Mohammed, I am not familiar with NCDF, but R ignores NA-s when computing mean if you use the option na.rm=TRUE:

tmp = c(1, 2, 3, NA, 5)

mean(tmp, na.rm=TRUE)

And if you replace NA-s with NULL, the most functions will just ignore them.

More here:

http://www.statmethods.net/input/missingdata.html

Tanks you Pavel, I use this and it's work good ..

for (i in 1:tr$n) {

v

I am assuming you are working with geographical data? If so, have a look at the raster package. It is the de facto standard for working with geographical fields and will likely be useful in the rest of your workflow too.

Cheers

Yes, it is image data, and for my example I used the GDAL and Raster package.

Then it should be fairly easy to produce a RasterStack from your ncdf fields on which you can use raster::calc() to produce a RasterLayer of pixel averages. See help(calc)