[GRASSLIST:2516] Re: Which projection?

[Rich Shepard]

On Fri, 28 Sep 2001, Gordon Keith wrote:

> I have a gridded data set I want to read into grass, but I'm not sure
> which projection to use. The header for the data contains the following
> information:

> Coordinate system: Lat/Long

> I can get the same data output using a different coordinate system with
> the header:
>
> Coordinate system: merc

> I'm not very familar with projections and don't understand the
> significance of the half axis and flattening terms.

Gordon,

  Lat/long is a coordinate system for locating points on the surface of the
Earth. It is not a "projection" but actual locations. Underlying all this is
the ellipsoid, or "datum", which is a mathematical model of the Earth's shape
and used to calculate both the coordinates and the elevation above some
reference level (usually either sea level or HAE, Height Above Ellipsoid).
As an aside, this mathematical model of the Earth is why GPS accuracy for
elevation is 2-3 times that of the horizontal position.

  The second set of values appears to be using the Mercator projection with
a coordinate system based on metres (or meters, let's not argue). A
projection (Mercator, Transverse Mercator, Unversal Transverse Mercator --
UTM, Albers Equal Area, Conformal Conic, etc.) are mathematical formulae to
convert -- project -- points located on the 3D Earth to a two-dimensional
piece of paper. Because this cannot be perfectly done, different projections
have varying accuracies and distortions, and each is best suited to a
different part of the globe and for areas of different sizes.

  r.proj will convert from a strictly-coordinate-system lat/lon to an actual
projection, and among projections. It will not do datum shifts (which are
changes incurred when one uses different models of the globe's shape).
Regardless, GRASS should read your lat/lon values quite well. Are you using
a sites (s.*) module? Do your data represent elevations? If so, look at one
of the DEM-importing modules.

> The data is gridded, but the grid coordinates don't follow lines of
> latitude and longitude, so I haven't had good results reading the data
> into grass using a latitude longitude projection.

  And they wouldn't if they were collected by satellite, aerial photography,
or someone walking/driving across the landscape. Just because they're on a
grid does not require the grid to be precisely laid out along lines of
latitude or longitude.

> I will need to work with this data in latitude/longitude coordinates.
> Does this mean I will have to reproject (r.proj) the data into a latitude
> longitude projection, if I get it into grass successfully?

  I believe that your problem is selecting the approprite import module. If
you need to work in unprojected coordinate units, then stay with that. If
you take your original data, project it, then un-project it back to lat/lon
you'll lose accuracy and precision.

HTH,

Rich

Dr. Richard B. Shepard, President

                       Applied Ecosystem Services, Inc. (TM)
            2404 SW 22nd Street | Troutdale, OR 97060-1247 | U.S.A.
 + 1 503-667-4517 (voice) | + 1 503-667-8863 (fax) | rshepard at appl-ecosys.com
                 Making environmentally-responsible mining happen.