather than the . . . ’ Until interrupted by a second local member: ‘As long as you didn’t pinch all of the bottom meadowland . . . ’ The first local member continued: ‘if it is going to get flooded up there and it is only on a temporary basis for say 2 ⁄ 3 days, it is not going to affect the silage anyway’. The second local member added: ‘The biggest thing of all is if with their agreement, and if necessary compensation, they would go for it straight away because the moor fellows aren’t making a lot of money. No farmer is at present anyway.’ While using the same tool in CG3 (January 2008) a local member in Break Out Group 2 noted: ‘I think we have to be concerned about the visual aspects. Also the wildlife . . . ’ And a second local member added: ‘It is all in a national park.’

Table VII

Examples of material that led to the encoding of bunding by the RFRG

Steer from RFRG

Implications for bunding

While marking up maps of the catchment with what you could try where (CG2, November 2007), local member: ‘Mini reservoirs and then you could then release that at the dry time. Now somebody is talking sense’. Response of second local member: ‘But then you don’t even have to have reservoirs as such. You could have a dam with a restricted outflow so it literally holds it back and releases it slowly . . . so the stream normally just runs straight through it and doesn’t get stopped ...’ During the same exercise (CG2, November 2007), local member: ‘Not necessarily one big dam but preferably a series of small dams.’ In a discussion about what we would need to model (CG2, November 2007), an academic member was explaining the different between hydrological and hydraulic modeling, Academic member: ‘And the hydrology is about working out how big is the discharge in your river. And the hydraulics, is about working, once you have got a discharge in your river, where does that water go on the floodplain?’ Local member responded: ‘Isn’t there speed, because the quantity of water will increase the speed, and the speed that the whole thing will happen’ (local member, CG2, November 2007)

Need to have outflows that let some flow through the bunds so they do not immediately fill and their effects are lost

model we produced was not general, so restricting its ability ‘to travel’ for a number of important reasons. First, it contains assumptions that enabled the

Need to be able to represent multiple bunds simultaneously Need to allow overland flow velocity to increase with water depth and hence event duration

model to be formulated as it was for Pickering, but which restrict its relevance to places (and possibly times) when those assumptions hold. Second, the


30

model’s production challenges those traditionally involved in determining what should constitute the model: the model as a product of conventional scientific activity and modelling as a process of development, undertaken by those who, convention states, have the authority to develop a model. The model could not travel until it had somehow been given greater legitimacy and it is quite possible that, in attempts to acquire greater legitimacy, the model would come across other sources of expertise who might not accept its assumptions. This process was distinctive in another respect. Both hydrological and hydraulic models are not simply applied to local datasets in order to generate predictions; local datasets themselves are actively involved in making a model perform in a particular place. Model simplification inevitably means that a model has to be forced, through calibration, to deliver empirical adequacy (see Oreskes et al. 1994) and to be made to perform (Winsberg 2003; Lane in press). This makes all models specific to particular places and less able to travel. What we have done here is moved the idea that a model should travel to one where the process of model building should travel. By simplifying a model to what is deemed to matter to the place being modelled a priori, including negotiating and evaluating the assumptions being made, we produced a simpler model, one more consistent with the data we had gathered for exploring model predictions, but also more suitable to the kinds of uncertainty analysis commonly resisted in modelling activities (cf. Beven 2002). In the RFRG, the burden of effort shifts from making an off-the-shelf model work in a particular place to developing a model directly suited to that place. The latter process may be much more cost effective than the former and we have estimated that the cost of doing the modelling using our approach may actually be substantially less than the way in which a consultant might conventionally pursue the task (see Ryedale Flood Research Group 2008).

Towards a sense of knowledge One of the primary assumptions of Callon’s PDM is that there is a group of people with knowledge, knowledge that they are willing to use discursively in response to consultation. When we tried this, in CG2, to our surprise, local members found it difficult to articulate their concerns over extant knowledge, in our case flood maps produced by the Environment Agency. They were not a source

S N Lane et al.

of debate, and the maps tended to be taken as given. The debate was over why nothing had been done. This has been observed more widely as a characteristic of public engagement in science, where those who are not certified as scientists tend to remain markedly committed to the authority of Science (e.g. Harvey 2007), as it is constructed as something that has a special position in solving problems in the absence of other interests. The problem is not Science (in our case, the maps produced by the Environment Agency’s consultants) but the failure to address what that Science is showing. However, from CG3 onwards, there emerged a newfound willingness to interrogate the RFRG’s own emerging knowledge, enabled through the practice of doing the modelling, including: 1 the limits imposed on flood risk reduction due to the finite availability of potential flood storage 2 the problem that slowing the water in Pickering Beck might cause the flood peak to become coincident with a flood from one of the upstream tributaries and 3 the fundamental nature of modelling as a practice. One local member concluded (CG3, January 2008), after using the bunding model: ‘You could say then that your model is never right. It is all pure guesswork.’ The RFRG’s own and emerging sense of knowledge went on to make a material difference to the practice of our modelling. Between CGs 4 and 5 we held a reading group to look at how to read and how to critique consultants’ reports. Despite us having introduced flow gauge data in general terms as important in CG2 and in detail for Ings Bridge, downstream of Pickering, as a model input during CG3, the data had never been questioned. During the reading group, we discussed why the consultants had not used it. Local members knew exactly why, something they had known all along: the man has taken the flood banks down . . . and he’s put a bridge across the river at that point, and the bridge is an old articulated lorry body, which enables him to go backwards and forwards with his farm implements when he’s making silage. [So] the information that they would be able to collect from Ings Bridge would be useless . . . Because it wouldn’t be going through the gauge, it would be going over land. (CG4)

The emergence of this information made a dramatic difference to the amount of flow that the



bunds had to reduce and the RFRG needed to remodel the system. The knowledge could have been brought to the RFRG in either CGs 2 or 3. What allowed the sense of knowledge to emerge was the practice of modelling, of co-producing knowledge. What we experienced was a shift from taking knowledge as a given, to being able to see the model predictions as just one set of knowledge, against which they were entitled to compare and to use their own knowledge. In many cases, the knowledge had been there all along (see Table IV, in the early CG meetings). What was different was that the position of, and confidence in knowledge, the sense of knowledge, had changed and we argue that this is a key consequence of co-production. The concern revealed in Table II over the experience of public engagement prior to our experiment could be attributed to a failure to appreciate the presence of this sense of knowledge, one present from early on in the RFRG (Table IV), but which had its own dynamic, worked through the assemblage of people and models in the RFRG itself. One local member reflected after the experiment: You can experience something without any knowledge at all, can’t you? A child can experience, but can still not know the reasoning behind. [pause] As I am a bit . . . I like to know the reasoning behind things.

Co-production, the Scientist and the event Our experiment is distinctive in that the academic researchers involved worked within and through a place and a period (Pickering, June 2007 to October 2008), an event, where it was the event that mobilised and shaped their research practices. This is unusual. Most of the statutory flood risk science in England and Wales is undertaken by specialist scientific consultants, most of whom have had a formal scientific training, and some of whom might have even completed a research degree. Academic scientific engagement with flood risk research is more focused on the development of underpinning technologies, including data acquisition and mathematical models and, to a lesser extent, with some of the wider debates regarding the causes of flood events (e.g. Defra project FD2114 on the possible role of rural land management in flood risk reduction). There is some further engagement through the provision of consultancy services to support research and development contracts awarded to

31

consultants, again often concerned with technological development, and in wider policy development and delivery such as through membership of Regional Flood Defence Committees. Academic scientific involvement in specific flood events is relatively more restricted, either to situations where an event impinges personally on a place (home, place of work)21 or to where, through the documentation of those events, knowledge can be obtained to assist in model development.22 The primary focus of academic flood risk science23 is on methods and model development, where the academic is inevitably separated and insulated from engagement with those impacted upon by flooding. Our experiment challenged several conventional assumptions: who was allowed to be involved in the framing of flood risk in Pickering; what kinds of solutions should be admissible (beyond flood defences); and how it was that Pickering had become a place that had experienced five floods in ten years but which was still one that might not appear on a prioritisation list for flood risk defence until 2018. In the mess of inter-relationships between people (academics, locals, professional partners), institutions (statutory organisations, local councils, government) and things (the river, flood events, properties), the RFRG became a temporary stabilisation of meaning (Carr and McCusker 2009), within which the Scientists involved in the project became scientists. For instance, we note above how initial plans to work with an established hydrological model were abandoned in the light of the experiment; the refocusing of the modelling moved away from managing infiltration and towards trying out upstream storage; the RFRG went through a material transformation from being an element of an experiment funded under a UK research council initiative concerned with different ways of trying out science to actually trying to reduce flood risk in Pickering; the testimony provided by the river gauge, which the academic scientists initially accepted, was questioned, necessitating remodelling of the system. Returning to Callon’s (1999) description of co-production, a much wider constituency was involved in many more elements of knowledge production than in normal flood risk science. Vernacular knowledge formed part of the process of model-building alongside scientific knowledge (Tables IV and VII). Vernacular knowledge was not just taken as given but debated in the same way as the knowledge held by academics (e.g. Table V). Much of this was achieved through


32

practice (e.g. Tables VI and VII). However, although co-production involved both academic Scientists and local members becoming scientists, there were still some roles, of retained expertise, that were left intact. The academic scientists, for instance, had to code the model and handle some of the large datasets required by the model. Local scientists had detailed knowledge of particular places and issues that a Scientist could not possibly bring to their work. The difference was that, in theory, any of these elements could be subject to wider interrogation and the means by which this interrogation happened was the practice of doing flood risk science and not only discussing or debating scientific reports.

Conclusion This paper has provided an account of an experiment in doing flood risk science differently, one that is distinctive because of: the agency given to an event (a flood) in mobilising the working of the associated Flood Research Group; the negotiation and reformulation of knowledge as distributed across both the ‘certified’ and the ‘uncertified’ members of the Group; and the repositioning of academic Science as science. The experiment represents a radical departure from the normal practice of flood risk science through the way in which it repositioned scientists in the process. The experiment revealed the widespread distribution of a deep qualitative understanding of flood hydrology, one that was not simply ‘local’, and which through working with the event could be harnessed to produce and to negotiate a new and collective sense of knowledge. In the case of flood risk reduction in Pickering, the RFRG identified and articulated a possible way forward after a period of impasse, one which is now being explored as part of a national demonstration project. This impasse had emerged because of resource constraints, manifest through the application of costbenefit analysis to particular sets of options. Through the co-production of knowledge in general, and the development of the bunding model in particular, the RFRG was able to subvert those options, identifying new ones for further exploration. The fact that the latter are now being taken forward, and without any change in the resource allocation model, shows how knowledge practices, including the question of how local people are

S N Lane et al.

positioned with respect to those practices, can have a material impact on the nature and form of flood risk management. There is, however, a final theme that emerges from our experiment, relating to the status and standing of bunding, the mathematical model that the RFRG produced. When we went public, the model had not been ‘validated’ in a conventional scientific sense. It had not been compared with any independent data, as none were available, nor with an extant model, as this required boundary condition information not yet collected. Although the model has now been successfully assessed, at the point at which the RFRG chose to go public and to make an intervention, the model was not ‘proven’. If the purpose of the RFRG was to produce a model, then going public at such a point would not have been wise. However, there are two counters to this caution. First, establishing trust in models is not just about the restricted kind of validation implied by comparison of predictions with data, but a wider set of practices through which a modeller makes a model perform (Lane in press). The active engagement of the RFRG throughout the modelling process, as reflected in everything from collective development of a detailed conceptual understanding through to assessing the model critically, was a means of reproducing the process of making a model perform, such that the RFRG had trust in its own model. Second, and more importantly, what the RFRG did, through developing a wider collective competence, was to reach a point where it was possible to intervene in a situation where decisionmaking progress, had largely stopped. The purpose of science, then, had shifted from problem-solving and analysis, in ways that tend to give Science a perceived hegemony in the decisionmaking process, to the practice of science as a means of making a political intervention, making Ryedale and its local community ‘heard’ and unsettling the established positions of institutions and professionals in the decisionmaking process. We were initially steered towards Pickering by the Environment Agency because of the difficulties that they were encountering in finding a way forward for the town, given the limitations placed upon them by national flood defence funding rules, and in the face of repeated flood events. In January 2010, we asked for some feedback from the Environment Agency as to what they felt our intervention had achieved. The Ouse Catchment Manager



commented: ‘this work enabled us to improve our relationship with the community in Pickering and one that will hopefully lead to a successful outcome for all’. Even if the bund solution had not been adopted, the practice of science in the way we describe became a means of enabling the knowledge of those excluded from (even ridiculed in, Table II) the decisionmaking process to be made admissible. In the words of Coeckelbergh (2006), it was a means of growing a sense of ‘moral imagination’ by opening things up. In so doing it challenges the supposed entitlement to professional autonomy of the Scientists and other ‘professional stakeholders’, and their tools and accepted ways of working (Jasanoff 1996), who tend to dominate decisionmaking processes surrounding flood risk and which may be technically and economically efficient but neither just nor fair (Johnson et al. 2007). Working with the hybridisation of science and politics (Jasanoff 2003), putting it to work through an event, rather than working to exclude politics from science in the ways forced through traditional accounts of what scientific method must be to make it a Scientific Method, is what makes this approach particularly radical.

33

4

Acknowledgements This work was funded by grant RES-227-250-018 from the Rural Economy and Land Use programme of three UK research councils (BBSRC, ESRC and NERC) and Defra, awarded to Whatmore, Lane and Ward. We acknowledge critical but constructive comments from three reviewers and an Editor, which resulted in major revisions to an earlier version of this manuscript. We are particularly grateful to local members of the Ryedale Flood Research Group, to Gillian Willis (Oxford University) and to those others who agreed to be interviewed for this research.

Notes

5

6

7

1 This is the official figure. Local people claim the figure to be approaching 100. 2 See http://www.environment-agency.gov.uk/cy/ ymchwil/llyfrgell/cyhoeddiadau/40557.aspx Accessed 10 July 2009. 3 Any flood defence project must be economically viable – i.e. the benefits of the scheme must exceed the associated costs (Ministry of Agriculture, Fisheries and Food [MAFF] 2001). Floods events are treated as stochastic in the sense that an event of a given magni-

8 9

tude has a likelihood or probability of occurrence in any one year and that event will also result in a certain set of economically quantified losses. Thus, the aim is to determine an expected annual loss (MAFF 2001), the probability of loss in any one year multiplied by the value of that loss, and standardised by a discount rate as different schemes may have costs and benefits with different timescales. Hydrological modelling is used to determine the probability of events of different magnitudes, generating a set of river flows that can be applied to hydraulic models. The hydraulic models then determine the spatial patterns of flood inundation associated with each flow for application in an analysis of consequent economic losses. When this analysis is complete, options appraisal begins wherein the economic benefits of different options are tested and the benefits that derive (reduced expected annual loss) are compared with the associated annualised cost of each option. As the determination of expected annual losses is based on the probability of events of different magnitudes and as the length of many flow records is small (c. 96% of England and Wales’ flow records are < 50 years in length), the frequent occurrence of large events can cause increases in expected annual losses and hence a shift in the benefits that might come from investment in flood defence. Similarly, since 2001 there has been recognition that a broader range of benefits need to be quantified (e.g. environmental benefits) and if these cannot be quantified, then decisions should be informed by a wider multi-criteria analysis (see Department for the Environment, Food and Rural Affairs [Defra] 2009). The latter is particularly important for schemes where benefits exceed costs, but where there is a limit to the total resource available to spend on all favourable schemes. Here we use the more general definition of experiment of trying something out rather than the more formal and restricted definition of Science as an Experiment. We expand on this below. We use the phrase ‘flood risk science’ here to reflect that we are focusing on one element of flood risk management decisionmaking: the generation of knowledge about the hydrology and hydraulics of a problem to inform the design of possible risk reduction schemes. This is notwithstanding the widely held view among scientists that science proceeds through falsification. http: ⁄ ⁄ www.deliberative-mapping.org ⁄ Accessed 8 July 2009 This distinction between the general and the particular is recognised in models of scientific knowledge (e.g. the ‘covering law’ model of Hempel, in which a ‘law-like’ or universal statement plus some (particular or local) empirical information is all that is needed to provide an adequate prediction).


S N Lane et al.

34 10 For the purposes of flood risk science, the difference between hydrological analysis and hydraulic analysis is between determining the magnitude and frequency of river flows, or the relationship between the size of river flows and associated rainfall events (hydrological analysis) and determining the relationship between a river flow and water level, and how river flows with a given water level translate into floodplain inundation (hydraulic analysis). 11 Since 1999, these have been appointed for 5-year periods under a Framework Agreement. Until 2009, framework consultants were not required to tender for contracts to do strategic flood risk assessments but were selected from those available by Environment Agency staff. The motivation for this change was the perceived need to achieve national consistency in modelling practice and standards in strategic flood risk assessment. Since 2009, these consultants have to enter a ‘mini-bid’ for contracts priced at greater than £25 000. 12 For instance MAFF recommends: ‘The Flood Estimation Handbook presents new methods for estimating design flows and is highly recommended for its realistic approach to flood estimation’ (2001, 14). Similarly, in 2001, MAFF commissioned a study to benchmark hydraulic models for flood routing and inundation modelling (SC000052 ⁄ 59, Benchmarking and scoping of river hydraulic models). 13 For instance, MAFF commissioned a special study in 2001 (Project SC000042, Reducing Uncertainty in River Flood Conveyance) to reduce uncertainty in the ways that modellers estimate channel conveyance, which directly impacts on estimates of how much water is delivered to floodplains. Not only did this lead to tighter guidelines, it also led to new technical devices (models) to assist in the homogenisation of practice. 14 This toolkit is also the one most frequently referred to by local Environment Agency staff when we have discussed this project with them. 15 The term ‘main river’ is a statutory one used to identify those water courses on which the Environment Agency has the power to carry out flood defence works. Any other channel, drain, sluice etc. that water flows through is called an ‘ordinary watercourse’. The Environment Agency is not entitled to defend properties flooded by ordinary watercourses, and this responsibility has traditionally been attributed to an amalgam of local authorities, internal drainage boards and private water providers. The Environment Agency’s other duties and functions do, however, extent to ordinary water courses. 16 Bunds are small dams, between 0.5 m and 2.0 m in height, placed tangentially to the river across the floodplain but not the river, such that some flow is allowed through the bund, to prevent it from filling too quickly.

17 Source: http://webarchive.nationalarchives.gov.uk/ 20081027092120/http:/www.defra.gov.uk/environ/fcd/ policy/funding.htm Accessed 18 October 2008. 18 Our use of the term experiment in relation to Experiment is an intentional parody of the way in which we describe science as the corollary of Science. 19 These things might include, for example, maps, models, photographs etc. 20 We do not have the space here to contrast these five principles of an Environmental Competency Group with wider forms of public engagement in science. 21 The activities of Tom Coulthard (University of Hull) following the Hull floods of 2007 and Geoff Parkin (University of Newcastle) following the Morpeth floods of 2008 are examples. 22 For instance, the NERC Flood Risk in Extreme Events (FREE) project commissioned a major synthesis of data sources from the UK 2007 flood events that could be used in future flood science research. 23 This is not the case for academic social science, with a number of academics involved in undertaking work with flood risk communities, both pre- and post-flood events, some of this as part of developing new policy approaches (e.g. Middlesex University’s Flood Hazard Research Centre is lead on project SC060019, Improving Institutional and Social Responses to Floods, between December 2006 and January 2009).

References Babtie 2003 Pickering flood alleviation scheme – project appraisal report Report by Babtie Group Ltd No BWA290023 to the Environment Agency of England and Wales August Beven K 2002 Towards a coherent philosophy for modelling the environment Proceedings of the Royal Society of London, Series A 458 2465–84 Callon M 1998 An essay on framing and overflowing: economic externalities revisited by sociology in Callon M The laws of markets Blackwell, Oxford 244–69 Callon M 1999 The role of lay people in the production and dissemination of scientific knowledge Science Technology and Human Values 4 81–94 Carr E R and McCusker B 2009 The co-production of land use and livelihoods change: implications for development interventions Geoforum 40 568–79 Cockerill K, Passell H and Tidwell V 2006 Cooperative modelling building bridges between science and the public Journal of the American Water Resources Association 42 457–71 Coeckelbergh M 2006 Regulation or responsibility? Autonomy, moral imagination, and engineering Science, Technology and Human Values 31 237–60 Collins H M and Evans R 2002 The third wave of science studies: studies of expertise and experience Social Studies of Science 32 235–96


Doing flood risk science differently Costanza R and Ruth M 1998 Using dynamic modeling to scope environmental problems and build consensus Environmental Management 22 183–95 Defra 2004 Community and public participation: risk communication and improving decision-making in flood and coastal defence Research and Development Technical Report FD2007 ⁄ TR, Department of the Environment, Food and Rural Affairs, London Defra 2005 Making space for water: taking forward a new government strategy for flood and coastal erosion risk management in England Defra, London (http:// www.defra.gov.uk/environ/fcd/policy/strategy/1stres. pdf) Accessed 31 March 2008 Defra 2008 Social justice in the context of flood and coastal erosion risk management: a review of policy and practice Defra R&D Technical Report FD2605 ⁄ TR, London Defra 2009 Appraisal of flood and coastal erosion risk management Defra Policy Statement June 2009 Environment Agency 2004a Building trust with communities: a background Report for Environment Agency staff Internal publication Environment Agency 2004b Building trust with communities: a toolkit for staff Internal publication Expert Group on Science and Governance 2007 Taking European knowledge society seriously Report of the Expert Group to the European Commission’s Science, Economy and Society Directorate Brussels: European Commission, Directorate-General for Research Funtowicz S O and Ravetz J R 1991 A new scientific methodology for global environmental issues in Costanza R ed Ecological economics Columbia University Press, New York 137–52 Funtowicz S O and Ravetz J R 1993 Science for the postnormal age Futures 25 740–55 Hale E O 1993 Successful public involvement Journal of Environmental Health 55 17–19 Harrison S and Mort M 1998 Which champions, which people? Public and user involvement in health care as a technology of legitimation Social Policy and Administration 32 60–70 Harvey M 2007 Citizens in defence of something called science Science as Culture 16 31–48 Hunter N M, Bates P D, Horritt M S and Wilson M D 2007 Simple spatially-distributed models for predicting flood inundation: a review Geomorphology 90 208–25 Institute of Hydrology 1999 Flood estimation handbook 5 vols ISBN 0 948540 94 X Irwin A 1995 Citizen science Routledge, London Jasanoff S 2003 Breaking the waves in science studies Social Studies of Science 33 389–400 Jasanoff S 1996 Beyond epistemology: relativism and engagement in the politics of science Social Studies of Science 26 393–418 Johnson C, Penning-Rowsell E and Parker D 2007 Natural and imposed injustices: the challenges in implementing ‘fair’ flood risk management policy in England Geographical Journal 173 374–90

35 Johnson C L, Tunstall S M and Penning-Rowsell E C 2005 Floods as catalysts for policy change: historical lessons from England and Wales International Journal of Water Resources Development 21 561–75 Johnson M S 2009 Public participation and perceptions of watershed modeling Society and Natural Resources 22 79–87 Ministry of Agriculture and Fisheries 1948 Harvest home: the official story of the great floods of 1947 and their sequel MAF, London MAFF 2001 Flood and coastal defence project appraisal guidance: overview (including general guidance) Ministry of Agriculture, Fisheries and Food, London Lane S N in press Making mathematical models perform in geographical space(s) in Agnew J and Livingstone D Handbook of geographical knowledge Sage, London Lane S N, Reaney S M and Heathwaite A L 2009 Representation of landscape hydrological connectivity using a topographically driven surface flow index Water Resources Research 45 W08423 doi:10.1029/2008WR 007336 Supplementary on-line material Latour B 2003 Why has critique run out of steam? From matters of fact to matters of concern Critical Inquiry 30 225–48. Odoni N and Lane S N 2010 Knowledge-theoretic models in hydrology Progress in Physical Geography 34 151–71 Oreskes N, Shrader-Frechette K and Belitz K 1994 Verification, validation and confirmation of numerical-models in the earth-sciences Science 263 641–6 Penning-Rowsell E C, Johnson C and Tunstall S M 2006 ‘Signals’ from pre-crisis discourse: lessons from UK flooding for global environmental policy change? Global and Environmental Change – Human and Policy Dimensions 16 323–39 Rayner S 2003 Democracy in the age of assessment: reflections on the roles of expertise and democracy in public-sector decision-making Science and Public Policy 30 163–70 Royal Society 2006 Science and the public interest The Royal Society, London Ryedale Flood Research Group 2008 Making space for people: involving local knowledge in flood risk research and management in Ryedale, Yorkshire Report of the Ryedale Flood Research Group Stengers I 2005 The cosmopolitical proposal in Latour B and Weibel P eds Making things public, MIT Press, Cambridge MA 994–1003 Stirling A 2008 ‘Opening up’ and ‘closing down’: power, participation, and pluralism in the social appraisal of technology Science, Technology and Human Values 33 262– 94 Sturgis P and Allum N 2002 Science in society: re-evaluating the deficit model of public attitudes Public Understanding of Science 13 55–74 Whatmore S J 2009 Mapping knowledge controversies: science, democracy and the redistribution of expertise Progress in Human Geography 33 587–98


36 Winsberg E 2003 Simulated experiments: methodology for a virtual world Philosophy of Science 70 105–25 Wynne B 1992 Uncertainty and environmental learning: reconceiving science and policy in the preventive paradigm Global Environmental Change 2 111–27 Wynne B 2003 Seasick on the third wave? Subverting the hegemony of propositionalism Social Studies of Science 33 401–17 Yearley S 2000 Making systematic sense of public discontents with expert knowledge: two analytical approaches and a case-study Public Understanding of Science 9 105– 22 Yearley S 2005 Making sense of science Sage, London

S N Lane et al. Yearley S 2006 Bridging the science–policy divide in urban air-quality management: evaluating ways to make models more robust through public engagement Environment and Planning C 24 701–14 Yu D and Lane S N 2006a Urban fluvial flood modelling using a two-dimensional diffusion wave treatment: 1 Mesh resolution effects. Hydrological Processes 20 1541– 65 Yu D and Lane S N 2006b Urban fluvial flood modelling using a two-dimensional diffusion wave treatment: 1 Development of a sub-grid scale treatment Hydrological Processes 20 1567–83
