OGC Temporal Domain Working Group ponders difficult requirements
Suppose you are interested in displaying, or processing, international historical data from some period in the last few centuries using OGC Map Services or other standards. Scanning the available data, you or volunteers or legions of underpaid students have extracted some interesting series of data. And you realise that some of the data is labelled 10, 11, 12 or even 13 days out. This is because various countries switched their calendars from the Julian calendar, established by the Roman emperor, Julius Caesar, to the Gregorian calendar, established by Pope Gregory XIII, in different years, between 1582 and 1923.
And of course, the geographical boundaries of those countries may have been significantly different then! For example, Bohemia, Lorraine and the Ottoman Empire switched in 1584, 1760 and 1917 respectively. Where are they now?
Suppose you are processing some detailed scientific or engineering data logged around June 2012. Did you know that at 30 June 2012, 23:59, the next minute had 61 seconds, not 60?
Suppose you are trying to geo-locate some data collected from a small Unmanned Aerial Vehicle (UAV) using a cheap and cheerful GPS receiver. You then want to correlate it with some other data from a different source. Do all the software and GPS receivers involved give correct timings to the nearest second?
Of course, everyone understands all of this, and after the lessons learnt from the Millennium Bug in 1999/2000, these problems never occur anymore, do they? And your software will not be running in 2038, but if it were, has that Unix bug in it already been fixed?
Suppose you have a deluge of space-borne multispectral images from a number of different satellites' sensors that you want to co-locate in a unique space-time "aquarium", and some of the satellites are in heliocentric orbits. As well as struggling to reconcile geocentric and heliocentric coordinate systems, what time coordinates are you going to use? Do not forget to consider that light, and any theoretical clock, is going to take more than 8 minutes to get from the Sun to Earth.
Suppose you have to deal with some 4D archaeological data models and you need to keep track of each of the six chronometric and stratigraphic temporal categories to develop a powerful time-aware GIS. The next artifact dug out of the ground could alter the dating of everything! Mapping centimetres or millimetres of ice or sediment depth cores to 'TIME' is usually straightforward if not absolutely accurate, but in archaeology, oldest is not always the deepest, because some tidy-minded person may have dug a rubbish pit for their waste!
Because of all these and other more technical issues, such as seamlessly combining space and time in coordinate systems, OGC established a Temporal Domain Working Group at the Technical Committee Meeting in Mumbai, December 2013. Its first full official meeting will be at the next TC in Washington, March 2013, where and when we hope to start establishing Best Practice for time in a geo-spatial context and formally register some temporal Coordinate Reference Systems. We have an open discussion mailing list at temporal [at] lists [dot] opengeospatial [dot] org and a wiki at http://external.opengeospatial.org/twiki_public/TemporalDWG/WebHome, and as ever, volunteer effort is needed. Make time and join in, there is plenty of space for you!
Peter Baumann at Jacobs University Bremen GmbH is the second author of this OGC Blog post, Piero Campalani, Jacobs University Bremen GmbH also provided input. Chris Little of the Met Office UK and Peter Baumann are the co-chairs of the OGC Temporal Domain Working Group.