A few days ago I commented about the sources of error in the direct/diffuse splitting method. In this case, other sources of errors exist: possible partial shading of the instrument due to obstructions, and ground albedo.
In addition, what is the source of your weather file? If it is TMY-type data, you cannot expect good correspondence of the modeled and measured data, since they are not for the same period.
We could comment better if you give more details: magnitude of the differences, location, source of data, ground albedo, differences between clear-sky and cloudy0sky conditions, etc.
--
Chris A. Gueymard, PhD
Solar Consulting Services
P.O. Box 392
Colebrook, NH 03576, USA
Florida's Office:
Tel. (386) 402-8949
www.SolarConsultingServices.com
I am interested in this question. How are you measuring the solar radiation?