Dear you Dear you can visit RG ,so I thank written for this subject and meet you

You can address this issue by creating custom filter(s), depending on your needs.

Think about it like this:

1- You have a first raster that has NaN values, but ENVI counts them as zeros.

2- You can create a 2 class raster from the original file where all values you want to keep will be assigned the label > and all the NaN values will be assigned a large positive or negative value (No-Data values can be assigned -9999 or 9999 for example, values can be much larger to serve your needs better).

3- By doing this you have one extra step after the change detection whereby you multiply the results by the layer created in -2-

4- Finally, you may want to clean the resulting output in 3 further. You will have the change detection result but also some pixels with very large values resulting from the multiplication in step -3-

You can create a new filter by classifying the change detection result raster into class (1) that includes all legitimate (or correct) values, and class (NaN) that basically groups all the pixels with large values, which correspond to the original NaN pixels.

Keep in mind that the values you assign in Step -2- might be changed to adapt to your needs. Also, I believe ArcMap 10+ does not remove NaN values and change them to >. I have used the same concept, i.e., creating intermediate filters by classification, to remove unwanted values (NaN or other).

Hope this helps.

Atef

Hello Atef,

This was very useful, thank you!

Cheers