For many years, the desorption data were used for calculation the distribution function using BJH theory. The results are more accurate and simply nicer. For mesoporous silica materials using desorption data seems to be enough. However, you should know, that these are different mechanisms and we get some different results. For comparision we always should use the same type of calculation. In my opinion the best in this case are desorption data.
The NLDFT are a great way to define the microporous distribution in the sample. Of course if they exist. Typically for mesoporous silica we should use model based on cylindrical geometry for oxides surface.