How to meaningfully manipulate the cell size of a Kriging or IDW interpolation from 0.09 to 30 meters or 100 meters resolution?

3754
5
04-11-2016 07:30 AM
Highlighted
New Contributor

Good morning! I'm running into a spatial resolution mismatch problem between the NLCD land cover raster (30 meters or 100 meters cell size) and my interpolated rasters (Ordinary Kriging and IDW with a default cell size of [X = 0.096577988, Y = 0.096577988] developed using bird point data from the lower 48 states. I have even tried to manually set the resolution for the interpolated rasters during the analysis to 30 or 100 meters, and by doing so, the interpolation analysis won't yield an output raster. It might be some technical aspect that I'm missing or conceptually something else. Any leads regarding this will be greatly appreciated! Cheers!

Reply
0 Kudos
5 Replies
Highlighted
MVP Esteemed Contributor

These numbers ... 0.096577988... strongly suggest that the units for that file are unlikely millimeters. I suspect that they are fractions of a decimal degree and you didn't project that layer prior to running the interpolation.  The best thing would be to project that file using your desired cell size in order to facilitate comparison.  It will take less time and there will be less issues that trying to get things to be matched.

Highlighted
New Contributor

Thank you Dan! I was afraid that could be the case. Now, if the point data is projected to GCS_NAD_83 with angular unit in degrees, which  Projected Coordinate System would you recommend for the continental USA that provides "meters" or "feet" in the angular units (maybe, NAD 1983 Contiguous USA Albers or USA Contiguous Albers Equal Area Conic?). Your advise is greatly appreciated!!

Reply
0 Kudos
Highlighted
MVP Esteemed Contributor

NAD 1983 Contiguous USA Albers seems to be popular in the U.S (it is conformal, if I recollect... which is a kind-of good at everything projection).  It will at least yield you results more appropriate than those derived from using decimal degrees

Highlighted
New Contributor

Hi Dan,

I have tried defining the projections to existing data points and even add new unprojected data points to an unprojected data frame to see if I can project them all to NAD 1983 Contiguous USA Albers. However, that projected coordinate doesn't align with the spatial reference of the NLCD layer (only GCS_NAD_1983 seems to work for all layers). None of the NLCD rasters, point features and subsequently interpolations seen to align meaningfully in any other spatial reference that isn't GCS_North_American_1983. This proves somewhat odd and puzzling to me since ArcMap continues to set the angular unit to degrees (0.01745329...) for anything I add not matter the defined projection, it force the cell size for the interpolated raster to [0.9548, 0.09548] as the only functional output.

>I do thank you for your time and effort in helping my case, if you or anyone else have gone through similar problems, please feel free to comment.

Reply
0 Kudos
Highlighted
MVP Esteemed Contributor

The Define Projection tool... should only be used when you get a warning that there is no projection for the layer.

The Project Tool Project—Help | ArcGIS for Desktop is the tool you use to produce a new file which has projected coordinates.

If the Define Projection tool was used erroneously... undo the definition.

When you project your data (say to Albers) I find it good practice to add the projected layer to a new data frame.  This does two things.

  • it ensures that the data frame is define appropriately... in this case it will have a projected coordinate system (lets say... Albers)
  • The data frame does not get polluted with interlopers that may be masked in a different coordinate system, taking over the projection of the data frame.,,, curses to projection-on-the-fly

So if any sounds familiar, it isn't too late to fix it.

One last thing... you should never do interpolation using unprojected data (aka data in decimal degrees).  Why? well the X,Y and Z values are not the same... and the difference in Y values for an equal increment is not the same as the X values (unless you are just at the equator) and the difference gets worse as you move poleward.

Good luck.