Tonight I attended an OpenStreetMap pub night in Toronto.
In addition to some of the regular local OpenStreetMap mappers (Heather Leson) was there. Heather is involved with Crisis Commons. Crisis commons is a groups that tries to use crowd sourcing to help in emergency situations. For example, following the earthquake in Hati they (along with OpenStreetMap) worked on using aerial imagery to make post quake maps of the area to assist aid agencies on the ground.
I’m starting to see some similar problems show up more frequently. Consider the following motivational stories
* A forest fire breaks out in some area (maybe near a forest) and the local responders (maybe government officials, maybe community volunteers, maybe strangers hundreds of miles away) chart the extent of the fire. They might use aerial pictures from single engine airplanes, they might use geocoded tweats of people saying ‘the flames have reached main street’ or they might be from slightly insane people chasing the flames in their truck. These reports track both the current location of the fire and the spread of the fire over time. This information is then used to direct firefighting resources and to give people a heads up to evacuate.
* An oil platform experiences a malfunction resulting in a spill (say on the Gulf of Mexico). As the oil slick spreads citizens mark the location of oil strewn beaches and sick birds using GPS units and the Internet. Aircraft fly over the spill each day and take geocoded pictures allowing its movement to be tracked. The maps of how where the oil has spread are then used in determning how to send wildlife rescue teams and where to deploy physical barriers.
* A meeting of world leaders is taking place in a major city. Police setup barriers, protesters take to the streets. As the event makes its way through the city core residents, protests and the police all report on the location of the protest and violent activities as they happen. The information is all collated on a constantly changing map letting people see where the troubles are so they can avoid them.
Each of these stories involves geodata that is highly temporal. A mountain will have sat silently for many millions of years waiting for someone to put it on a map. Once that mountain was mapped its location on the map will stay the same for the entire life of the map. In each of my stories above the event being mapped came about rather quickly and maps of it are needed almost immediately. The data might only stay current for a short period of time (tens of minutes for the protest, hopefully not more than months for the post quake damage).
Databases like OpenStreetMap or a traditional shapefile based GIS system are geared towards mapping mountains, they have no concept of how the data changes over time. Sure researches in the sciences of mapped how some measured attribute at a particular location varies over time but mapping physical features that move or change over time is rare. There are techniques for indicating time based validity on the attributes of the data (‘valid from June 1 to June 5 2010′) but these tend to be adhoc and awkward.
I see a lot of demand for a GIS infrastructure geared towards capturing, storing and serving up geographic data in a temporal fashion. Not only do you need to show the map as it exists at this moment but you also want the history of the data (how did the forest fire move over time) to be available later for analysis or viewing.
When these events happen you don’t have time to go out and build an infrastructure for the data. Something needs to be sitting there ready and waiting. This infrastructure needs to allow for easy and fast adding of data (even from a mobile device) and it needs to allow for people to easily create temporal mashups of the data along side more stable map features (like roads, rivers and parks). Preserving how this data changes over time is key, and both the time intervals and the where (and how) the data is collected will vary from point to point.