Have to convert raster... but need to build attribute table!

[LauraCockrell]

Hello folks-
I am using the raster calculator to build the NDVI expression (using Arc10.0) because I want the raw values of the NDVI calculation. However, in order to use the raster calculator and get the raw values, I have to change my original landsat input into float images (I used the Math toolbox>Float tool to convert). The resulting output from the raster calculation of NDVI is a 32 bit float image. The NDVI looks great, I can use classified symbology to break up the color ramp and I really like the display, but I want to work with the pixel values to calculate the mean pixel value for the habitat that I have selected. Obviously with the NDVI being a 32bit float I can't build a raster attribute table.
Is there a good way around this?
Ultimately, I want to clip out the areas I sampled, then calculate the mean NDVI across each area. Then I want to remove all the negative NDVI values and calculate the mean of the remaining positive NDVI values and see if either of those NDVI's can be correlated with site use.
Thank you for your help!
~Laura Cockrell

[JeffreySwain]

The usual step to maintain data would be to multiple the float values by a constant like 1,000 or more depending on how many decimals you want to maintain. Then once you are done processing then you can divide to get back down to the float values. From your description you may not need to see the values. I am curious about your workflow for data. Is your original data single band integer data?  If so how are you performing the NDVI on less than four band imagery?  Which NDVI calculation are you recreating, like this one to get values between  -1.0 and 1.0?  If that is the case, short of trying to convert the data to a vector format and then spatially joining them, I am not sure if what you are looking to achieve is possible for each area. If you had polygons originally and then wanted to add the data based on the NDVI raster you create then you could do that with the Zonal Statistics or if you had point you could harvest the NDVI values for the individual points via the Add Surface information or Extract Values to Point.