I have been working with multiband images and on calculating standard deviation I am not sure what happens? Usually it's a single band operation so how is it dealt with in ARC GIS?
There are two separated ways in order to calculate standard deviation which have distinctive purposes. First is texture (window-based), well, I think its too obvious to talk about. The second is the standard deviation of a fixed-pixel in different bands. Well in that case, "Map algerbra" is very convenient in ARCGIS. For an instance, I was willing to determine in which places water vapor changes so drastically.Therefore I prepared a set of 200 images in different date, and by calculating the standard deviation of a pixel over all time, I realized in which places it changes. The results demonstrated in tropical regions water vapor changes dramatically, since it had high standard deviation. Do I have to explain it more?
I wanted to know what is the equation followed to calculate standard deviation for a multiband image in ArcGIS? i.e. i want to know the algorithm that happens to calculate standard deviation in a multiband image using this procedure of ArcGIS
Toolbox > Spatial Analyst > Focal Statistics > standard deviation
Well, your question was somewhat nebulous. In this case, you just perform some operations for a pixel within a specified neighborhood around it. For an instance, you define a 3X3 kernel (rectangle kernel) and you opt "SUM" operation. The output of this pixel will be the sum of whole values in 8 pixels around and itself. Therefore, when you choose standard deviation, you will calculate the standard deviation of all these values. As I told you before, the standard deviation calculated within a kernel is regraded as a texture in multi-band sensors. There are many applications for this output: 1-cloud masking, 2-classification ,3-population estimation and so on. Please see the attachment too.