Mathematical models have gotten pretty good at predicting daily and monthly GHI based on satellite images and temperature or humidity measurements. Estimating hourly values seems a lot harder, with most models having a 25% error margin.
The Canadian Weather Energy Engineering Dataset (CWEEDS), published by the federal government's department for Environment and Climate Change Canada, provides hourly GHI values for dozens of locations in each province. But when you read the release notes, you find that most hourly values are derived from a model and have not been directly observed:
Solar global horizontal irradiance (field 102) observations are available for some of the period of record for 35 of the 143 CWEEDS locations. The MAC3 model was used to estimate solar global horizontal irradiance for the 108 locations and times for which observations were unavailable. It is important to note that the modelling errors can be substantial for any given hour. The root-mean-square error for hourly solar global horizontal irradiance amounts are typically around 30% (Morris and Skinner, 1990).
X Value was observed (that is, not derived with a model and not altered)
E Value was missing and has been replaced by a hand estimate.
I Value was missing and has been replaced with one derived by
interpolation from neighboring observations.
M Value was missing and has been replaced with one derived
with a model (model used depends on element).
Q Value is derived from other values (e.g. illuminance data which are not observed).
9 Value is missing; data positions contain 9s as well
|Hourly sample origin percentages for Alberta CWEEDS data|
Only Stony Plain and the Edmonton municipal airport had actual measurements of Global Horizontal Irradiance. At the municipal airport, 17.5% of all observations between 1953 and 2004 were based on measured values, whereas Stony Plain measured 81.0% of all hourly GHI values.
If you need hourly GHI values to validate a photo voltaic model, it is probably wiser to use the data from the Stony Plain location, rather than a location nearer you.
Environment Canada published a Solar radiation data and model review in 2010, recommending an update of the CWEEDS data for the benefit of the solar industry. The automation of most observation stations caused cloud observation data to be incompatible with the models for estimating GHI, so more measurements (or re-staffing observation stations) will be required.
Le me know if you are interested in the data for your province.