A customer has written us with the following question about using DAYMENT data within IDL: I am analyzing DAYMET data from mosaiced images for the entire western U.S. for the years 1993-2014 For 93-2012, I extracted values for 100 plots using lat/lons from ArcGIS and Python. I recently acquired 2013 & 2014 data but no longer have an ArcGIS license so I used IDL instead. As a test, I re-ran the extraction for one variable (precipitation) for 2005 in IDL to compare against the prior ArcGIS output. Approx. 6% of the values were different and the correct values were from the GIS extraction. In a random review of selected differences, the lat/lon point was close to a boundary, so I think the problem is a numerical precision issue. I know that the precision of the lat/lons was the same in both GIS and IDL. Is there a known numerical precision restriction? Or perhaps there is a better way to extract DAYMET values (other than xy = Map_Proj_Forward(pt_lon, pt_lat, Map_Structure=mapStruct) pt_x = xy[0] pt_y = xy[1] xindex = Value_Locate(uvec, pt_x) yindex = Value_Locate(vvec, pt_y) ?