How to select years separately for calibration and validation?

Humaira,

The calibration and validation years for your model should be selected based on the years where you believe you have the most complete and reliable set of observed data. Of the years with reliable data, I would think that the more recent years would be the most useful, as they would better reflect the present system. You could also think about trying to capture a variety of hydrology years (wet, dry, etc), particularly focusing on the type of year you most want to explore in your model (perhaps drought for example).

In terms of how many years to pick for calibration and validation, I would suggest researching literature on this topic.

Good luck- hope this helps!
-Doug