1

I am currently designing a geospatial product that will be primarily used for large and intermediate scales that cover an entire state of the US or more.

The UTM projection would be ideal because of its conformal properties.

But, I am not crazy about it for the product I am designing because of states that may encompass more than one UTM zone. Of course I could render all of the imagery in one zone or the other, but it seems problematic, especially for possibly combining the products together some day in the future into a national US dataset. My source data is in variety of projections, that I will be reprojecting to a common projection.

So I need to decide on my common server-side projection, and possibly a different projection for my clients (if needed but not desirable). Is Albers Equal Area a good choice?

EDIT: The product shall be in the form of topographic maps in both Large Scale (1:24k) and Intermediate Scales (1:63k).

EDIT: I should like my national projection to minimize distortion when projecting to a client projection, as I will have pre-rendered raster images - not just vector data, so I don't think geographic is best.

EDIT: Here is another way to ask the question (I'll figure the rest out on my own or ask as a second question). For a Large Scale topographic viewer, that could view across multiple UTM zones and multiple State Planes at any time depending on its center, what projection would you recommend serving to the view?

Jeshua Lacock
  • 171
  • 1
  • 8
  • 2
    Something similar has been asked at http://gis.stackexchange.com/questions/104005/which-projected-coordinate-system-should-be-used-for-mapping-the-us – PolyGeo Oct 11 '14 at 05:22
  • It may help to explain the purpose of the maps or the calculations to be performed upon the data. – Martin F Oct 11 '14 at 20:29
  • See edit, please let me know if you have any specific questions. – Jeshua Lacock Oct 11 '14 at 21:15
  • Why include such nonsense as "UTM projection because of its non-linear anisotropic scaling in both easting and westing."? – Martin F Oct 12 '14 at 01:09
  • Why do you consider that as nonsense? It is an attribute of the UTM projection that I find desirable. I like the fact that the projection is based on a regular grid. For instance, If I have a raster image in the UTM projection, all I need to know is the corner easting, westing and resolution to determine any cell's coordinates. That isn't the case with say Lonlat. – Jeshua Lacock Oct 12 '14 at 01:22
  • 3
    But that applies to all projected CRS. You can have projections with good results in distance, area or angles. It is up to you what is important. – AndreJ Oct 12 '14 at 06:17
  • That is true, but I was stating what one of the desirable projection parameters are. – Jeshua Lacock Oct 12 '14 at 19:21
  • Anyways - thanks for the link PolyGeo, it looks like my hunch about using Albers Equal Area is correct. – Jeshua Lacock Oct 12 '14 at 19:22
  • 1
    If area calculations are important for your analysis, Albers Equal Area is an excellent choice. Additionally, the Albers projection is ideally suited for landmasses oriented E-W, rather than N-S. Keep in mind that you can create a custom Albers Equal Area projection by adjusting the standard parallels to your specific area of interest. This will further decrease distortion. – Aaron Oct 15 '14 at 04:19
  • That statement, @martin f, is not nonsense--it's just wrong. Because UTM is conformal, its scale factor is isotropic (the same in all directions), as you point out in your (nice) answer. Nevertheless, the thrust of the question seems clear. – whuber Oct 15 '14 at 19:00
  • Well then I suppose one of you guys might want to update Wikipedia. ;) – Jeshua Lacock Oct 15 '14 at 19:14
  • Also, @martin f is not so much an answer to my query as it is a diatribe on projections, but it is nice I will give you that. State plane does not solve my issue. I think I made that clear in my post citing the UTM zones being an issue, and the first word of the title is Regional which I believe can encompass state lines. – Jeshua Lacock Oct 15 '14 at 19:19

1 Answers1

4

Traditionally, topographic maps are plotted using a conformal projection, not an equal area projection. A conformal (or orthomorphic) projection preserves angles and, at any point, has isotropic scaling. Such qualities allow relatively easy computations involving angles and distances.

en.wikipedia.org/wiki/Map_projection#Conformal
www.progonos.com/furuti/MapProj/Dither/ProjConf/projConf.html

There is still variation in scale, however, throughout the mapped (projected) region. To minimize the overall scale distortion, conformal projections are usually chosen with two standard (or secant) lines, and those are oriented along and placed on either side of the region's "main axis", if it has one.

States in the USA are of different size, shape and orientation. Each one uses it's own "State Plane" projection for topographic mapping, typically a transverse Mercator or a Lambert conic.

en.wikipedia.org/wiki/State_Plane_Coordinate_System
vterrain.org/Projections/spcs.html

I'd suggest you investigate the State Plane Coordinate Systems already in use and follow in the footsteps of prior experts.

To map a very large region, such as the entire USA, the Lambert conic conformal is often used, with standard parallels at, say, 33 and 45 degrees north.

en.wikipedia.org/wiki/Lambert_conformal_conic_projection
www.wired.com/2013/11/projection-lambert-conformal-conic

I should like my national projection to minimize distortion when projecting to a client projection

Projection is the conversion of geographic coordinates to planar coordinates. If your source data are already in a known projection then the process of converting them to another projection, often called reprojection, is really two conversions: reverse-projection from the source planar system back to geographic and then normal projection again to the target system. Thus, assuming your tools are projecting and reprojecting correctly, your source projection system -- your single "server-side" projection -- has no affect on the distortions present in your target (i.e., client) projections. It can, however, affect performance significantly.

Google Earth uses the simple cylindrical (or equirectangular) projection for its imagery base. Since such a projection is trivial it will have negligible affect on performance during reprojection.

support.google.com/earth/answer/148110?hl=en

whuber
  • 69,783
  • 15
  • 186
  • 281
Martin F
  • 8,948
  • 36
  • 58
  • I realize most topographic maps use conformal - as most every topographic product I have seen is in UTM. I was considering Albers Equal Area for a common national projection because of the inherit issues with multiple UTM zones in one state. – Jeshua Lacock Oct 12 '14 at 23:27
  • What happens if I have each state projected in its own State Plane and I happen to want to map centered on a state border? I will have to re-project correct? That is what I am trying to avoid for my server side common projection format. – Jeshua Lacock Oct 12 '14 at 23:30
  • @JeshuaLacock - See addendum. – Martin F Oct 13 '14 at 00:19
  • So you would suggest using Lambert conic conformal for the server side projection, but do you think it is suitable for the topographic client (could be printed product or digital) which will more often than not be viewed at large scales? Or would you suggest reprojecting my potentially national dataset to a different client projection such as State Plane? – Jeshua Lacock Oct 13 '14 at 00:39
  • For a national DB you should use a single coordinate system -- even geographic might be good. For display/printing of local regions, yes, you're best to use whatever local projections are common/convenient. – Martin F Oct 13 '14 at 00:54
  • I should like my national projection to minimize distortion when projecting to a client projection, as I will have pre-rendered raster images - not just vector data, so I don't think geographic is best. That is kind of the root of my question, I apologize for not being more clear. – Jeshua Lacock Oct 13 '14 at 02:20
  • 2
    I want to accept your answer because I do appreciate you trying to help. On the other hand, it seems as if you did not understand my query before you answered. So the way your answer is phrased does not seem to be a high quality answer to my query. You essentially state what is ideal for topo maps (not the question) and that sure use any projection I fancy for my server side (as an afterthought). – Jeshua Lacock Oct 14 '14 at 05:22
  • My question was stated quite clearly and yet instead of actually trying to understand what I was asking you choose try to belittle me. This was my first and last question posted here which is a shame - I really like Stackoverflow. – Jeshua Lacock Oct 15 '14 at 03:24
  • I'm trying to understand. Your question may be clear to you but not to me. I'll alert others to it, in case they can help... – Martin F Oct 15 '14 at 03:28
  • see latest edit. – Jeshua Lacock Oct 15 '14 at 04:15
  • I disagree about what you have stated about raster images not being distorted (some more than others) by reprojection. The more severely a raster image is distorted the detrimental the operation is to the resulting image. – Jeshua Lacock Oct 15 '14 at 04:29
  • See added note about Google Earth – Martin F Oct 15 '14 at 14:31
  • You may be an Expert but I don't think you are understanding my question. I rephrased my question for you, that is very straightforward, and you give me a completely different answer than I asked. – Jeshua Lacock Oct 15 '14 at 18:58
  • Google Earth (or rather Google / Web Mercator) is worth for nothing: http://www.hydrometronics.com/downloads/Web%20Mercator%20-%20Non-Conformal,%20Non-Mercator%20(notes).pdf – AndreJ Oct 15 '14 at 19:08
  • @martn f you state Google Earth uses Cylindrical for its imagery base, but what projection is served in to the clients? Do you think the projection served is appropriate for viewing large scale topographic maps? – Jeshua Lacock Oct 15 '14 at 19:30