How to Hide a 395-Acre Oil Terminal Moa Karolina Carlsson PhD Candidate in Architecture Massachusetts Institute of Technology, MIT [email protected] abstract Examining the evolution of Occidental Petroleum’s oil handling terminal on the island of Flotta in Scottish Orkney (1972-1978) reveals how a new planning procedure was invented to justify a controversial industrial development in a seemingly undisturbed scenic landscape. One of the UK’s first Environmental Impact Assessments (EIA) was undertaken with a specific focus on visual amenity. To reduce the visual impact of the 395-acre facility in the island’s flat moor landscape, techniques of computerized visual analysis were developed. The article thus traces the invention of a new digital technology designed to simulate human visual perception by replacing the observing subject with a “seeing” algorithm. The significance of its use in the terminal project is the advent of one of Britain's most widely used tools for planning and development control: Visual Impact Analysis (VIA). Ultimately the article argues that the VIA of the proposed terminal on Flotta served as the official justification for a planning decision that had already been granted. Introduction Since the early 1970s, planners and policymakers have come to recognize scientific methods for making decisions on new developments in the British countryside. This approach has proven particularly fruitful in locations where competing visions of the landscape (for example those held by pro-industry and nature conservation groups) have turned developments into issues of public controversy. Systematic, repeatable methods often produce information that allow decision-makers to legitimate their preferred outcomes by relying on scientific, or pseudoscientific, facts.1 Most established landscape analysis techniques were developed for specific projects such as the siting of power stations, transmission lines, roads and other large projects, and typically involve the use of perspective drawings, scale models and photomontages. Photomontage techniques may include superimposed photographs or projected images in order to present a better visual appreciation of a proposed development scheme prior to construction. More recently, digital computer mapping and graphics are employed to more accurately simulate 1 As Brian Martin and Evelleen Richards point out, this process is typically far from a frictionless. See, Scientific knowledge, controversy, and public decision-making. In Sheila Jasanoff, Gerald E. Markle, James C. Petersen, and Trevor Pinch (Eds.), Handbook of Science and Technology Studies (Newbury Park, CA: Sage, 1995): 506-526. 1 the visual intrusion of a development in a landscape. These picture-making techniques are mainly designed to enable impacts to be visualized but do not justify the location of the development. This article focuses on the evolution and early experiments with a different kind of digital technology that promises to simulate without subjective bias visual changes in landscape caused by a new development – a claim that gives the produced outcome mandate to serve as proxy for human visual perception in the planning control process. Visual Impact Analysis (VIA), as the technology is commonly called, involves a procedure invented to precisely and geometrically determine and control who can see what from defined locations in the built environment. It produces maps not only of landscape but renders visible the extent (uninterrupted area) of a human observer’s visual field, positioned therein. Consistent with Daston and Galison’s notion of “mechanical objectivity,”2 this endeavour deliberately replaces the observing subject (and thus aims to get rid of inevitable bias and unpredictability) with an explicit and repeatable procedure that allows the “nature [of landscape] to speak for itself.”3 It measures, so to speak, the quantity of “seeing” or of visual changes allegedly without the bias characteristic of the empirical gaze. The adoption of this technology in countryside planning ultimately renders the act of seeing landscape as a disembodied, veridical, and realistic process of classifying the world according to that “in view” and that “out of view.” Challenging this reduction of human visual perception into a binary condition (digitally represented as 0/1), this paper illuminate that questions of who can see what from where in the built environment --- or more recently: what can see what4 --- for what purpose, and for whose benefit or gain is arguably a through-and-through political project with very high stakes, particularly for the individual. In my formulation, the possibility for 2 Lorraine Daston and Peter Galison, “The Image of Objectivity,” Representations 40 (1992): 81–128; Lorraine Daston and Peter Galison, Objectivity (New York: Zone Books, 2007). 3 Daston and Galison, “The Image of Objectivity” (1992, 81). 4 Today, integration of technologies of this type is becoming more widespread and invisible in different sectors of society. Automated or semi-automated processes of machine and computer vision are used, for example, in surveillance and predictive analytics, in law enforcement to identify suspected criminals, and in developments of self-driving vehicles. In design and environmental planning, a semi-automated procedure known as Landscape and Visual Impact Assessment (LVIA) has become a widely adopted tool to analyze potential visual impacts to landscape and landscape views resulting from proposed developments or management actions. (See, for example, Guidelines for Landscape and Visual Impact Assessment 3rd edition, 2013.) Similarly, so called, viewshed or isovist analysis are common functions of most geographic information system (GIS) software. The technology is often used to assist in planning deliberations and to preserve strategic view corridors in the built environment. 2 individuals to independently interpret visual stimuli is not simply a cognitive or artistic idea but one arguably essential to individualism, transgression of norms, and delight.5 To substantiate this claim, I examine how techniques of visual analysis and simulation were invented and put to use in one of the UK’s first Environmental Impact Assessments (EIA)6 undertaken for the development of a 395-acre oil handling terminal on the island of Flotta off the northern coast of the Scottish mainland. While an EIA is designed to consider many different aspects of landscape, the process known as and Visual Impact Assessment (VIA) is put in place to “ensure that effects of change and development both on landscapes and on landscape views, are taken into consideration in formal decision-making processes.”7 However, as Brian Clark et al point out, there is an intrinsic difference between impact analysis on the one hand and assessment and evaluation on the other.8 In EIA matrices are used to systematically record environmental impacts but these matrices are themselves not comparable. Arguably, one main idea of EIA was that the procedure would complement cost-benefit-analysis by providing means of assessing non-quantifiable environmental factors. Thus, the task still remains of comparing and judging one matrix against the others, the only way which is trained judgement. In the case of VIA the distinction between “mechanical objectivity” and trained judgement is even less clear because the digital procedure has already been programmed to undertake a particular way of “human” seeing. The central problem in focus is thus not that digital computers replaced human beings in 1970s planning debates, although this was arguably often the case, but rather that digital “seeing” became understood as empirical measurement processes that sufficiently epitomize human vision. The consequence of replacing an observing subject with a seeing algorithm is that human interpretation and evaluation of visual content, what Nelson Goodman calls “worldmaking,”9 5 Leon Battista Alberti, De Architectura (“On architecture”, published as Ten Books on Architecture). 6 The basis of Environmental Impact Assessment is a procedure that seeks to ensure that the environmental implications of decisions are taken into account before the decisions are made. 7 In the UK, the current approach and methodology to visual impact assessment is set out in a Department of Environment (DoE) ‘Blue Book’, entitled Guidelines for Landscape and Visual Impact Assessment (GLIVA). The first edition of these guidelines, GLIVA1, published in 1995, was co-produced by the Landscape Institute and the Institute of Environmental Management and Assessment. 8 Brian Clark et al. 1981. Assessment of Major Developments. 9 Ways of Worldmaking (1978). 3 come to play no role in matters that ultimately concern the public. This gives rise to a dilemma where, despite the rhetoric of visual perception that surrounds VIA and its claim to simulate what a person can see from a defined location, human beings are effectively removed from decision- making processes. As I will demonstrate, this was indeed often an explicit goal of the early computerized experiments, or as Mark Turnbull and Graeme Aylward who developed digital computing aids for the Flotta terminal put it: “any approach for assessing the visual impact of a new development should remove the uncertainty of individual judgement so far as the presentation of evidence for visibility is concerned.”10 Thus, this article examines the emergence of a new vision-based technology that promised to quantify changes in visual amenity11 and to simulate visual prospects