I have an Ambient Weather station which measures irradiance too. What I have noticed over the years is that the station's sensor is aimed straight up, which is probably fine for getting a reading of average sun from the sky. However my panels are aimed S and tilted to better catch the sun. The results of this are that from late spring til Oct or so the weather station and PV output correlate fairly well (speaking of mid-day readings). But in winter, the panels produce far more than the irradiance values from the weather station would indicate they could/should. I suspect this is due to the PV panels being aimed more closely at the sun, while the weather station's sensor is aimed straight up.
And as far as cleaning goes, much easier to clean my ground mount PV panels than the weather station which is up a pole 11' off the ground.
BTW - I had suggested to Ambient Weather that they should integrate the irradiance numbers they graph to show WH's, as that would be quite useful for us solar folks. But after seeing how far off their sensor and numbers are from a properly aimed set of panels, not so sure that it would be all that useful after all.