Reflections on Eye on Earth summit: the integration of GIS in Environmental Information Systems

The Eye on Earth Summit took place in Abu Dhabi on the 12 to 15 December 2011, and focused on ‘the crucial importance of environmental and societal information and networking to decision-making’. The summit was an opportunity to evaluate the development of Principle 10 from Rio declaration in 1992 as well as Chapter 40 of Agenda 21 both of which focus on environmental information and decision making.  The summit’s many speakers gave inspirational talks – with an impressive list including Jane Goodall highlighting the importance of information for education; Mathis Wackernagel updating on the developments in Ecological Footprint; Rob Swan on the importance of Antarctica;  Sylvia Earle on how we should protect the oceans; Mark Plotkin, Rebecca Moore and Chief Almir Surui on indigenous mapping in the Amazon and man others. The white papers that accompany the summit can be found in the Working Groups section of the website, and are very helpful updates on the development of environmental information issues over the past 20 years and emerging issues.

Interestingly, Working Group 2 on Content and User Needs is mentioning the conceptual framework of Environmental Information Systems (EIS) which I started developing in 1999 and after discussing it in the GIS and Environmental Modelling conference in 2000, I have published it as the paper ‘Public access to environmental information: past, present and future’ in the journal Computers, Environment and Urban Systems in 2003.

Discussing environmental information for a week made me to revisit the framework and review the changes that occurred over the past decade.

First, I’ll present the conceptual framework, which is based on 6 assertions. The framework was developed on the basis of a lengthy review in early 1999 of the available information on environmental information systems (the review was published as CASA working paper 7). While synthesising all the information that I have found, some underlying assumptions started to emerge, and by articulating them and putting them together and showing how they were linked, I could make more sense of the information that I found. This helped in answering questions such as ‘Why do environmental information systems receive so much attention from policy makers?’ and ‘Why are GIS appearing in so many environmental information systems ?’. I have used the word ‘assertions’ as the underlying principles seem to be universally accepted and taken for granted. This is especially true for the 3 core assumptions (assertions 1-3 below).

The framework offers the following assertions:

1. Sound knowledge, reliable information and accurate data are vital for good environmental decision making.
2. Within the framework of sustainable development, all stakeholders should take part in the decision making processes. A direct result of this is a call for improved public participation in environmental decision making.
3. Environmental information is exceptionally well suited to GIS (and vice versa). GIS development is closely related to developments in environmental research, and GIS output is considered to be highly advantageous in understanding and interpreting environmental data.
4. (Notice that this is emerging from combining 1 and 2) To achieve public participation in environmental decision making, the public must gain access to environmental information, data and knowledge.
5. (Based on 1 and 3) GIS use and output is essential for good environmental decision making.
6. (Based on all the others) Public Environmental Information Systems should be based on GIS technologies. Such systems are vital for public participation in environmental decision making.

Intriguingly, the Eye on Earth White Paper notes ‘This is a very “Geospatial” centric view; however it does summarise the broader principles of Environmental Information and its use’. Yet, my intention was not to develop a ‘Geospatial’ centric view – I was synthesising what I have found, and the keywords that I have used in the search did not include GIS. Therefore, the framework should be seen as an attempt to explain the reason that GIS is so prominent.

With this framework in mind, I have noticed a change over the past decade. Throughout the summit, GIS and ‘Geospatial’ systems were central – and they were mentioned and demonstrated many times. I was somewhat surprised how prominent they were in Sha Zukang speech (He is the Undersecretary General, United Nations, and Secretary General Rio +20 Summit). They are much more central than they were when I carried out the survey, and I left the summit feeling that for many speakers, presenters and delegates, it is now expected that GIS will be at the centre of any EIS. The wide acceptance does mean that initiatives such as the ‘Eye on Earth Network’ that is based on geographic information sharing is now possible. In the past, because of the very differing data structures and conceptual frameworks, it was more difficult to suggest such integration. The use of GIS as a lingua franca for people who are dealing with environmental information is surely helpful in creating an integrative picture of the situation at a specific place, across multiple domains of knowledge.

However, I see a cause for concern for the equivalence of GIS with EIS. As the literature in GIScience discussed over the years, GIS is good at providing snapshots, but less effective in modelling processes, or interpolating in both time and space, and most importantly, is having a specific way of creating and processing information. For example, while GIS can be coupled with system dynamic modelling (which was used extensively in environmental studies – most notably in ‘Limits to Growth’) it is also possible to run such models and simulations in packages that don’t use geographic information – For example, in the STELLA package for system dynamics or in bespoke models that were created with dedicated data models and algorithms. Importantly, the issue is not about the technical issues of coupling different software packages such as STELLA or agent-based modelling with GIS. Some EIS and environmental challenge might benefit from different people thinking in different ways about various problems and solutions, and not always forced to consider how a GIS play a part in them.

Some important questions about the usability of geospatial technologies

At the beginning of May, I gave a lecture at the UCL Interaction Centre (UCLIC) seminar titled ‘Interacting with Geospatial Technologies – Overview and Research Challenges’. The talk was somewhat similar to the one that I gave at the BCS Geospatial SIG. However, I was trying to answer a question that I was asked during a UCLIC seminar in 2003, when, together with Carolina Tobón, I presented the early work on usability of GIS for e-government applications. During that talk, the discussion was, as always is in UCLIC, intensive. One core question that remained with me from the discussion was: ‘What makes geospatial technology special or is it just another case of a complex and demanding information system that you should expect difficulties with and spend time to master?’

Over the years, I have been trying to improve the answer beyond the ‘it’s special because it’s about maps‘ or ‘geospatial information comes in large volumes and requires special handling‘ or similar partial answers. In the book Interacting with Geospatial Technologies different chapters deal with these aspects in detail. During the talk, I tried to cover some of them. In particular, I highlighted the lag of geospatial technologies behind other computing technologies (an indication of complexity), the problems of devices such as SatNavs that require design intervention in the physical world to deal with a design fault (see image), and the range of problems in interfaces of GIS as were discovered in the snapshot study that was carried out by Antigoni Zafiri.

There was an excellent discussion after the presentation ended. Some of the very interesting questions that I think need addressing are the following:

• In the talk, I highlighted that examples of spatial representations exist in non-literate societies, and that, therefore, the situation with computers, where textual information is much more accessible than geographical information, is something that we should consider as odd. The question that was raised was about the accessibility of these representations – how long does it take people from the societies that use them to learn them? Is the knowledge about them considered privileged or held by a small group?

• For almost every aspect of geospatial technology use, there is some parallel elsewhere in the ICT landscape, but it is the combination of issues – such as the need for a base map as a background to add visualisation on top of it, or the fact that end users of geospatial analysis need the GIS operators as intermediaries (and the intermediaries are having problems with operating their tools – desktop GIS, spatial databases etc. – effectively) – that creates the unique combination that researchers who are looking at HCI issues of GIS are dealing with. If so, what can be learned from existing parallels, such as the organisations where intermediaries are used in decision making (e.g. statisticians)?

• The issue of task analysis and considerations of what the user is trying to achieve were discussed. For example, Google Maps makes the task of ‘finding directions from A to B’ fairly easy by using a button on the interface that allows the user to put in the information. To what extent do GIS and web mapping applications help users to deal with more complex, temporally longer and less well-defined tasks? This is a topic that was discussed early on in the HCI (Human-Computer Interaction) and GIS literature in the 1990s, and we need to continue and explore.

In my talk I used a slide about a rude group in Facebook that relates to a specific GIS package. I checked it recently and was somewhat surprised to see that it is still active. I thought that it would go away with more recent versions of the software that should have improved its usability. Clearly there is space for more work to deal with the frustration of the users. Making users happy is, after all, the goal of usability engineering…

GIS Systems ?!? from acronym to a term…

Yesterday, for the first time, I came across the phrase ‘GIS Systems’ in an academic paper, written by geographers (not GIS experts). I have also noticed that the term is being used more often in recent times when people talk about packages such as ArcGIS or Mapinfo.

On the face of it, talking about a ‘GIS System’ is ridiculous – how can you say ‘geographic information system system’? However, people have a reason for using this phrase and it makes some sense to them.

Maybe the reason is that GIS now stands for a class or type of computer software that can manage, manipulate and visualise geographic information, so GIS system is the specific hardware and software that is used. Personally, I’ll continue to find it odd and use GIS for what it is…

Maps are wonderful, but GIS are hard to use. What can we do about it?

These are the slides from the presentation that I gave to the BCS Geospatial SG.

The talk abstract is:

Here is a useful party trivia: as a form of human communication, maps pre-date text by thousands of years – some early spatial depictions are 25,000 years old, whereas writing emerged only 4000 years ago. When it comes to computing, the reverse is true: the first wide use of computing is from the early 1950s, whereas the first effort to create a GIS only started in 1966. There are good reasons for this, chief among them is the complexity of handling geographical information in digital computers. An adverse impact of this challenge is that for many years geospatial technologies developers focused on functionality and not on the interaction with end-users. The result of this focus is that while word processors and spreadsheets became popular in the early 1980s, only with the emergence of ‘Web Mapping 2.0′ in 2005, GIS and geospatial technologies became more popular, albeit far from universally usable.

The talk covered interaction and user aspects of geospatial technologies, pointing to issues that permeate the usability and usefulness of geographical information itself (e.g. why ESRI shapefile is a popular format despite its drawbacks?), the programming of geospatial technology (e.g. why OGC WMS did not spark the mashup revolution, while Google Maps API did?) and the interaction of end users with desktop and web-based GIS.

And the talk happened at the same day in which the excellent Third Workshop on the Usability of Geographic Information was running at the Ordnance Survey.

Usability of Geographical Information – the case of Code-Point Open

One of the surprises of the Ordnance Survey OpenData release at the beginning of April was the inclusion of the Code-Point Open dataset, which lists the location of all postcodes in England, Wales and Scotland. This was clearly a very important dataset because of the way postcode geography drives many services and activities in the UK. Before the release, the costs of using postcodes in geographical analysis were prohibitive for many small organisations.

So how usable is this free Code-Point data? The principle of ‘do not look a gift horse  in the mouth’ doesn’t apply here. The whole point of releasing the data is to make it as useful as possible to encourage innovation, so it should be made available in a way that makes it easy to reuse. I evaluated it while analysing a dataset of 11,000 volunteers’ postcodes that I received from a third sector organisation.

The download process is excellent and easy, apart from the fact that there is no clear and short description of the products in a non-technical manner next to each product. To find a description, you need to go to the product page – so you are at least 2 clicks away from the product details. It would be better to have a link from each product and include a brief description in the download page. We will see in a second why this is important…

The next step was the download itself and the opening of the zip file, which was clear and easy. There is an oddity with all Ordnance Survey data that they have a redundant sub-directory in them – so in this case the data resides under \codepo_gb\Code-Point Open\ . The fact that the files is broken up into postcode area instead of one big file of 157MB is fine, but it can be helpful to remind users that they can concatenate files using simple commands – this is especially necessary to less tech-savvy users. So an explanation for Windows users that you can open the Command window using ‘cmd.exe’ and run ‘type a.csv b.csv > common.csv’ can save some people plenty of time.

But the real unpleasant surprise was that nowhere in the downloaded package is there a description of the fields in the files! So you open the files and need to figure out what the fields are. The user manual is hides 4 clicks away from the download page and luckily I knew that the ‘user manual’ is stored under ‘technical information’ on the product page, which is not that obvious at first visit. Why not deliver the user manual with the product ?!? The Doc directory is an obvious place to store it.

The user manual reveals that there are 19 fields in the file, of which 9 (half!) are ‘not available in Code-Point Open’ – so why are they delivered? After figuring out the fields, I created a single line that can be attached to the files before importing them to a GIS:

Postcode,Positional Quality,PR Delete,TP Delete,DQ Delete,RP Delete,BP Delete,PD Delete,MP Delete,UM Delete,Easting,Northing,Country,Regional Health Authority,Health Authority,County,District,Ward,LS Delete.

Of course, all the fields with ‘Delete’ in the name mean that they should be deleted once imported.

Interestingly, once you delete these fields, the total size of Code-Point Open drops from 157MB to 91MB – which means that it can save the Ordnance Survey bandwidth and carbon emissions by making the file smaller.

Another interesting point is that the user manual includes detailed instructions on how to change the postcode to a ‘single spaced postcode’. The instructions are for Excel, Mapinfo and ArcGIS. This is the type of information that can help end-users start using the data faster. Finally, you can use this wonderful information to create lovely maps.

All these problems are minor, apart from the description of the fields which is a major usability error. Similar analysis can be carried out for any of the Ordnance Survey datasets, to ensure that they are useful to their users. There are some easy improvements, such as including the user manual with the distribution, and I’m sure that, over time, the team at the Ordnance Survey will find the time to sort these issues.

http://www.ordnancesurvey.co.uk/oswebsite/media/news/2010/April/OpenData.html

The source of the assertion that 80% of all organisational information is geographic

The question from Jeremy Morley ‘An often quoted figure estimates that 80% of information contains some geographic reference.’ – anyone got the source reference for this?! led me to search for an answer. This assertion is indeed often quoted in governmental documents, academic papers and trade magazines.

So, what is the source? While V1 magazine suggests that it links to a magazine article from 1992, a search on Google Scholar shows that William Huxhold’s 1991 book ‘An Introduction to Urban Geographic Information Systems’ is mentioned when this factoid is used. For example, here, here or here, although the last one includes an independent assessment that uses an 80% value.

Let’s look at what was said in the original book, on pages 22-23:

‘A 1986 brochure (Municipality of Burnaby) published by the Municipality of Burnaby, British Columbia, reported the results of a needs analysis for an urban geographic information system (GIS) in that municipality: eighty to ninety percent of all the information collected and used was related to geography.’

On page 236, the following statement can be found:

‘Chapter 1 reported that 80-90 percent of all the information used by local government is related to geography.’

And the latter is probably the source of the famous statement. So for about 20 years, the GIS community has been using a powerful assertion which is actually based on a brochure and not on a rigorous analysis of evidence. Maybe, as John Fagan suggested, it wasn’t a good idea to look too closely!

Workstations or PCs for GIS? The long memory of the Internet

Over the past decade, different people either hailed or criticised the growing inability to forget in the digital age in which we are living. Logging on to Google Dashboard and seeing every single search that I carried out since 2006 is rather scary – especially as there is no guarantee that if I ask to delete my web history, it will be also deleted from Google servers – just anonymising the information which is not much, really. An interesting point of view on the virtue of forgetting in today’s digital world is available in Viktor Mayer-Schonberger’s recent lecture at the RSA .
And then there is all the public information about you that is already on the open web and that is going to be there for as long as the servers and the web continue to be around. While looking for my earliest internet trails, I came across a posting to the usenet group comp.infosystems.gis from 1994. Back then I was working on a large-scale GIS project for the Israel Electric Corporation and, as far as I can recall, I was asked to write a briefing about the direction that we should take regarding the hardware and software platforms that would be used by the client in the roll-out of the system, which was designed for IBM RS/6000 workstations.  The requests that I sent to the list and the discussion are summarised in a posting that is still accessible on Google Groups – so anyone can find it and read it …

In terms of internet memory, it does expose certain aspects that I’m now much more aware about – such as my control of English back then. Glossing over the grammar and spelling mistakes, the analysis makes interesting reading from 15 years perspective.

Firstly, it is interesting to note that the need for high-end computing in terms of operating systems and hardware for GIS remains a relevant issue. See, for example Manifold GIS’s use of 64-bit operating system or the issue of graphic capabilities and the use of General Processing of Graphic Processing Units (GPGPU) in GIS and Remote Sensing packages such as Geomatica. Another indication of the continued need for processing power is illustrated in the description of ‘who might need this?’ for high-end workstations – although in 1994 no one in PC Magazine ever mentioned GIS.

However, for the absolute majority of end-users who are using GIS for basic map making and processing, this is not true anymore and many are using standard desktop or laptop computers quite well. Over the next few years, as more of the processing migrates to the ‘cloud’, the number of GIS users who need high-end machines will continue to decline. In 1994 the expectation was that most users will need a workstation, whereas very soon they will happily use a low-powered netbook.

Secondly, it is interesting to see the changes in data sizes – I note in the text that 1GB data caused us difficulties in backups and the local network (10BASE-T). I recall complaints from the rest of the company, which was running mainframe systems with millions of alpha-numeric operations, when we ran performance tests because of the bandwidth that GIS processing consumed. This aspect of geographical information handling is still challenging, usually not at the local level – even for large-scale processing, the cost of storage is so low that it’s not a problem. However, for the people who manage the backbone of large-scale applications, say Yahoo! Maps, this is still an issue – I assume that video, images and maps are now major consumers of bandwidth and disk storage that require special handling and planning.

Thirdly, there is a lesson about ‘disruptive technologies’. The PC was one such disruptive technology and, even over a decade after their introduction, PCs were not comparable to workstations in terms of memory, processing, multitasking and networking. The advantage of workstations was clear in 1994. Even as late as 1999, when we ran the Town Centres project on Sun workstations, there was still an advantage, but it was disappearing rapidly. Today, UNIX workstations occupy a very small niche.

This is an issue when we think forward to the way GIS will look in 2015 (as the AGI Foresight study is doing) or 2020. Some of the disruptions to the way GIS operated for many years are gathering pace, such as the move to software and data as services where organisations will receive the two bundled from a provider, or using more crowd sourced information.

So sometimes it is useful to come across old writing – it makes you re-evaluate the present and consider the future. At the same time, it is only because I forgot about the post that it was interesting to come across it – so Victor Mayer-Schonberger is correct that there is a virtue in forgetting.

WUN Global GIS Academy Seminar – What’s So New in Neogeography?

These are the slides from the Worldwide Universities Network Global GIS Academy Seminar from the 22nd October. The seminar’s title is ‘What’s So New in Neogeography?’ and it is aimed largely at an academic audience with background in GIScience.

The aim of the talk is to critically review Neogeography: explain its origins, discuss the positive lessons from it – mainly in improved usability of geographic technologies, as well as highlighting aspects that I see as problematic.

The presentation starts with some definitions and with the notice that mapping/location is central to Web 2.0, and  thus we shouldn’t be surprised that we’ve noticed a step change in the use of GI over the past 3 years.

By understanding what changed around 2005, it is possible to explain the development of Neogeography. These changes are not just technical but also societal.

The core of the discussion is on the new issues that are important to Neogeography’d success, but also raising some theoretical and practical aspects that must be included in a comprehensive analysis of the changes and what they mean to Geography and geographers.

The presentation is available below from slideshare, and the (very rough and without proofing) notes are available here.

OpenStreetMap: User-Generated Street Maps – IEEE Pervasive Computing paper

Earlier this year, in April, John Krumm from Microsoft Research, the editor of IEEE Pervasive Computing commissioned me to write a paper about OpenStreetMap for the magazine. The paper was written together with Patrick Weber, and it is finally out. It went through the magazine peer review process, and it is part of a set of articales in the October-December issue of the magazine that are dedicated to aspects of user-generated content.

The article was written for a general audience, and aims to provide an easy to understand introduction to OSM that is suitable for technically minded readers (such as the readers of IEEE Pervasive Computing!). It provides some history, a description of the OSM geostack and how it operates and ends with some open issues and challenges that the project is facing.

You can access the article from IEEE website, and it’s full citation is
Haklay, M. and Weber, P., 2008, ‘OpenDtreetMap: User-Generated Street Maps‘, IEEE Pervasive Computing, October–December 2008, pp. 12-18.

Snapshot Study – a Useful Usability Engineering for GIS Method?

For a while now, I have been calling for the development of usability engineering for GIS – a set of techniques and methods that will ensure that geographical technologies are learnable, efficient, easy to memorise, error tolerant and provide a satisfactory and enjoyable experience. While general usability guidelines are as relevant for GIS as they are to other computer-based systems, the use of maps as an interface and the nature of the data require the development of specific methodologies for GIS.

An example of such a method is a snapshot study. This is a simple usability study where we ask current users of GIS to take a snapshot of their computer monitor (by pressing Print Screen), fill in a very short survey and send the results by email. It requires very limited investment from the participants, but actually reveals a lot about the software packages that they are using.

The study that I conducted with Antigoni Zafiri in the summer of 2005 helped in understanding common screen resolution, the way users set toolbars on their interface and what type of maps they are looking at.

This type of study can be effective for the revision of products, to understand how they are used. It can also be used to understand the common computing environment in which the product is used.

The study has now been published in the Cartographic Journal – and can be downloaded from here. The proper citation for it is Haklay, M., and Zafiri, A., 2008, Usability Engineering for GIS: Learning From A Screenshot, the Cartographic Journal, 45(2) 87-97.