I recently posted a great visualisation showing 24 hours of shipping in the Baltic. I liked it for its cinematic appeal (was a bit less keen on the music though), and said that such work goes a long way to broaden the appeal of data visualisation. 422 are the masters of this art, producing a great number of innovative visualisations for TV programmes around the world. I first saw their work on the “Britain from Above” series shown on the BBC and have been amazed at what they have been able to produce ever since. The video above is a montage of some of their projects. Enjoy!
The video above by the MIT Senseable City lab is one of the nicest I have seen to illustrate train and passenger flows along a rail network. The network itself, the SNCF high speed trains in France, is fairly sparse in terms of the number of trains passing along it each day so the visuals can be kept relatively clean and simple. As the quote from the team below suggests, this animation is a nice way of quantifying the impact of train delays both in terms of their duration and also the number of passengers affected.
“Trains, at times, do run late. While a rail network operator is interested in reducing overall delay as such, an especially critical aspect relates to the number of passengers directly affected by such delays and their location.
In this visualization we combine data on the time trains run behind schedule with the actual number of passengers on any train at any moment. This information is represented at the actual location of a train on SNCF’s high speed rail network. With this, a rail operator can quickly understand where many passengers are affected by train delays and use this information to take appropriate action, ultimately limiting delay per passenger and increasing overall passenger satisfaction.”
This video, produced by Chris McDowall, shows the journeys that the buses (teal), ferries (blue) and trains (red) take each day in Auckland. Chris’s description on Vimeo summarises what’s going on brilliantly:
“The animation begins at 3am on a typical Monday morning. A pair of blue squiggles depict the Airport buses shuttling late night travellers between the Downtown Ferry Terminal and Auckland International. From 5am, a skeleton service of local buses begins making trips from the outer suburbs to the inner city and the first ferry departs for Waiheke Island. Over the next few hours the volume and frequency of vehicles steadily increases until we reach peak morning rush hour. By 8am the city’s major transportation corridors are clearly delineated by a stream of buses filled with commuters. After 9am the volume of vehicles drops a little and stays steady until the schools get out and the evening commute begins. The animation ends at midnight with just a few night buses moving passengers away from the central city.”
We know that knowledge networks and intensive competition within cities boosts innovation. There are also further scales to this dynamic. The networks and competition between cities at regional and global scales promotes the adoption of new ideas- as cities buy, borrow and adapt ideas from their competitors. It’s this latter global dynamic that we’re exploring in this post, investigating the spread of new ideas in a sector that’s intrinsically urban in nature- public transport. After widespread decline in the second half of the 20th century, transit has recently undergone an impressive renaissance linked to the dramatic growth of urban populations, high density forms and sustainability policies.
The spread of new ideas between cities is clustered in space and in time, as cities are strongly influenced by nearby competitors, as well as economic investment cycles. Therefore a natural way to visualise these spatial and temporal patterns is through animated cartography. This is the technique used here with the help of Processing and the MapThing library by Jon Reades (allows GIS data to be imported into Processing).
So first up we’re going to head back in time to the invention and dispersion of the underground/subway metro (data from metrobits.org; best viewed HD fullscreen)-
London celebrated 150 years of the Underground this year, and it was three decades after 1863 before other cities in Europe and North America had their own high-frequency high-capacity city centre networks. This delay can be linked to varied levels of industrialisation between countries, as well as the time taken to improve the metro concept with electrical power (the original Underground amazingly used steam locomotives). It’s interesting that the youthful American metropolises of Chicago and Boston were quicker off the mark to build metro systems than many European capitals.
Buenos Aires in 1913 and Tokyo in 1927 (now the world’s largest metro) were early exceptions to the European and North American monopoly on metro systems. Yet it took until the 1980’s onwards with the rise of Newly Industrialised Countries like Brazil, Russia, India, Mexico and Turkey for metro systems to become truly global. China is now in a league of its own with gigantic metros in Shanghai, Guangzhou, Beijing and Hong Kong.
Underground metros may seem like the best answer to cities’ transit demands, but they are highly expensive and disruptive to build, and are pricey to maintain also. These difficulties underlie another key innovation in the global rise of public transport- bus rapid transit. The use of segregated roads, specially designed stations and articulated buses enables BRT to have similar capacity and speed advantages of subways at a much lower cost. We can see from the animation that BRT begins as a Brazilian innovation (data from brtdata.org)-
Initially BRT adoption is highly clustered in Brazil’s major cities, with a few early adopters including Santiago de Chile, Quito, Pittsburgh and Essen in Germany. Then in the late 1990’s the dynamic changes with a burst of new systems in Central America, Canada, Australia, and mainly second-tier cities in Europe. Taipei has spearheaded the adoption of BRT into China, with many new large systems emerging. Sizeable BRTs also recently opened in Istanbul, Tehran and interestingly in Lagos where hopefully further investment in African cities will follow.
In our highly connected globalised world, new city innovations are likely to spread more quickly, and that seems to be the case with BRT. Indeed this acceleration effect is even more marked in the last innovation we’re going to investigate- the bike sharing phenomenon. Now bike share schemes are of course small investments compared to city-wide metro systems, yet they are still an interesting recent advance with similar global dispersion dynamics (data from Bike Sharing World Map and O’Brien Bike Share Map)-
The original pioneer of bike sharing is not as clean cut as the BRT and Underground examples as there have been several generations of innovation (see pdf article). In 1995 Copenhagen successfully created a reasonably sized (1,000 bikes) coin operated system with specially designed bicycles that tried to reduce theft. A small number of cities in Germany and France followed suit. The next generation began in Lyon in 2005 with a larger (4,000 bikes) system using smart card technology that greatly reduced theft. Subsequently bike sharing has exploded globally across Europe, North America, China and South Korea.
Paris has by far the largest system in Europe with 20,000 bikes. But even Paris’s Vélib’ is small compared to two huge Chinese systems in Wuhan (90,000 bikes) and Hangzhou (70,000 bikes). China’s strong cycling tradition has recently been in decline with rising car ownership, and hopefully the Bike Share boom will reverse this trend.
So to conclude, we are experiencing an age of truly global transit adoption with innovations spreading more rapidly through global city networks. While innovation has traditionally arisen in Western European and North American contexts, by far the greatest urban growth is in Newly Industrialised Countries, increasing demand for innovations like BRT. The rapid rise of bike share systems shows that relatively modest innovations can have a global impact when the innovation is popular and effectively implemented.
The ridership and scheme size data relates to current passenger levels rather than the size of the system at the time of construction. Would be great to do this visualisation with time-series ridership data, but this is not to my knowledge currently available.
The definition of metro and BRT systems used here comes from the database providers, and there is some ambiguity, e.g. in defining when a regional urban rail system can be classed as a metro (see metrobits.org).
Access to credit/ debit card data is, quite rightly, heavily restricted, so there aren’t many data visualisations out there that can match the detail of the one above, produced at the MIT Senseable City lab. It shows how Spanish spending patterns vary both geographically and temporally in the run up to Easter.
It is inspiring technically and of course visually. Still, even though there is a whole story told in the animation the sound is not pushed in to the background, the two elements perfectly join up to one great piece of entertainment.
The Randevous album can be bought on iTunes HERE. They also feature the music video there and some additional stuff.
We have now finally also an animated NCL (aNCL) version using the same dataset. This part of the project was only developed earlier this year in collaboration with Anders Johansson at CASA and we are trying to catch up on the different cities we have data for. A series of aNCL visualisations has already been realised.
Image by urbanTick for NCL / Showing four screenshots taken from the aNCL visualisation for a weeks worth of Tweets in and around London. The timings are midnight, morning, afternoon and evening. Each do is a tweet, re-tweets show a lin between sender and re-sender.
There are only very few features we are using for these visualisations. A characteristic landscape feature to roughly describe the urban area and the 30 km collection radius parameter to provide scale. Other than that there are only the individual Twitter messages that were collected over the period of one week. THe animation superimposes all seven days in to 24 hours.
With the visualisation we are highlighting the way information disseminates through re-tweeting of messages. An RT message will show a thin yellow line between original sender and re-sender. The information travels at some speed, which is based on the time it takes between sending and resending.
London, even though the data is already a year old is compared to other cities a very busy place in Twitter terms. We have a lot of individual messages, but more interesting there are quite a lot of different interactions happening simultaneously. Where as other cities don't show a lot of interaction, in London the sharing of information is quite an important part of tweeting. An interactive, but static activity map can be found at London NCL.
Its great to see how London wakes up between 07h30 and 09h00 in the morning after a moderat night. Then there is however, not very much sharing at this point of the day. Only after lunch and especially later in the afternoon the sharing really starts in London. It is almost as if the city was to digest the information it had created earlier in the day, reprocessing it and passing it on.
The highest point is just over the Placa de Catalonia with a steep slope down la Rambla to the Roca Columbus. Other places of high activity are around the parliament, here the 'Monte di Parliament Catalonia' and around the Olympic centre on Montjuic.
Image by urbanTick for NCL / Barcelona New City Landscape map generated from location based tweets collected over the period of one week. The area covered is within a 30 km radius of Barcelona.
The Barcelona New City Landscape map has already been published earlier, but it needed an update because of some problems in the processing and labeling. This new version also goes in line with the adjusted layout and design.
Thanks for the help with the map go to Narcis Sastre, who kindly worked it through.
Barcelona New City Landscape
Image by urbanTick using the GMap Image Cutter / Barcelona New City Landscape Use the Google Maps style zoom function in the top right corner to zoom into the map and explore it in detail. Explore areas you know close up and find new locations you have never heard of. Click HERE for a full screen view. The maps were created using our CASA Tweet-O-Meter, in association with DigitalUrban and coded by Steven Gray, this New City Landscape represents location based twitter activity.
Barcelona is very active in the afternoon hours. There is a peak around 15h00, 18h00 and 21h00, after which it quickly drops off. The mornings are very pronounced right after six, however overall far less than the afternoon. Over lunch there is clearly a dip with lesser activity.
Spanish is clearly the dominating language, followed by English. Indonesian, French, Portuguese and Italian are sort of the runner ups. ALso Esperanto is there, this is surprisingly often present in the top ten list and it seems that a lot of people are using it as a statement, since it is not really a spoken language.
Image by urbanTick for NCL / The rose shows the twitter activity per hour of the day, starting at 00:00 at the top, displayed as local time. Barcelona is a afternoon city with more activity between three and nine than through out the rest of the day. The graphs show the platform of preference used to send the tweet and the language set respectively.
Also, we have the animation ready for the Barcelona data set. This one is put together in collaboration with Anders Johanson. The animation also shows the interaction between the users based on RT and @ tweets with thin yellow lines. This indicates a direction and provides a sense for the distribution of flows.
The data used is the same as for the San Francisco New City Landscape (NCL) map. Where a virtual landscape was generated from the tweets. More details HERE. This new animated version shows in detail how the different centres ebb and flow as time passes. There are distinct characteristics between the location over times of the day. Basically the Bay bridge keeps it all together.
Interestingly the RT's are a very specific day thing. During the night this information channel is not ver active and people seem to be busy tweeting their own stuff. In a sense this could be hinting at a more formal and business use of the RT function.
This animation is developed in collaboratively Anders Johansson and urbanTick. The data was collected using our CASA Tweet-O-Meter tool, coded by Steven Gray, in association with DigitalUrban.
There is more to come. We will be working our way through the NCL data collection of over 70 cities from around the world. Within the next week will be posting the next city to continue this aNCL (animated New City Landscape) series.
Looking at the activity on twitter during the tsunami we are on the search for clues about the relationship between twitter and an unfolding natural disaster. As an inspiration to serves the XKCD PhD comic 'Seismic Waves'.
In this close look at twitter activity related to the tsunami resulting from the earthquake, Anders Johanson has animated the messages for one hour before and one hour after the expected arrival time of the tsunami wave in Honolulu on Hawaii. The messages are collected through the usual NCL collection method and are focusing on actual geo located tweets that contain lat/long information. Johanson explains "At the time instant when each new tweet is posted, a bright red blob appears on the map, and this blob is then decaying in intensity and size. Re-tweets are shown as an arrow, pointing from the original source of information. Interestingly enough, the information wave has the same direction as the seismic wave. However, there are obviously way too few data points to enable a rigorous spatio-temporal analysis in this case."
IMage by urbanTick for NCL / The graph shows the number of geo located tweets sent per hour from Honolulu, on 2011-03-11, in a radius of 30 km on the day the tsunami resulting from the earlier earthquake in Japan was expected to reach the Hawaiian coast. In white are the overal tweets and in purple the tweets containing the key words wave, tsunami and earthquake or Japan. The first dotted line from the left is the expected arrival time of the wave on the coast of Hawaii at around 13h07 UTC. The second dotted line is the arrival time of the wave on the coast of Mexico.
The tsunami arrived in Hawaii and hit hard, causing damages estimated to be in the tens of millions of dollars. On the twitter scape on the other hand, there is a slight increase of activity after the wave, but actually very little. However, as you can note in the graph above there are more tweets using the keywords related to the natural disaster unfolding than there are thereafter, especially after the wave arrived in Mexico.
Processing the twitter data some further, the spikes on the keywords do fade out nicely in the hours afterwards. This time we are working with the larger data set containing all the located tweets including the geo located tweets. This data set differs from the one used earlier as that it included reverse geocoded locations, eg places, but not necessarily pure lat/long messages. This set contains some 260'000 messages as compared to some 20'000 geo located in the earlier dataset.
Looking at the event over longer period shows the pattern much clearer. There is a lot more activity around the expected tsunami wave and the dying out of the keywords can be observed in the following hours and days. However it also confirms that to some extend the purely geo located tweets, as a sub set, folow largely the same pattern and are not
Note, there is a baseline tweet containing the term 'wave' that we picked up as part of the collection. This is a weather boy just of Honolulu tweeting the current status of the water, wave and wind.
Image by urbanTick for NCL / The graph shows the number of geo located tweets sent per hour from Honolulu after 2011-03-11, in a radius of 30 km on the day the tsunami resulting from the earlier earthquake in Japan was expected to reach the Hawaiian coast. In white are the overal tweets and in purple the tweets containing the key words wave, tsunami, earthquake or Japan. The first dotted line from the left is the expected arrival time of the wave on the coast of Hawaii at around 13h07 UTC. The second dotted line is the arrival time of the wave on the coast of Mexico.
A fairly romantic picture, but definitely a basic diagram explaining connection. Also it servers very well as a starting point for questions. For example for a six year old, why is the water salty in the ocean, but not the rain? How does the water know the way to the ocean? And why are the couds not drifting of to the moon?
This could go on and on developing into a full blown session on earthly systems, but it is all about the basic idea of a systemic concept. It is not about what happens it is more about what happens with it.
So or an ad explaining what a water supplier company does, this is a gold one, back to school.