Levi John Wolf, Alison Heppenstall, and Rich Harris
A Research Agenda for Spatial Analysis (RASA) is a collection (approximately) 5000 word perspectives on the future of spatial analysis published by Edward Elgar Books.
Back to Berry and Marble’s early Spatial Analysis handbook, handbooks in quantitative geography or spatial analysis have provided a venue to take a snapshot of our field as it is. And, while they are produced often, it is both an honor and a rare opportunity to contribute to these Handbooks. More rare than this, however, are opportunities to envision a future for spatial analysis.
In this edited volume, A Research Agenda for Spatial Analysis (RASA), authors stake a claim on the future of their part of spatial analysis, writing a “manifesto” for the future of the area. These short chapters (approximately 5000 words, about as long as a Progress in Human Geography “Progress Report”) are intended to be clear, opinionated, and speculative perspectives on the future of a concept or a domain of practice in spatial analysis.
The edited volume is split into two halves. In the first part, senior researchers reflect on the future of “big ideas” or “concepts” that have been central to their research agenda, as well as any classics that weren’t: concepts, books, or papers that have been particularly influential in your thinking but that are not quite as prominent as the ought to be. The second half of RASA contains essays from a variety of scholars who write about their chosen domain of expertise.
Table of Contents
Essentially Contested Concepts in Geography
Geographical analysis involves a wide set of academic domains that all share a set of central concepts, concerns, and practices. Chief among these are space, place, scale, and pattern. More recently, reproducibility has re-emerged as a core concern for geographic analysis. This section of A Research Agenda will present how these concepts are conventionally represented and applied. Further, this section will discuss how these concepts may evolve as the discipline itself changes over time.
Click on the titles below to expand the chapter abstracts.
"Pattern," Trisalyn Nelson (UC Santa Barbara)Geographers often describe the “structure” of a geographical process as exhibiting pattern when attempting to represent geographical processes in statistical models. But, what each geographer means, exactly, by pattern has historically been contentious, and some patterns are more frequently analyzed, conceptualized, and problematized than others. This chapter will discuss different notions of pattern in geographical analysis, and take a perspective on how pattern might evolve in new areas of geographical study.
"Reconstructing the Map," James Cheshire (UCL)Denis Cosgrove wrote ‘the only true map is the territory itself’ (Cosgrove, 2008, p. 168), and yet the idea of maps as impartial scientific impressions of the world is a pervasive (and persuasive) one. In many contexts we have no choice but to trust maps: they navigate us off hillsides, route us to unfamiliar places and are the basis to life and death decisions in times of crisis. They offer certainty at times of uncertainty, so we don’t want to hear that they may be wrong. This perspective is also essential for those who critique maps – the presumption that the cartographer believes they are always correct makes it much easier to point out the mistakes and partiality. This chapter reflects upon the current and future status of maps in spatial analysis, considering what they are now and how we can possibly realize their full potential for social justice and spatial analysis.
"Space: towards a global sense of place" Luke Bergmann (UBC) & David O'Sullivan (Te Hereng Waka | Victoria University of Wellington)A spectre is haunting spatial analysis—the spectre of geography. But doesn’t spatial analysis study geography? Spatial analysis is often a matter of analysis of phenomena over space. This analysis presupposes a certain kind of space, an independent space not entangled with the phenomena under question. Yet geography also studies space, and not only empirically, but conceptually. What if understanding space, both conceptually and empirically, was itself a step in spatial analysis?
"How to solve the scale problem in spatial analysis," Stewart Fotheringham (Arizona State University)Although geographers are perhaps most familiar with scale as a cartographic concept relating the size of objects in an image or on a map to their real-world dimensions, the notion of scale has much broader interpretations across a vast range of disciplines. Scale is a fundamental concept in spatial analytical research, in whatever discipline such research is promulgated, and has been the cause of much concern ever since the genesis of the field. The notion of scale in spatial analysis incorporates such diverse issues as the geographic frame of reference for the analysis, the definition of the areal units for which spatial data are reported, and the geographic domain over which processes vary. All three conceptual views of scale are also sources of problems for the spatial analyst. This chapter shows how the three problems are related and can be better understood by moving away from the traditional view that they are a product of data properties and recasting them in terms of the properties of processes.
"Reproducibility," Chris Brunsdon (Maynooth University)Reproducibility is an important contemporary concept in spatial analysis. For a published article or report, reproducibility is the ability of a third party to reproduce any numerical or graphical output appearing in the report. Typically this can be thought of as extension of ‘traditional’ publication. In a traditional publication, a written article is available. In a reproducible publication the written article is still available, but in addition any data used in the analysis, and the computer code used to perform the analysis are also available. In this chapter, I outline the need for reproducibility in spatial analysis, focusing on a few of the specific advantages that a reproducible spatial analysis will bring to geography. I then discuss a few ways how reproducibility can be improved in spatial analysis, and also present a few tools and frameworks useful for reproducible and open spatial analysis.
Agendas for Domains of Practice
Geographical analysis, as a domain in itself, involves many different modes of inquiry. Often, these quantitative analyses are distinguished by their methods, but also sometimes by their aims, purposes, and focus of study. In this section of A Research Agenda, authors will discuss the current state-of-the-art of their field. They will also give a perspective on the future of these areas.
Click on the titles below to expand the abstracts.
"Geographic Data Science," Daniel Arribas-Bel (University of Liverpool) & Anita Graser (AIT)Geographic Data Science (GDS, (G. Andrienko, Andrienko, and Weibel 2017; Singleton and Arribas-Bel 2021)) is broadly defined as the space where Geography and, in particular, GIScience (Goodchild 1991), intersect with Data Science (Donoho 2017). As such, GDS has emerged around a community trying to solve challenges with new forms of data, computational approaches, and where there is a clear spatial/geographic (and often temporal, (G. Andrienko, Andrienko, and Weibel 2017)) dimension in the answer. Neither of these three elements are new in themselves, and hence GDS does not claim originality in each of such aspects individually. In fact, an explicit goal of its agenda is ensuring that existing contributions of each field are appropriately communicated across the disciplinary divide. In this chapter, we define geographic data science, focusing particularly on its necessity for contemporary spatial science. Then, we provide some discussion of the challenges and critiques for geographic data science, as well as the substance of of what geographic data science will look like in the future.
"Causal Inference," Gareth Griffith (University of Bristol), Gwylim Owen (University of Liverpool) & Meng Le Zhang (University of Sheffield)
The most interesting and useful questions in spatial science are often truly about cause and effect: if we do X what will happen to Y. If spatial data scientists care about better understanding society and designing policies to improve it, we must concern ourselves with why spatial processes function as they do, not simply how. We propose doing so requires greater focus on research design, not ever more arcane spatial methods or datasets.
There is no doubt that ongoing developments in spatial data analysis and computational capacity have enabled a breathtakingly interdisciplinary scope of spatial enquiry. However, the causal questions they tend to answer are very limited, perhaps just shy of being deemed “mostly pointless”. Rather than continuing to lament this, we outline a manifesto for why spatial analysts should care about causal inference, and provide guidelines for the converted.
In essence, understanding causal relationships requires us to create, assume or discover random variation. It is not usually practical or ethical for spatial analysts to create random spatial assignment, so we commonly rely on assuming randomness (via exchangeability) using statistical adjustment. Better yet, we may search for so-called natural or quasi-experiments to remove the need to rely on statistical assumptions. We do so by identifying situations where someone (sometimes ‘God’; but more often a bureaucrat) has, without realising, cast geographically informed dice which allow us to investigate spatial processes.
We demonstrate the benefits of explicitly considering causal research questions, rather than avoiding the “C-word” (and furtively hoping readers infer it regardless). Throughout, we provide examples of studies which have explicitly exploited such designs, from considering the impact of neighbourhood homicides on exam results, to the impact of localised restrictions on COVID-related mortality. We use these studies as a roadmap to identify the most fruitful future research avenues for spatial analysts.
"Generative Modelling," Clementine Cottineau (TU Delft)Generative modelling has entered the field of geography and spatial analysis some 35 years ago. Besides allowing spatial analysts and geographers to build operational models for transportation and planning, it has represented the opportunity to take causal inference from the traditional (statistical) analysis of empirical models to the design of causal mechanisms simulated in virtual environments. It is a major epistemological shift, whose scientific contribution still needs advocacy in the wider community of geographers and spatial analysts. In parallel, the expansion of both computing power and available data have made it easier for isolated teams and individuals to build ad-hoc models fit for their specific research questions, leading to a cacophonous development of generative models unrelated to one another. Since the introduction of generative modelling in geography and spatial analysis, hundreds of models have been developed: all may have been wrong, but some were surely useful. However, most models built over the years have been abandoned, their program either inoperable with today’s technology or lost entirely In 35 years, this should strike us as vastly wasteful of ideas, time and energy. In this chapter, I will lay out possible paths towards sustainable, modular, and painless generative models, and the expected impacts in terms of geographical theory development and scientific reproducibility. In this way, generative modelling done right can both contribute to a new form of causal inference and to the larger programme of social sciences: the simultaneous search for generalised explanations of social phenomena and recognition of the uniqueness of historical events.
"Machine Learning," Stephen Law, Yao Shen, and Chen Zhong (UCL)The field of artificial intelligence have expanded rapidly in recent years permeating to many application domains including medical science, climate science, finance, and geography. In Geography, these advances have culminated in the new subdomain of GeoAI which was driven by advances in deep learning, optimised computational tools and the availability of large scale spatially embedded data. In this chapter, we will describe a couple of techniques in deep learning for analysing image, text, graph and point data. We will then provide some projections on the near future for the topic, including increasing application of deep learning on traditional geographical problems, open data practices as well as cross disciplinary engagements and teaching. We envisage the use of deep learning in geography will continue to grow leading to hopefully new spatial insight, knowledge and methods to be discovered in the future.
"Earth Observation," Michelle Stuhlmacher (DePaul)Earth Observation—the gathering of information about the planet via in-situ and remote sensing technologies—has allowed us to look at the Earth in new ways. The view from above provided by aerial and satellite imagery has been an essential source of spatial information in areas where we have very little data, such as out in the ocean or remote reaches of the rainforest. With the advent of big data, however, there has been a shift toward utilizing Earth Observation (EO) technologies in areas where data are abundant. Urban areas, in particular, have large quantities of spatial data that often remain unleveraged or underutilized. This chapter focuses on the future of EO in urban areas: a technological and analytical frontier that has the potential to improve the lives of over half of the world’s population and the environment we all rely on.
"Integrated Science of Movement," Urska Demsar (St. Andrews)Recent years have brought unprecedented advances in movement data acquisition and movement is now being analysed in such disparate disciplines as human geography, computer science, transportation research and movement ecology. While mathematical concepts for movement analysis are the same across all disciplines, there still remains a barrier for sharing methods, despite similar research objectives. Recently attempts have been made to bridge this gap and establish an overarching interdisciplinary science, the Integrated Science of Movement. This essay introduces this initiatve. In the first part, I provide an interdisciplinary overview of contemporary movement analytics. In the second part, I discuss challenges arising from both scientists’ views on interdisciplinary work and from new developments, including new sensors and types of data. I further outline how spatial analysis, as part of Geographic Information Science (GIScience) and with its focus on space-time, could play an integral part in this exciting new interdisciplinary science.
"Spatial Interaction Modelling: A Manifesto" Francisco Rowe (University of Liverpool), Robin Lovelace (University of Leeds), and Adam Dennett (UCL)Spatial interaction models (SIMs) are a core tool in spatial data modelling to predict spatial flows and understand their underpinning factors. SIMs have been applied to provide data insights and support decision making in multiple settings, notably in transport, human mobility, migration, and epidemiology. While considerable progress has been made on advancing the theoretical and methodological underpinnings of SIMs, key challenges remain to facilitate the application of SIMs, extend existing modelling approaches, and leverage the opportunities afforded by big data. We identify three key challenges: reproducibility, calibration, and Big Data Modelling. We propose a blueprint to tackle these challenges by identifying our areas of development: (1) to enable essential infrastructure to facilitate the training, calibration, and reproducibility of SIMs; (2) to embrace modelling frameworks to capture spatial, temporal, and population heterogeneity; (3) to enhance statistical inference to accommodate big data analysis; and (4) to integrate data science approaches to enhance SIM-generate predictions and statistical inference.
"The neighbourhood: where Wilson, Schelling, and Hägerstrand meet" Ana Petrović, Maarten van Ham (TU Delft), and David Manley (University of Bristol)There is a longstanding interest in the causes and consequences of socio-spatial inequalities in cities. A large literature has emerged on so-called neighbourhood effects, which seeks to understand how living in neighbourhoods of concentrated poverty affects a range of individual outcomes, such as health, income, education and general wellbeing (Galster, 2012). for both theoretical and empirical reasons, the term neighbourhood effects should be replaced by the more encompassing term ‘spatial context effects’, as many of the assumed spatial effects are not confined to residential neighbourhoods and the contestable meaning of neighbourhood distracts (Petrović, Manley, & van Ham, 2019). Despite the substantial advances that have been made in defining and measuring the spatial context of individuals, we argue that to further our understanding of spatial context effects it is necessary to go much further. In this chapter, we outline how weaving together different contemporary strands of thinking about socio-spatial inequalities can help us develop an approach to better understand spatial context effects.
The full draft is in production, expected Spring 2024.