Saipan data is in a sad state

Posted by guamGIStech on 25 May 2013 in English (English)

I've noticed that while the data for Guam is generally pretty good, the data for Saipan is in a pretty sad state. In particular, the streets are distorted, and up to 60 meter off from their true location. You can see this by loading the streets on a Bing or other satellite image. This is starting to create problems for folks using GPS and other tools that use OpenStreetMap as a base source.

Most of this appears to be due to bad TIGER data during the 2008 import. Back in the bad old days, Saipan and the rest of the CNMI used a variety of old datums and projections, some of which weren't clearly documented. The TIGER data was on one of these, as was a lot of our old data from USGS, and the roads and related data are offset and distorted as a result. Even Google Earth uses bad old data.

My goal over the next six months is to replace all the distorted road data with TIGER 2012 data, or with equivalent data sourced directly from the government of the CNMI (i.e. Saipan). I'm frequently on Saipan and have worked with their data and analysts, so am confident that I can make this work.

Most of the TIGER data for Saipan is unreviewed and unedited, so doing a batch replace will have limited impact on existing users' work. A similar project was done for parts of Alaska by pnorman.

Things to do: 1. Update shoreline with NOAA data 2. Update shoreline with 2013 imagery to reflect ongoing erosion 3. Locate appropriate data 4. Figure out the scope of the problem (e.g. only roads, all TIGER linework, GNIS?) 5. Make sure folks in the area/community are on-board. 6. Prioritize fixes (probably start w/ roads, as that's most annoying to me) 7. Start implementing fixes by priority 8. On-the-ground GPS work to check fixes 9. Invite other folks to review

Location: West Harmon Industrial Park Road, Tamuning, Guam, 96913, United States of America

Comment from Wynndale on 28 May 2013 at 17:18

Latitude and longitude can vary depending on what geodetical system is used. There was a problem with the original TIGER import that, although the mainland coordinates used WGS84 which we use, some islands did not.

We have also had problems with imagery being offset so if there any GPS traces for Saipan you should check what you are doing against them.

TIGER data is also of variable quality so it could simply be wrong.

Comment from guamGIStech on 28 May 2013 at 21:02

Thanks for taking a look! I'll take some extra GPS traces when I head up to Saipan next week; so far the imagery has been pretty good but I'll make a point of doing some systematic comparison work. I'll try to pull in some other imagery options, too.

I'm going to check with the government on Saipan to see if they've got a publicly available data layer we can use instead of TIGER; I suspect the 2012 TIGER data reflects updated data provided by them (much like Guam did a few years back). The 2012 TIGER is a huge improvement over the 2005 TIGER data, from what I've seen.

Login to leave a comment