2

My question is about how coordinate systems affect comparison of different layers - I have tried to refresh my memory about coordinate systems in GIS, but am still unsure about this situation.

I am comparing the land cover classifications (estimates of the % each land cover type represents in the given area) from a raster layer versus from a vector shape used in photo-interpretation. This comparison will be done on many raster-classification/photo-interpretation combinations throughout the United States.

The area of the vector layer comes from delineating a watershed (generally 25 to 250 square kilometers in area), and that delineation and vector layer is then used to clip that area out of a larger land classification raster layer.

The vector layers used in photo-interpretation are all in lat/long (unprojected). The photo-interpretation is already done, so I need to keep these source shapes as they are. These vector layers have the following coordinate system:

Geographic Coordinate System:   GCS_North_American_1983
Datum: D_North_American_1983

All of the raster layers I'm working with have the following coordinate system:

Spatial Reference:  Albers_Conical_Equal_Area
Datum: D_North_American_1983

Are the areas of these 2 GIS layers the same, given they have the same Geographic Coordinate System, or am I in error to compare the vector layer to the raster layer, given that the vector layer is not projected and the raster layer is? If the latter, then can I simply project the vector layer to the same coordinate system as the vector layer, and if so can that be done after clipping the raster to the vector shape or must the projection be done before clipping?

  • 2
    You question mentions the "same PCS as the vector layer" but your vectors don't have a PCS, they have a GCS (Geographic coordinate system, not Projected). If you care about area, do not deproject the rasters. Project the vector to Albers (the same exact Albers, since the projection differs with different parameters). You can deproject (carefully) back to the GCS after analysis is complete. – Vince Dec 1 '15 at 17:31
  • 1
    Your want/hope cannot be realized. Area comparison in Cartesian degrees is useless. Deprojecting the rasters will warp them and render them unfit for comparison (each row of pixels will have a different area). Photo interpretation done in a GCS coordsys is likely of questionable value; the areas would have to have been mensurated geodetically (which deprojecting should retain, but without the extra bother). No projection can preserve size and shape perfectly (especially with respect to orientation), but an Albers over a large study area doesn't do an awful job. – Vince Dec 1 '15 at 18:41
  • 1
    I must say I second Vince, for area calculations, it is best to convert your vectors to the Albers EA projection. Apart from being appropriate it is easier to communicate your results in m2, km2, miles2 etc... rather than dec. deg. squares, which nobody likes. – yanes Dec 1 '15 at 18:44
  • 2
    It sounds like an incorrect procedure, for the reasons already given. – Vince Dec 1 '15 at 19:34
  • 1
    @yanes I did as you suggested. I don't care to keep the rasters in albers equal area. My goals are to keep the vectors in their original coordinate system to ensure the photo-interpreted land cover %s remain true to the shapefile; and to get the rasters in a coordinate system equivalent to the vectors so their land cover %s can be compared. So, I re-projected the raster layer into the GCS defined by the vector layer. – cr0 Dec 1 '15 at 20:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.