The twin conferences AGIT and GI-Forum took place in Salzburg three weeks ago, complemented by the German language FOSSGIS conference. This fully packed conference week had a lot to offer (see my Twitter diary ) and definitely was an inspiring week. With a short time lag in between I’d like to reflect on a topic that popped up at various occasions and is very relevant to my PhD project: geography’s contribution to mobility and transport (research).
To make it very short, these are my key take home messages:
- The geo-space is very powerful in integrating various data/information layers and facilitating holistic approaches for research, planning and operation.
- Technology driven arguments are annoying. It’s always about people.
- Geography supports system thinking, which is required in any mobility and transport topic.
Harvey Miller from Ohio State University opened the GI-Forum conference with his keynote on “Big Data for Healthy Places”. Referring to Pollocks article in Nature , Harvey made a strong case for how the built environment affects mobility and subsequently public health. In his keynote Harvey identified two major challenges in the context of healthy cities: firstly, cities, which are human systems, are complex systems and secondly, policy interventions can have unclear or even counter-intuitive outcomes. In order to tackle these challenges, Harvey proposed what he termed Geographical Information Observatories (GIO), which facilitate opportunistic GIScience. A GIO is a way to constantly monitor certain areas or phenomena and link the sensed data to other data or information sources. Here, the geographical coordinate plays a central role as common denominator for all data or information layers (‘spatial index’). So called urban dashboards (such as CURIO ), which are fueled by GIOs, are the basis for opportunistic GIScience, a framework for spatial science which is able to adapt to spontaneous events, combine real-time with historic data and to simulate planned interventions in a virtual environment. This way, complex systems can be studied, monitored and influenced in a naturalistic setting and intended measures can be tested for their effect on the whole system prior to implementation.
Some of the keynote’s topics had already been discussed before in an interesting panel discussion on the relation between GIScience and Data Science, organized by Peter Mandl . Besides Harvey, Petra Staufer-Steinnocher and Josef Strobl discussed as panelists.
Peter argued for the integration of recent developement in GIScience, namely linked data, open data and semantics, into “Spatial Data Science”.
Harvey made two crucially important points: Data scientists tend to go for correlations (predicting and control paradigm) instead of focusing on causalities in complex systems; for the latter domain experts are needed who interpret correlations in the respective (spatial) context and transform data into information. Conceptually related to this observation, Harvey pointed to the fact that not all decisions should be made quick and purely data- or algorithm-based (the reference to the Jevons paradox is highly interesting in this context). This critical statement is often missing in Smart Cities debates!
gicycle (@gicycle_) July 06, 2016
Similar to Harvey, Josef made a few conceptual statements, which are often overlooked in “data-positivistic” discussions. In his opinion, correlations and pattern detection are only ways to make sense of massive data (streams); they have little value for themselves but act as filters and hypothesis generators. Again, he underpinned the role of domain experts, who are indispensable when exploratory studies are lifted to explanatory ones. In analogy to this conceptual difference and referring to the relation of GIScience and Data Science, Josef stated, “Data leads to explorations, science leads to findings”.
Being affiliated with the Vienna University of Economics, Petra put a focus on Business Analytics (which has, of course, a lot in common with Data Science!) and called for a tight coupling of data driven approaches to theory-based science. In her opinion, Business Analytics is currently too often only about dehumanizing people (clients) and turning them into data.
On Wednesday Anita Graser kicked-off the German language AGIT conference with her keynote on “OpenSource, OpenData and OpenScience”. In the afternoon I first attended a session on Urban Geoinformatics (I co-authored one of the presented papers ), which was nicely wrapped up by Joao Porto . He stated very clearly that Urban Geoinformatics is the intersection of people (urban), technology (informatics) and place (geo). This rather simple definition is blanked out ways to often in current discussions!
After that, this year’s special session on GIS-T (“Spatial perspectives on transport systems” ) took place with three excellent presentations and lots of discussion. The session was opened with a session keynote by Harvey Miller, who provided an overview of the role of GIS in transport (research). Referring to his article from 2015, Harvey talked about the fast changing environment of our discipline (presentations slides are available here ):
- Data availability and computational power have been increasing constantly over the last years.
- Despite the predicted abolition of space through the Internet, progressive urbanization is changing the human sphere radically (urban metabolism ).
- The success of the smart phone, which is constantly connected to the Internet, facilitates new applications, methods for data capturing and business models; most of them are location-based.
The other two contributions to the session were rather technical: Mario Dolancic, the winner of this year’s student paper award, presented an approach for lane detection from floating car data. Mario’s motivation for his work, which is part of the LaneS project, was humorous, “I’m a student and don’t have the money. But I want a realistic road graph.”
gicycle (@gicycle_) July 06, 2016
Anita Graser provided insights in current algorithms for realistic pedestrian routing across open spaces and presented an efficient approach for OpenStreetMap data (for more information visit the PERRON project website).
Thursday was like a roller coaster ride. The day started and closed with sessions on authoritative spatial (transport) data. I had never expected to attend a GIP forum where the majority of contributions discussed how authoritative data can be made available to the public. The digital road graph can be completely downloaded via data.gv.at . This rich dataset can be nicely coupled with national address data that were made available just recently. In the afternoon OGD strategies on various administrative levels were discussed in a GeoTalk (the presentation slides have been made available on the organization’s website , scroll to GeoTalk #10), organized by the local GIS cluster.
In between these sessions I attended a special forum on autonomous driving. Although some of the contributions where innovative and relevant (for instance Benno Bock’s presentation on car sharing patterns ), the forum was dominated by automotive lobbyists who demonstrated a very narrow perspective on mobility. It was a bit frustrating to see how much money is put into R&D with an exclusive focus on the car. There is little effort to completely re-think mobility as a system. Here is just one example: Graham Smathurst from VDA was asked how to understand BMW’s slogan “Freude am Fahren” (pleasure in driving) in times of autonomous vehicles. His answer spoke volumes: On Monday when he drives to work and roads are congested (!) he prefers the autonomous mode, while on a sunny Sunday afternoon he enjoys to drive himself. There was not a single trace of rethinking commuting patterns or mobility behavior. Nothing. Similarly, Christian Kleine from Here presented the company’s ambitions and technology, illustrated with a picture of a self driving car in a massive congestion.
The sessions I attended on Friday nicely demonstrated the potential of the spatial perspective and GIS technology in models, applications and participatory planning processes:
- Enrico Steiger gave an update of the excellent OpenRouteService .
- Nikolaus Krismer gave a presentation on his PhD project about multimodal isochrone calculation.
- Stefan Herbst demonstrated the Mobility Optimizer , a multi-layer information and analysis tool for evidence-based (planning) decisions.
- In the very last presentation of the conference, Dennis Groß presented his thesis where he combined bio-physical sensor data with locations and produced maps of increased stress for cyclists.
What all these contributions have in common, is the added value of an explicit consideration of spatial information. And because transport systems and mobility are spatial by their very nature, geography has a lot to contribute to a better understanding of these complex and dynamic systems. This is why we will definitely organize another GIS-T session for the GI-Forum conference next year. It would be great if you could consider this in your publication and dissemination plans for 2017 (the CfP will be published in December 2016).