Notes on the GSA Annual Meeting 2010 in Denver, where I presented a NCL paper during the Community Remote Sensing session. The session was organised by the Susanne Metlay for the Secure World Foundation and drew together a range of people working on crowd sourcing, online data mining and collaborative mapping. The range of work presented was very broad, but the overarching idea was to bring together these experts and in a concluding plenary discussion the way forward for an emerging field. As an overarching title the term Community Remote Sensing CRS was suggested, hence the title of the session.
Speaking of the variety of presented projects, it ranged from crowd sourcing projects, mainly smart phone based, for mapping and observing the nature environment. For example observations contributing to the national species registry, the mapping the spreading and growth of invasive weeds or the tracking of sessional changes from crowd sourced reports on the state of flowers and threes. A major group of project was looking at disaster management, with data and information gathering and rapid response. The focus here was on using the crowd as well as the wider science and academic community to quickly mobilise and coordinate recourses in the wake of a disaster event. A third group of projects focused on distributed systems to generate a wealth of detailed local data by piggybacking on existing resources, such as collecting weather and road information data through a network of vehicle based sensor information. A fourth area was the mining of existing or online generated information of web services such as social networks.
The Session was opened with a presentation by William Gail from Microsoft Startup Business Group who talked about his vision of a CRS discipline, high lighting the potential as well a the challenges and problems that need to be faced. Among the main aspects are the question of privacy and ethics, the validation and accuracy of the data as well as the question of ownership. These questions or topics continuously popped up during the whole session in various combinations and contexts. Gail gave an overview of existing projects and efforts without getting too attached to a certain direction or solution. While trying to setting up CRS as a emerging fields this was essential.
The Virtual Disaster Visualiser VDV software was on the examples focusing on rapid disaster response and offers a platform for viewing, mapping, sharing and handling large quantities of aerial photo based surface information. It for example assists with remote sensing disaster damage assessment and supports the collaboration of numerous contributors and experts. The VDV was developed at the EPIcentre at UCL by Enrica1 Verrucci, Beverley Asams and Tiziana Rossetto. The software was successfully deployed in the wake of the Haiti earth quake in 2010. The platform was used to coordinate surveys and allowed a team of 150 experts to asses the area of 1000km square within only 96 hours. For this the base data from the satellite processed by SERTIT which documented especially the port au prince area just four hours after the earth quake with a resolution of about 25cm square.
Two projects from UCLA, presented and co developed by Eric Graham demonstrated the potential of crowd sourced botanical data collections. The 'What's Invasive!' weed project made use of an Android platform application to map and monitor the distribution and growth of invasive weed species in the United States. The idea is to let volunteers submit location based and image information on the weed they happen to find. To do so the software would offer a location based list of weeds that can be reported. With one click the information is submitted to a central server and mapped in real time. The Android app can be downloaded HERE. The second project called 'Project BudBurst' is focusing on the tracking of the seasons by collecting plant state information. With this application, also Android based, volunteers can log the progress different plant species make over the course of the year. The plant state is submitted via the app to a central server. The idea is to get people to observe howler example threes grow buds and start flowering on their daily commute and that these observations can be simple and conveniently reported to simplify the contribution to a nation wide project. The projects have to show some very promising initial results where the mapping is very successful and accurate as comparison studies show. Furthermore the crowd sourcing aspects allow for a widely distributed collection setup with a lot more reports than individual experts would be able to archive. However the project team has identified the challenges to be with the motivation of volunteers for a continuos contribution rather than one off submissions.
Image taken from Project Bud Burt / A loacation mapped with description popping up on te interactive map.
A massive environmental disaster hit the American Gulf Coast around New Orleans in May this year and the extensive oil spill will dominate the area for years in many ways. From food production, for example sea food businesses, to every day natural impact and animal deaths and large scale loss of habitats. A local initiative to enable the local population to deal with the intense impact is the NGO Luisiana Bucket Brigade. They where in this from day one and they are locals, which means they know the area and especially they know the people. They got involved with the Grassroot mapping project and implemented it as part of their activities in the Gulf area. The idea is to produced community sourced mappings of the oil spill. The technique was developed at the MIT by With cheap parts and simple technique the project manages enable nearly everyone to start doing quality mapping. Their Wiki describes all the details of what is required and how to put it in practice.
The images captures b the flying camera are later on stitched together to a high resolution image that can be used to produce the maps. The flight device not necessarily needs to be a helium balloon, but also can be a kit, large enough to carry the additional weight of the cam.
Image by Cesar Harada & Hunter Daniel / Bay Jimmy, Detail 04, 20100722, for : LA Bucket Brigade | labucketbrigade.org/, with : grassrootsmapping.org/, location : 29.468400 ; -89.911300, time : 2010 July 22 @ 10:30AM
This is of course only a very short and incomplete selection of the presented work. The single-blogs contribution can be found HERE. The potential of community based data collection has been realised and can be taken a step further coming from the web 2.0 idea. Once more this session demonstrated the diversity of the possibilities for applications and utility. On the other hand it has also clearly visualised the questions that need to be addressed, starting with ethical and privacy issues all the way to ownership and authority.
I will be at the CRESC 'The Social Life od Methods' conference in Oxford today. The conference takes place at St. Huges College and covers different aspects of methods in social sciences.
I am presenting a joint 'CASA' paper focusing on data mining and crowd sourcing. The team behind it is Andy Hudson-Smith from Digital Urban, Richard Milton, Steven Gray and myself. We are focusing on the development of tools that make use of the recent wave of digital social networking tools and methods.
For the first time it is possible to gather large scale social data sets containing detailed information about activity, interests and network of thousands of individuals. This is extremely exciting and interesting for spatial analysis. The city is not just a built artefact, but a living structure continuously changing with the inhabitants shaping it. Using this data we can start to understand the immediate connection between the environment and individual a well as collective decision making and opinion.
Image by single-blogs / A visualisation of my over 1200 facebook connections and how they are connected, using the Gephi software.
To illustrate this I will be using three recent projects that we have developed here at CASA. It will feature the 'Mood Mapper' turned into MapTube. The brand new Survey Mapper system that allows everyone to set up a survey and get the result on a map and the New City Landscapes generated from tweets sent in an urban area.
From the abstract:
The paper describes how we are harnessing the power of Web 2.0 and related technologies to create new methods to collect, map and visualise geocoded data as an aid to further our understanding of social, spatial and temporal change in cities. The authors begin with an insight into the ‘Ask’ survey system developed as part of the National infrastructure for e-Social Simulation (NeISS) project. ‘Ask’ provides social scientists with series of online tools to collect and visualise data in near real-time allowing the creation of ‘mood maps’ linked to a backend geographic information system. We examine the systems use to date, specifically by the BBC, and the implications of allowing anyone to survey the world, continent, nation, city or indeed street via our social survey system.
The authors expand on the concept with the additional of data mining social networks such as Twitter to collect, map and analyse social related data. Developed around the populist name ‘Tweet-o-Meter’ we have developed a system to mine data within a 30 km range of urban areas, focusing on New York, London, Paris, Munich, Tokyo, Moscow, Sydney, Toronto, San Francisco, Barcelona and Oslo. The system mines all geo-located Tweets creating a vast database of social science data and numerous challenges for both visualization and analysis. The paper concludes by arguing that data mining has notable potential to aid our understanding of the complex social, spatial and temporal structures of the city environment.
Image by single-blogs / Moscow New City Landscape, explore areas you know close up and find new locations you have never heard of. Click HERE for a full screen view.
A writeup (not mashup) or write along of the one day CASA conference at UCL - Advances in Spatial Analysis and E-Social Science. Comes in three sessions with presentations on current CASA research work and a punchline panel session at the very end.
Session One
Online exploration of cultural regions, migration and ethnicity using the geography of personal names by Paul Longley & Pablo Mateos Paul Longley is talking about the Surname-Profiler at www.spatial-litteracy.org The first slide is showing a very early map outlining the origin of welsh names in 1881 and the change through to 1998. He then goes on to the profiler websites where they attempt to map the world surnames with origin and location. This database contains some 8 million surnames. Using some screenshots, Longley talks through the options and features of the page. The projects potential really is the visualisation of the spatial dimension of names currently but also over time. However this is not well demonstrated here. The project h sparked of a number of detailed related investigations, for example research into realties between surnames and genetics at the University of Oxford or the definition of regions based on Surnames in James Cheshire's PhD here at UCL Geography. Pablo Mateos takes over to demonstrate some applications for this work. The first bit is looking at neighbourhoods, with an example of Kreuzberg in Berlin. Observations based on door bell signs, People tend to cluster by building. Similar ethnic groups seem to prefer to vie close. Over al Mateos' idea is to create a surname based ethnicity classification. This is the basic idea of onomap.org For the clustering the group has also started to use network clustering algorithms producing some intriguing graphs. Applications for this are in public health, cancer research, political party representation and residential segregation. In Germany and the UK they are working on projects to analyse the residential segregation. The main problem they are facing is the lack of public available ethnicity data across Europe.
Spatial Interaction Models for Higher Education by Alex Singleton & Ollie O’Brien They are working on partial interaction models and it seems they are summarising a lot of the ongoing work at CASA and related researchers. Mainly the focus is on commuting and how citizens get to a destination, might this be work or shopping. The project in a second stage is focusing on education and the impact of the move from schools to university. For this they are using a partially constraint model because there are more pupils than places at university. One of the motivation is the question of how the geodemographics affect the pattern of university choice. Regarding the data they are using the NPD national pupils data base for schools and the HESA individual students records for universities. Because of the large number of individual indicators and factors, they have simplified the the source aggregating some factors. For example they ignore school types or assume that pupils go to school within the ova authority they live in or make the universities a one subject destination. This is really a dramatic simplification of the problem, for me Some additional factors would have made the model more exciting. What does it represent if there is no longer a reason to choose a specific university, based on subject, quality or financial possibilities. The initial goal to research the impact of the geodemographic background of the pupil on the choice of university moves further away. The model is written in Java. The presented results of the model compared to the actual data are surprisingly similar, well actually there are no differences to be made out from the back of the ranks. I am guessing that we are dealing here with a lot of social factors, such as demographics background, financier circumstances, knowledge and performance levels that are not taken into account. The fact that some students simply go to the university they are offered a place is here not represented and will increasingly not be if there will be a stronger focus on cost diastase relationships taken from the Wilson entropy models.
The Dynamics of Skyscrapers Scaling, Allometry, and Sustainability by Michael Batty The main focus of the talk will be on the distribution of skyscrapers in the city. Batty describes that the distribution will not at al be normal, since it is based on growth and competition. The data is based on the Emporis database on high building. Interestingly Batty points out that the development of skyscrapers in size is somehow interlocked without economical cycles. A first finding he points out is that fact that the floor highest in Paris are only 3.25 meters compared to Dubai with a floor hight of 4.32 meters. This is really a dramatic difference which I can not believe to be true. Skyscrapers are purely based on engineering and financing parameters and a difference of one meter, especially over the 160 floors of the tallest building, are a trey dramatic difference. Batty points out that there are some basic differences between the distribution of skyscrapers compared to cities. Tall buildings mainly become more, they get not destroyed often, and they do not grow as such. One of the major problems Batty is dealing with here is the fact that the Zip law really is only looking at the top end of the graph. This was because the calculation was without computers much too tedious to go much beyond the top 200. Looking at 200000 examples show a different graph, where the tail dramatically drops off. The interesting part of the talk really is the bit about the rank clocks where Batty visualises the change in size over time.
Image taken from digitalurban / Rank Clocks
Development of an urban growth model using high-resolution historical data by Kiril Stanilov Stanilov is looking at growth data of a section of western London he had digitised in painful detail from historic maps. He introduces the topic with a note on the increasing detail research in other fields of science but not in urban modelling. It is still largely based on the Triers concentric model of the city. The data collection is abased on OS historical maps since 1875 with equip intervals of 20 years, resulting in seven time layers. From the maps Stanilov interpreted also the land use type. From the growth he identifies three types, nuclei, scattering and infill to characterise the different stages of growth. In detailed patterns he points out that there must be a relationship between infrastructure and new development. He demonstrated this with maps and graphs highlighting roads and rail way stations. However the question of the actual relationship between the two elements are not specifically investigated. The question of what ceps first, the road or the development, the rail way station of the housing area remains in the air. The model, as demonstrated, is extremely close to the reality. With each increment in time the model follows the actual patterns even though the growth was only constraint by the overall amount of land use to grow over each time unit. Stanilov points out what hasn't changed is that relationship between land use and accessibility. One of the difficult elements of this model really is that the historic data shows a period of filling land as the city grew outwards. The current conditions are completely different, since the city now is existing. Dealing with existing urban fabric is definitely different.
Session Two
The research frontier in urban modelling: the agenda and the challenges by Alan Wilson He is demonstrating retail model for London and how the interaction with the software is structures. He then makes a comparison between the existing conditions in the city to the cities DNA. This is a very tricky jump in thinning I am not quite able to follow. Wilson the goes on to refer to classical scenario elements such s the possibility corridor. He the develops the DNA idea into the 'genetic medicine' indicating an universal cure for any retail related decline. To follow this thought you not only have to accept the universal concept of genetic cures, the concept of the DNA needs to be adapted. The existing condition is a state and not a basic unit containing the information on how the individual units are structured and built. However Wilson goes on to show two further studies, one is a detailed retail predicting model again for South Yorkshire and a transport model for the Western part of the United States centred around Chicago, focusing on the development of the railway lines between 1760 to 1860
On-line Spatio-Temporal Analysis of Network Data and Road Developments by Tao Cheng The starting point is the observation that todays cities are increasingly congested and the demanded mobility explodes. The detail case study is the Greater London area also in light of the 2012 olympics. The major of London has reduced the congestion charge zone, this project is employed to develop methods to reduce congestion despite the smaller charging zone. The approach chosen is a network modelling route with the obvious challenges of data management. The amount of data rise to 20 GB monthly. The main point of the methodology it the possibility to integrate space and time. For the analysis Cheng is using time-space aquariums to visualise the amount of traffic over time on specific roads.
Twitter Tags and Real-Time Visualisation of Complex Geographic Data with MapTube by Richard Milton Introducing the tweet-o-meter developed as a part of the genesis project by Steven Gray. Milton shows the mapped location tweets in London that we produced in a collaboration with digitalurban, single-blogs and Steven Gray. The video plays nice, but the sound is not suitable for a presentation. It is impossible to hear the explanations. Anyway, for details head over to single-blogs/twitter and watch the full series of posts on the topic. Milton then moves on to show the development around the UK snow map developed by Steven Gray using the # tag as a collecting parameter. People were tagging their snow related tweets with a #uksnow code followed by the postcode and the amount of snow fall between one and ten. Miltons aim here is to compare the crowd sourced snow weather data with the accurate Met Office weather data. The conclusion on this experiment is not quite clear, but partly it is that the amount of snow indicated by the crowd is not reliable and can not be used as a scientific data set. The third and final work Milton demonstrates is the new version of mapTube, now with the integrated ability to directly import CSV files from a web source. Also since there now are some 800 or 900 maps on the mapTube service it is in need of a organisation method to find the maps desired. The last bit, since there is one more, are the mod maps developed at CASA for the BBC. The one pressuring point on this visualisation and mapping technique really are the temporal aspects of the data, especially since today Milton was talking about Twitter, GPS and mobile device generated data. This might be a future step for the mapTube development.
Session Three
Tales of Things and electronic Memory – Creating and Mapping The Geography of Everything by Andrew Hudson-Smith
Image taken from taesofthings.com / Project logo.
For the final bit of the day the long awaited and pretty much kept secret project presented by Andy Hudson-Smith finaly comes to the stage, 'the World of Things' - "What if we knew the geographical location of everything?', 'What if everything in the world was tagged?', 'What if everything could tweet?' All this is part of the TOTeM project that came out of a Sandpit pitch as a collaboration between the Brunel University, Edinburgh College of Art, University College London, University of Dundee, University of Salford. A more than a million pund project concerned with the internet of things. The project is built around the idea of linking multimedia data to real object via a machine readable tag in the form of a QR code. The vision is that personal memories in tis way can be linked to physical object and narratives can be passed on together with the object. Hudson-Smith demonstrates the different aspects of the website and how to tag object via QR Code or RFID tag, using a self-made RFID to Bluetooth reader.
He then also introduces a few possible partners, the British Library London Olympics 2012 with whom a collaboration on tagging and linking could be beneficial. For the Future Art Festival they are working with artists and an Oxfam shop for a week tagging and recording for all the incoming and outgoing object. Anyway we'll probably hear a lot more about this project in the coming days and weeks. There was an article on the New Scientist's front page on Tuesday 2010-04-13 and the website talesofthings.com will launch on Friday 2010-04-16.
Panel Discussion with Mike Goodchild, Keith Clarke, David Maguire and Carl Steinitz The last bit of the conference is a panel discussion on the aspects of GIS. Carl Steinitz gives a short introduction. He gently hits out to the earlier shown projects, from railroad modelling to mapping existing large scale data sets. It is quite clear from the beginning that he aims at a very different kind of scientific research. He points out the importance of scale and context of the research. It matters what the question is and it matters what the He then again hits out on the work produced at CASA, saying there if far too much visualisation produced without understanding the meaning or the message. He makes it very clear that the output often is crap and can not communicate its content. He criticises everything from music, to colour codes. The approach he presents is a division of different elements that can be combined at will to get different outputs. Maybe the world has moved on since the elements were freely dividable and analysed separately. But the main point he is making here, is that not one single model can explain everything, but different models can explain different aspects. He makes an important separation between design decisions and growth structures and calls his approach geoDesign. The other members of the panel react and respond to his critical position with different versions. Some clearly agree, others point out that there is something to this new approach of crowed sourcing and data collection outside the established reliable official bodies. This session was clearly the best bit of the conference, naming and pointing out the 'hot' points, beside the usual talking and presenting. In this sense it was a good round of and shake up of this one day conference.
Other comments and reviews can be found on twitter using the #casa tag or on knowwhere.
I will be at the Stadtkolloquium workshop for the next two days. The Stadtkolloquium is an international PhD workshop for and with PhD students here at UCL. A divers range of research topics related to the city and urban aspects of space will be up for discussion. I will be presenting a paper with he title ‘UrbanDiary - The Temporal Narrative of Space or the Construction of the Collective and its Visualisation’. The main focus will be on the last part of the manufactured title construction, the construction of a collective and visualisation aspects of temporal data.
A brief summary will form the introduction to enable everyone to understand the extent of the investigation as well as to grasp what field work has so far been undertaken. This set the background for the explanations and ultimately the discussion to follow. The focus of the explanatory part is held on the experience and perception of space. The aim is to build up, on one hand the methods and ways of investigation and mapping , as well as on the other hand introduce the concept of the narrative. Along the story of individual experience, the narrative of the space and ultimately the city is redrawn. It is not so much the mark on the city, but the mark the city leaves on the individual that interests. It is hoped that this also clarifies the position. However, the aspect of the body and is left out of this discussion. This is mainly to not confuse the direction of the discussion. The Discussion part is dived in to two section, the possibility of a collective construction and the visualisation of the findings. Earlier is illustrated with track records shown on a city level including and connecting to the immediate urban morphology, followed by the UD London map. The new twitter weekend map is then used to illustrate other sources of data that could be interpreted as collective. However this is mainly thought as a starting point for the discussion around the construction of a collective and its value compared to the individual data. With a activity graph based on time and amount of activity, the transition to the topic of the visualisation is introduced. Also the time-space aquarium will be up for debate and an animated UD London map.
POPFest needs you, POPFest needs your paper! This years POPFest has now put out a call for papers. Already the new website is worth a visit and you will get the paper details details HERE as well as submission guidelines HERE. POPFest, in case you haven’t heard of it before, is the annual Population Studies Conference for postgraduate students. It says on the website: “It is open to postgraduates from all disciplines studying any aspect of population and demography and provides a supportive and stimulating environment for students to meet, present their work and exchange ideas.” The conference is organised by postgraduates for postgraduates and this year it will be hosted by the University of St. Andrews together with the University of Dundee. It will be over three days from 28th to the 30th of June 2010. I did attend this years POPFest at LSE in London and presented the paper ‘The Spatial Extension of Everydaylife’. It is a great experience for me and a very supportive context. Oral presentations will be 15 minutes long with 5 minutes for questions. There is usually also a session with posters. However important here; #POPFest now also is on twitter! The only way to get them to actually twit is by subscribing numerous, click @HERE.
Image by POPFest - sorry no better graphic available.... I’ll ask Tom - Toooooooommmm where is the nice graphic?
