Thanks Mr.Joos. I used the NASA SSE meteo data for various cities in India and obtained the energy production values while simulating for a thin film PV module. PVsyst uses a CREST spectral correction model as a function of air mass and clearness. Comparing the production vs altitude in a linear regression, statistically significant data was only recorded for monsoon season. Is this a correct hypothesis?
The spectral correction is not all, and does not depend on just elevation. Sites at different elevations might have different climates (more or less cloudiness). The cloudless atmosphere becomes normally clearer (less aerosols and water vapor) as elevation increases. These two constituents directly impact the spectrum. Furthermore, higher elevation does not necessarily mean less cloudiness: one side of a mountain can be very cloudy whereas the other side is clear above some threshold elevation, depending on wind patterns, etc.
Finally, temperature tends to decrease with elevation, thus improving PV efficiency.
The NASA-SSE database may not be sufficient in this case because of its coarse spatial resolution and relatively poor accuracy.