This is information you should be able to get from the data provider.  For example, for ASTER thermal data, you can go to the https://lpdaac.usgs.gov/ where they describe the different products.  It tells you what units each product is in.  Once you know the original units, you can use the appropriate scale factor to input to get to the expected units of W/(m^2*sr*micrometer) for the emissivity normalization tool.

Thank you mminari,  -The MASTER data have been processed in ATCOR and I believe the thermal channels are in mW/m2/sr/um so need to be divided by 0.001?  - If this is correct my temperature values are very large, should I divide them by 10000?   thanks  

Those look like the expected units, so I would think your scale factor would be 1.0.

I thought ENVI Emissivity normalization method required a data scale factor of  W/m2/sr/um and the ATCOR4 output is mW/m2/sr/um? I may be confused. Please let me know if I should use 0.001 or 1.0 Thanks

Sorry, I thought you wrote W/m^2 but you wrote um/m^2.  I don't know what units ATCOR uses but um/m^2 would be unusual.  Units are typically either W/m^2 or um/cm^2.  You should confirm this in the ATCOR software.  If it is really W/m^2, no scale factor should be needed so you would use 1.0.

sorry I wrote mW/m^2 for the ATCOR output, is that still unusual?

Hi Mminari, sorry to keep asking but I wanted to clarify that if I am using mW/m^2 do I need a scale factor of 0.001 in the emissivity normalization method to get it to W/m^2? Thanks  

As I said before, that would be an unusual unit.  According to this website: http://www.rese.ch/support/atcor4_faq.html The output units from ATCOR are mW/cm^2 not m^2 which would make the scale factor 0.1.

But in the same document: "For spectral bands in the thermal region (8-14 micron) the radiance unit [mW m-2 sr-1 micron-1]" so should I still use 0.1?

Sorry, you are correct.  Those are the units that atcor states it returns for thermal data.  In any case, you should use the scale factor (which is multiplied actually) that converts the data to W/m^2*sr*micron.  Since you tried 0.001 before and the results were not reasonable, it may be that your radiance cube is not in these units or perhaps were not converted correctly.  ASTER data, which are already in the correct units, return data in K in the expected range (290-320) using this tool.

Thank you. The scale factor of 0.001 makes the emissivity results look noisier, but the temperature results are now correct (in kelvin). Thanks for your help