Hi I have Land Surface Temperature (LST) data from MODIS. The data are in .hdf format, and have a geographic lat/long equal area projection of 0.05 degree x 0.05. ArcGIS 10.1 is able to open and display the data as hdf no problem. However, when I try to do any processing in Arc (e.g., copy raster, project raster), I have to change the output format to something like .img, .tif, ESRI grid, etc, and this act of converting the file type alters the pixel values. I know that reprojecting a raster necessarily alters the values because it is interpolating the data over different areas, but I don't understand whey changing the format of the imagery would change the values so much. For example, in hdf format the values range from 9.67 to 64.85. When I convert the raster using copy raster, the values are -51.33 to 69.29. When I try reprojecting the raster, the values become -31.01 to 84.19. The latter values also result when I try using the MarineGeospatial Ecology Tools. I tried converting the dataset in ENVI using the save as function, and also using the Modis Reprojection Tool, and in these cases the range of pixels values are closer to the true values, but the NoData value is lost / altered, so it doesn't seem like a good solution. Any ideas and suggestions would be greatly appreciated!!
Altering the range of values (the statistics min/max) you see in the TOC is quite different from your assertion that the software actually alters the pixel values when doing a copy. Most likely during the conversion process, pyramids are being created along with statistics (which are built off a pyramid for performance) if you have default environment settings. The statistics on a pyramid are basically always different than statistics built on the source value. Have you used Identify button to confirm that the original raster and the copied raster have different pixel values for any given pixel location? If you want statistics to be built off the source data, then you have to delete the pyramids and stats, then build stats, then build pyramids if you want them. To avoid this in the future you can change your environment settings and turn off build stats/pyramids.
Altering the range of values (the statistics min/max) you see in the TOC is quite different from your assertion that the software actually alters the pixel values when doing a copy. Most likely during the conversion process, pyramids are being created along with statistics (which are built off a pyramid for performance) if you have default environment settings. The statistics on a pyramid are basically always different than statistics built on the source value. Have you used Identify button to confirm that the original raster and the copied raster have different pixel values for any given pixel location? If you want statistics to be built off the source data, then you have to delete the pyramids and stats, then build stats, then build pyramids if you want them. To avoid this in the future you can change your environment settings and turn off build stats/pyramids.
Hi I have Land Surface Temperature (LST) data from MODIS. The data are in .hdf format, and have a geographic lat/long equal area projection of 0.05 degree x 0.05. ArcGIS 10.1 is able to open and display the data as hdf no problem. However, when I try to do any processing in Arc (e.g., copy raster, project raster), I have to change the output format to something like .img, .tif, ESRI grid, etc, and this act of converting the file type alters the pixel values. I know that reprojecting a raster necessarily alters the values because it is interpolating the data over different areas, but I don't understand whey changing the format of the imagery would change the values so much. For example, in hdf format the values range from 9.67 to 64.85. When I convert the raster using copy raster, the values are -51.33 to 69.29. When I try reprojecting the raster, the values become -31.01 to 84.19. The latter values also result when I try using the MarineGeospatial Ecology Tools. I tried converting the dataset in ENVI using the save as function, and also using the Modis Reprojection Tool, and in these cases the range of pixels values are closer to the true values, but the NoData value is lost / altered, so it doesn't seem like a good solution. Any ideas and suggestions would be greatly appreciated!!
Hello,
I have MODIS HDF files but I don't know how can I convert them to ArcMap 10.1. I want to the average of LST for June 2005 in some African countries. May I get some help? Thank you
