NetCDF to Raster File Extent / Cell Size Change - raster

Question

In using the ArcMap tools, as well as an arcpy script, I am getting the same results. When I try to convert a NetCDF (NC) file to a TIF, the file extent increases by a very small amount which also slightly increase the Y cell size.
The only way I can get the images to match is to use the Export Data menu item when I right click the file layer on the Table of Contents.
Here is my workflow:
NetCDF to Raster Layer
Copy to Raster or Project Raster (both tools create the same problem)
Here are the original layer properties in the initial NetCDF to Layer:
Cell Size (X,Y): 0.099990845, 0.099998474
Here are the extent values:
Top: 89.9972572327
Left: -179.999992371
Right: -179.967048645
Bottom: -89.9999961853
Here are the values after I copy to Raster or Project Raster:
Cell Size (X,Y):0.099990845, 0.1000464
Top -90.0973036279
Left -179.999992371
Right- 180.167030334
Bottom -90.0862541199
I have tried to include a NoData value, background value, snap to image, and even include the original extent values of the NC file within the raster tools, but it always ends up increasing the extent which in turn increases the cell size, but a tiny bit - < 0.01 deg, but it's enough that my pixels all shift a tiny bit downward.
I have also tried to change the compression to none and it still does not help.
I have included a transformation and output coordinates in the tools as well. The datums of the files appear to be the same from the source info. Otherwise, using default settings.
It's a single variable file and it is already saved out by a date so it's pretty straight forward. Any ideas why this is happening?

Answer 1

Finally got the image situation figured out. For Arc and ENVI, we just need to specify the cell size because the original extent of the image is just outside the normal extent of a map. Arc goes ahead and adjusts the cell cize to maintain the normal extent (-180,180,90,-90). It's because of the floating point lat and lon coordinates and extra cells on the edge of the file.
Just as in R, once everything is specified, the floating point values are rounded and the image loads in just fine.

Answer 2

This is not a coding question, but you could try a question like this on https://gis.stackexchange.com/.
Here is a way to convert a netcdf file to geotiff using R, that you can use as comparison:
library(raster)
library(ncdf4)
r <- raster("3B-MO.MS.MRG.3IMERG.20140401-S000000-E235959.04.V05B.HDF5.nc")
x <- writeRaster(r, "file.tif")
Now let's look at the dataset:
r
#class : RasterLayer
#dimensions : 3600, 1800, 6480000 (nrow, ncol, ncell)
#resolution : 0.1, 0.1 (x, y)
#extent : -90, 90, -180, 180 (xmin, xmax, ymin, ymax)
#coord. ref. : NA
#data source : 3B-MO.MS.MRG.3IMERG.20140401-S000000-E235959.04.V05B.HDF5.nc
#names : precipitation
#zvar : precipitation
The extent is global and clean: -180 to 180 longitude and -90 to 90 latitude. The resolution is 0.1 degrees in both directions.
That is also what the ncdf metadata say:
print(r)
# ...
# Grid.GridHeader: BinMethod=ARITHMETIC_MEAN;
#Registration=CENTER;
#LatitudeResolution=0.1;
#LongitudeResolution=0.1;
#NorthBoundingCoordinate=90;
#SouthBoundingCoordinate=-90;
#EastBoundingCoordinate=180;
#WestBoundingCoordinate=-180;
#Origin=SOUTHWEST;
Somehow, Arc seems to get it wrong