Timeline for Choosing UTM zone to use for large country?
Current License: CC BY-SA 3.0
23 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Apr 18 at 15:00 | comment | added | Joooeey | The easiest way to sanity-check your results is to use the web mercator projection for everything - calculating pythagorean distances with that projection should give you values not more than 1% off for Indonesia since it's all near the equator. Web mercator is for the equator what the universal transverse mercator is for a meridian. But it uses a spherical instead of an ellipsoidal model of the Earth, leading to extra inaccuracies. | |
| Nov 28, 2017 at 1:45 | history | edited | PolyGeo♦ | CC BY-SA 3.0 |
deleted 8 characters in body; edited title
|
| Apr 13, 2017 at 12:34 | history | edited | CommunityBot |
replaced http://gis.stackexchange.com/ with https://gis.stackexchange.com/
|
|
| May 1, 2015 at 0:46 | history | tweeted | twitter.com/#!/StackGIS/status/593939456962514944 | ||
| Apr 7, 2015 at 20:24 | vote | accept | marquisdecarabas | ||
| Apr 7, 2015 at 19:59 | comment | added | Chris W | That's easy enough. I've converted my comments into an answer. | |
| Apr 7, 2015 at 19:55 | answer | added | Chris W | timeline score: 6 | |
| Apr 7, 2015 at 13:24 | comment | added | marquisdecarabas | @ChrisW -- I wish I could pick this response as the answer, because I think that is what I will need to do. Yesterday I reached out to a contact in Indonesia that I forgot I had, and he e-mailed me back this morning essentially saying that I need to calculate the length separately for each UTM zone to minimize distortion. | |
| Apr 6, 2015 at 10:11 | answer | added | mdsumner | timeline score: 4 | |
| Apr 6, 2015 at 7:02 | comment | added | Chris W | Note you don't need to split it to separate layers to do that. Also note you'd Intersect with UTM zones first, then intersect it with district boundaries, and then go zone to zone calculating distance. If you don't do both intersects first, the length you calc won't be for the sections only within a district, at least for any roads crossing a border. In ArcGIS you'd use Summary Statistics to total roads per district, but in QGIS I think you need the GroupStats plugin or something similar. | |
| Apr 6, 2015 at 6:57 | comment | added | Chris W | The more zones you are away, the more it matters. I found this page with a graphic example. I'm still not confident in suggesting a CRS, but maybe EPSG:3001. You could also find a dataset that is the UTM grid, Intersect that with your roads, and then start reprojecting the data into each UTM zone only calculating the length for records that have that UTM zone ID, then reprojecting (NOT DEFINING) to the next zone. Be wary of spatial join, especially if roads cross borders - you need intersect to cut them at the border. | |
| Apr 5, 2015 at 17:54 | comment | added | marquisdecarabas | @Spacedman -- I plan to spatially join the roads shapefile that contains the newly-created length_km variable with a shapefile of the country's district boundaries (I was hoping to get more granular than district, but I could not locate any sub-district shapefiles on the Web), and compute the total kilometers of roadway within each district. I will then import the data table into Stata and merge with my analytic dataset and use total roadway kms as a control variable in my regression models. | |
| Apr 5, 2015 at 17:33 | answer | added | Bryan Waller | timeline score: 2 | |
| Apr 5, 2015 at 17:24 | comment | added | Spacedman | If you have a large area and you want distance measurements accurate to a high precision then you should stick to WGS84 Lat-Long and compute the distances via great-circles on the WGS84 ellipsoid. What are you doing with the distances once you have them? | |
| Apr 5, 2015 at 16:03 | comment | added | marquisdecarabas | Thanks, @GerardoJimenez, what would you recommend instead of a UTM in my case? | |
| Apr 5, 2015 at 14:47 | comment | added | Gerardo Jimenez | Being near the equator also matters since it does have a double false value to asign the north value. For features north of the equator its false value is 0 meters. For features south the equator its false value is 10,000,000 meters. In your case, you have both cases, areas to the north and south of the equator making the use of a UTM CRS a little bit problematic | |
| Apr 5, 2015 at 8:28 | history | edited | marquisdecarabas | CC BY-SA 3.0 |
added 2 characters in body
|
| Apr 5, 2015 at 4:53 | comment | added | marquisdecarabas | @ChrisW thanks for your response and the explanation. I guess my question is how much does it matter when the country is on the equator (as opposed to a country closer to the poles), but obviously, my knowledge is limited. I suppose I could spatially join the OSM layer with a UTM layer and generate separate streets layers for each UTM zone, stitch the layers back together when I'm done creating the length_km, and then create the roadway-miles-per-district variable at the very end | |
| Apr 5, 2015 at 2:24 | comment | added | Chris W | You use a UTM zone when your area of interest fits completely within it or very nearly so. A UTM zone is not appropriate when your area of interest spans several zones like that. The further away from the zone you pick, the more distortion there will be. You actually want a projection designed to cover that area (and I can't suggest one which is why this is a comment and not an answer - we do have some experts who may respond, or I'll try to dig one up later). And yes, once you have an appropriate projection it's a simple matter to calculate the length in that projection to whatever unit. | |
| Apr 5, 2015 at 1:41 | history | edited | marquisdecarabas | CC BY-SA 3.0 |
added 188 characters in body
|
| Apr 5, 2015 at 1:26 | history | edited | marquisdecarabas | CC BY-SA 3.0 |
formatting
|
| Apr 5, 2015 at 1:21 | review | First posts | |||
| Apr 5, 2015 at 2:04 | |||||
| Apr 5, 2015 at 1:18 | history | asked | marquisdecarabas | CC BY-SA 3.0 |