1

I have a Landsat ETM+ image, I also have 200 sample points with coordinates,canopy density, different band values and NDVI values.

If I build OLS model I only got sample points regressed.

How can I join the image and the points so I can do regression and produce a map in ArcGIS for Desktop?

1

In ArcGIS I'll give you a blueprint to follow.

  1. Convert your sample points into a shapefile.
  2. Project / Reproject your shapefile to match your ETM+ image, mostly likely UTM but not always.
  3. Use the Sample or Extract multi-value to points tool to bring the values from the ETM+ image to your shapefile.
  4. Run the OLS in ArcGIS.
  5. Render the shapefile with the OLS values. Usually some form of residual map.
  • Thanks for answering!This is how I did yesterday:1),exported all sample points by first ‘Display XY data’ and then ‘Export data’; 2),use "Sample" extracted all points from ETM+ image, then I got a table and exported it to a DBF file. 2), I copied all band values from DBF to original sample .xls file, calculated NDVI; 3), I also did linear regression in .xls, since I'm asked to apply the equation to the EMT+ image for mapping, I joined the .xls (with NDVI) to sample points shape file; 4), ran OLS, failed. – ZhangYao Jiayin Sep 23 '14 at 6:11
  • Between steps 3 & 4. Do not run 0LS on the joined data. Export to a new shapefile and run OLS on this new file. – If you do not know- just GIS Sep 23 '14 at 12:34
  • still not working efter join→export....only got sample points regressed... – ZhangYao Jiayin Sep 23 '14 at 15:21
  • Got a PDF from my teacher, we have to use Image Analysis in ArcGIS to calculate NDVI for ETM+ image, then applying NDVI raster layer in regression equation in Map Algebra. I can produce the map now except it seems I got different Band values after 'Sample': I got NP_2_30513,14...17, I assumed they are band3,band4...band7 and used them calculated NDVI in Excel. Did I interpolate them wrong? Is that why my regression is different from my techer's? – ZhangYao Jiayin Sep 23 '14 at 17:16

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.