The accuracy of solar irradiance calculations used in mesoscale numerical weather prediction

In this paper, solar irradiance forecasts made by mesoscale numerical weather prediction models are compared with observations taken during three air-quality experiments in various parts of the United States. The authors evaluated the fifth-generation Pennsylvania State University-National Center for Atmospheric Research (PSU-NCAR) Mesoscale Model (MM5) and the National Centers for Environmental Prediction (NCEP) Eta Model. The observations were taken during the 2000 Texas Air Quality Experiment (TexAQS), the 2000 Central California Ozone Study (CCOS), and the New England Air Quality Study (NEAQS) 2002. The accuracy of the model forecast irradiances show a strong dependence on the aerosol optical depth. Model errors on the order of 100 W m(-2) are possible when the aerosol optical depth exceeds 0.1. For smaller aerosol optical depths, the climatological attenuation used in the models yields solar irradiance estimates that are in good agreement with the observations.