I am using Erdas Imagine 2014 to reproject 4-band raster data from GRS 1980 NAD83 Lambert Conformal Conic (ft) to UTM GRS 1980 NAD83 North UTM Zone 11 (m). Unlike ArcGIS, Erdas provides the option to reproject irregular cells (where x != y) or force square pixels on reprojection (where x = y). What are the pros and cons of accepting the default irregular shaped cells versus forcing square pixels on the reprojection? I am particularly interested in how these options could effect processing efficiency, accuracy and geoprocessing within ArcGIS.

enter image description here

  • The size of the area (is it completely within zone 11 north?) may affect the answer. – mkennedy Jul 25 '14 at 23:40
  • The size of the area is ~ 1/4 the size of Oregon and it is entirely within UTM Zone 11. – Aaron Jul 26 '14 at 1:48
  • ArcGIS will do rectangular cells. To get the option open the layer in ArcMap right click and go to export which will bring up a dialog with lots of options. To reproject this way set the frame reference to the to spatial reference and then select match data frame in the export dialog. – Michael Stimson Jul 28 '14 at 1:22

It depends on the type of analysis you are planning and how irregular your cells are. Anytime pixels are forced to become square, some amount of resampling and deformation is going to take place. You would need to provide more information about the satellite sensor your data originated from and what you're going to be doing with it in ArcGIS before I could tell you if it's appropriate for what you're doing.

But, ArcGIS will force raster pixels to be square upon import anyway. For MODIS swath data, this can significantly change the extent of the layer and make results of many analyses completely unreliable, even if not projecting the data. If you started out with square pixels, projecting may change the shape, and forcing them to be square upon reprojection will involve resampling or deformation. For a Lambert to UTM reprojection, any distortion or resampling shouldn't have a large effect on results.

What I normally do is keep everything in feature format. I.e. represent each cell by the centroid point, and use the Thiessen polygon tool to create polygons for each cell after projecting the points to the new coordinate system if I need them. This also makes it possible to use certain tools in ArcGIS that won't work with raster data, allows you to store more attributes for each point and do quick calculations using the reflectance values in each of the bands just through the attribute table, and can lead to speed improvements for certain analyses.

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.