OpenStreetMap
Anyone using OSM data at utility companies?
Posted by Alfred Sawatzky on 26 January 2011 in English.My day job is to work as a Smallworld consultant to large utility companies (gas, electric, telco, water). I'm trying to figure out how to sell these large companies on using OSM as their landbase. By nature, these utilities are very conservative and slow to change. I did a presentation at the Smallworld Users Conference this last year (PowerPoint presentation here: http://bit.ly/f3X1Cg), but I also need to figure out a better way to get utility users hooked.
My ultimate goal is to get an organization like British Telecom or Deutsche Telekom to use OSM actively in their GISes. But I might need to start off with a small electric cooperative first. I think that having a large company use OSM would only help drive the OSM project forward.
Any ideas? Anyone had experience using OSM at a utility or convincing a utility to proceed with OSM?
Discussion
Comment from Harry Wood on 27 January 2011 at 01:54
Nice presentation. I have some corrections/thoughts on some slides
Slide 22 - ...and bing imagery too now!
Slide 25 - Osmarender (perhaps more correctly, the tiles@home project) is no longer updating faster than the mapnik renderer. In theory distributed rendering might be a boost, but in practice mapnik on one machine can blast through the rendering tasks fast enough to cope with minutely diff updates, meanwhile tiles@home gets slowed down by upload bandwidth.
Slide 34 - Not sure if CloudMade are actually maintaining better uptime than the main OSM tile server these days. A reason for using CloudMade is to have paid-for support & guarantees. The central OSM tile server is not a guaranteed service (Tile usage policy)
Slide 16 - I would've thought the key point to add here, is that most free "Cloud" services provide flexible mash-up APIs (the basemap and geocoding examples of the previous slides) but do not provide access to their raw underlying data map data. OpenStreetMap does
I'm trying to picture end-uses which utility companies might have. I tend to think of them deciding where to dig a hole in the road using ultra-detailed "mastermap" GIS systems. Clearly OSM doesn't offer the kind of detail and accuracy (e.g. we don't have the width of sidewalks and the position of the curb down to the centimetre, and we're not really geared up to deal with that level of accuracy even if we did have the data) So perhaps you should mention that limitation up-front.
A selling point you didn't mention though: Licensed datasets, which companies may use internally, cannot be so easily re-published (e.g. on the web) without hitting restrictions. Some publishing tasks may not require ultra-detailed base maps, e.g. a map of where roadworks are being carried out in a city. Web mapping services are ideally suited to this.
I was going to suggest you get in touch with Peter Batty, but I see you're already referencing him on slide 33. Another suggestion for you SmallWorld experts. It could be helpful if you created a wiki page on SmallWorld. There's a red link on this page just waiting to be filled in. On Slides 29 and 30 you seem to be saying there's some built-in ways of working with OSM. That kind of thing could be documented on the OSM wiki. Also look at this from the point of view of explaining to OSMers about the needs of SmallWorld users. How could working with OSM be made easier? Detail SmallWorld data formats for which we might be able to develop converters.
Comment from daregusta on 27 January 2011 at 02:20
Really trying to think of a usecase here.
Big companies already have their GIS systems in place, often proprietary, with tailor-made extensions, tying in to all their other systems.
They pay a lot to create this data and to keep it accurate and up to date.
This dataset is very closely connected to their core business. As a rule, you do not outsource your core business.
So OSM sounds lovely, but it isn't.
- you can't rely on other people to map area's or changes because you don't know when they'll do it.
- you have nothing to say about the data quality. People will not follow your standards.
- The system doesn't connect to all your other systems
- Most importantly: You already have a system with high quality data in place. Why would you change it?
I can think of two applications though:
1. They can donate data. It'll need to overcome some internal resistance, but especially public companies may decide that their data is publicly owned anyway.
They can help to improve the quality of OSM data, which in turn may improve the quality of commercial data (by cross-checking) and lower the commercial prices (by competition).
2. In 3rd world or otherwise unmapped countries where there is no data available yet. OSM may be the system with the most data available.
Comment from Alfred Sawatzky on 27 January 2011 at 03:22
@Harry: Thanks for the comments and review of the slides. I will update them. I know @pmbatty quite well. At different times we have worked together and other times at competing consultancies :) But I always enjoy visiting with him. You know what they say in my industry... "It's a Smallworld..." Thanks for setting up the Wiki link. I will start filling out the page soon and then promote it in the Smallworld community.
@daregusta: Thank you for the points to ponder. I appreciate your suggestion of the two applications.
The big companies I have worked for are predominantly in the gas and electric space. And they all seem to have the same problem with their underlying landbase data. They get data from a vendor (e.g., Navteq, Teleatlas) and then get updates periodically (ever year or two). In the meantime, these utilities need to update the landbase with new estates, subdivisions, roads, etc. That all works well until they get the annual update from their landbase provider. Now what... how to integrate the landbase provider's new data with the new stuff that the utility is aware of. Maybe it made it into the new commercial landbase... maybe it didn't. So there is always a big integration effort whenever a utility wants to update their landbase. That daunting task often drives utilities to delay the landbase upgrade for years and only makes the inevitable upgrade even more painful. I see OSM as a great way to have a single source of data where utilities can add their local knowledge together with the knowledge of citizen mappers. At the end of the day, the utilities are not in the business of mapping landbase. They are in the business of maintaining their electrical/gas/water facilities. At least that is how I see it... trying to get the utility to leave management of non-core-business data to OSM.
Anyways, I do appreciate your comments. As I discuss this with more utilities and learn what their actual use cases could be, I will publish more about it. I think I might start a Utilities page on the wiki to track use case ideas.
I haven't created any pages on the wiki yet, so if you have a suggestion for page name or category, I would appreciate it greatly.
Comment from z-dude on 27 January 2011 at 04:05
What if your utility info is inaccurate? But people rely on it as fact?
http://envirolaw.com/burnaby-oil-spill/
If you add a pipeline to OSM, how accurate is the placement? Are you using accurate data in the first place? (ie data from 1957) What if natural ground movement moved the location of the pipeline? (all mountains move, even if it's a mm a year) What if the 'as built' did not match the as drawn engineering drawing?
Comment from blackadder on 27 January 2011 at 09:19
Being involved in the utility industry I can put another slant on this. The accurate position of underground utilities is not well known, even by the utility companies themselves. That’s why the industry is investing in research and further development of underground mapping tools using GPR (Ground penetrating Radar), ultrasonics and other methods. For some the situation is better than for others but the myriad of old and disused utilities alongside live utilities in the same street space means that for major urban centres the situation is far from clear and results in each instance of an excavation in a street requiring careful investigation, notification and often hand searching to prove that the way ahead is clear.
The second issue is one of asset security and commercial sensitivity. Generally speaking the utility companies do not want us to know where their assets are. They would rather keep this information to themselves and is one of the reasons why in the UK the National Street Gazetteer (The NSG) is a closed environment open only to the utility companies and Local Authorities for the purposes of Street Works co-ordination.
Finally there is accuracy. Utility companies are always improving the accuracy of their data relating to the utilities themselves, both in terms of the information on the asset and its location. This requires a robust system that is built on the basis of quality first.
Looking at OSM there are some opportunities in the utility sector as there are for OSM generally in the old-school GIS industry. As a base map it’s possible to easily recreate land detail, even for the full width of a public right of way (where the utilities are principally located) though at the moment we do not do so. Take a look at my local area which is now principally mapped out ( http://osm.org/go/euzaICBbz-- ) and you will see that while the majority of the landmass is mapped as it is on the ground (traced from BING imagery) the roads and other highway ways are still the centreline vector data rather than the edge detail required to define the true arrangement on the ground. Of course though I could add these highway edge details and tag them appropriately and then ignore the traditional highway ways when using the data in another application.
If I were to do this I would then be left with the issue of positional accuracy. Companies that use OS MasterMap base mapping are relying on data that is positionally accurate (relative and global) to within just a few centimetres. With OSM even when tracing from BING we probably can't realise global accuracy below 2m at best. We might do better with relative accuracy, perhaps even sub metre where aerial imagery quality is good, but anything better requires more accurate surveying techniques. For these reasons OSM is not that useful where precise positioning is important. For general data (eg for utilities and other organisations that need to record/display general asset location) then OSM has some merits, for others its doubtful at best.
I’m sure with time we will see new organisations take on OSM as a landbase source, much in the same was as Surrey Heath has done, ie where the product fits. For those that need positional accuracy of features I don’t see how OSM can ever compete, not unless some third party is able to take OSM data and verify and adjust the accuracy to some guaranteed level. Where accuracy is less important but change updates are vital then OSM may offer a better alternative to the traditional commercial suppliers though being able to guarantee as to what date updates for a given area are valid will always be difficult without third party verification.
Cheers
Andy
Comment from Tom Chance on 27 January 2011 at 10:53
Andy, that's some seriously detailed mapping! I thought I was obsessive with my house numbering, I'm one of only a few urban villages in London to do this:
http://www.openstreetmap.org/?lat=51.46107&lon=-0.07265&zoom=17&layers=M
I don't think I'm willing to do gardens and highway edges, too much bother for a voluntary effort.
Isn't it very unlikely we'll get that level of detail across the whole country from volunteers? Our national base map offering is more likely to be at the level of road centrelines and landuse areas for the forseeable future.
Comment from blackadder on 27 January 2011 at 14:58
Tom, Dulwich & Peckham looking great :-)
Regarding the level of detail I think its very likely that in pockets the full details will appear very soon. It only takes one mapper to get the bug and a town is as good as done. As with getting the basic road structure into OSM many places will lag several months or even years behind other locations, but eventually It will appear to some degree or other. As an example, this last week one of the Staffs mappers (iain1940) mapped Penkridge ( http://osm.org/go/euzVRUIO-- ), which had been white space till now. It's take 5 years to get to it. Where might we be in 5 years time? Perhaps Penkridge will have just had its houses and addressing done, its certainly possible.
Comment from Jean-Marc Liotier on 27 January 2011 at 15:52
A large French operator I work for officially uses WMS servers with proprietary data they license for internal use. The WMS background in the GIS software used there can be customized and I have also spotted users with Bing Maps backgrounds - but that is of course not official.
I bounced around the idea of using OpenStreetMap directly and got people quite impressed at the map, but the very idea that it is potentially incomplete and not an "official" source for which they can point fingers at a culprit when problems arise is putting them off - even though no one is deluded enough to think that having a culprit at hand will actually do anything.
I'll try again - I believe that packaged offerings from a reputed vendor such as Maquest or Cloudmade could make acceptance much more likely. At the scale of a large ISP, the cost is not the most important variable and no one has ever been fired from buying from the Institut Geographique National...
Comment from Jean-Marc Liotier on 27 January 2011 at 15:57
@Blackadder - For the local technical details of digging trenches to lay fiber, precisely surveyed plans are indeed required and OpenStreetMap cannot compete - we hand Autocad files to the digging crews.
But for higher level planning, deployment and management of the network, OpenStreetMap is totally sufficient - the proprietary map background I have seen in current use are worse than OpenStreetMap and their only value is that they do provide more even coverage in rural areas where OpenStreetMap still lacks.
Comment from blackadder on 27 January 2011 at 16:02
Jean-Marc Liotier, I agree, for network planning and management OSM is a very real alternative, especially if the map data is enhanced where required by the utility company or others along just the corridors and easements they are interested in.
Comment from IslandTime on 27 January 2011 at 20:49
I work with smaller utilities, primarily water/sewer and local multi-server operations. Many do not have internal GIS capabilities to update/maintain proprietary base mapping and rely on commerical maps or handouts from local governemnts. These utilities frequently run into to issue of what to do with new subdivisions or streets that are not on their basemaps. I have been suggesting that OSM might be an economical option for them to "fill in the gaps" at least until their main basemap service catches up. Not a perfect solution but a viable low-investment option. Hope to have someone take the plunge and implement soon.