Methodology and uncertainty
The models used to generate the maps are based on 1974-1993 (CERES, Environment and Climate Change Canada) monthly mean daily global insolation data from 144 meteorological stations across Canada. Data from an additional 8 stations in Alaska (U.S. National Solar Radiation Database, 1961-1990) were also used to improve the models in that region. Insolation values were interpolated over the country in a regular grid (grid size: 60 arc seconds ~2 km grid) using thin-plate smoothing splines as operationalized in the ANUSPLIN model. In addition to monthly mean daily global insolation, the interpolation models are based on position (longitude, latitude) and precipitation (used as a surrogate for cloudiness).
Photovoltaic (PV) electricity generation potential for grid-connected photovoltaic systems without batteries was estimated from the insolation models for each grid cell using a performance ratio of 0.75. The performance ratio quantifies and takes into account overall system losses due to operation under non-ideal conditions: climatic factors, inverter operation and so on. The photovoltaic potential represents a first order approximation of the expected lifetime average system production for each month and for the entire year. It indicates the amount of electricity in kilowatt-hours produced per kilowatt of installed photovoltaic DC capacity rated at Standard Test Conditions (STC).
Two main metrics were used to quantify the uncertainty in the solar resource estimates presented here, namely the square root of the Generalised Cross-Validation statistic (RTGCV) and the root mean square model error (RTMSE). The standard deviation of the true error will typically lie somewhere in between RTMSE and RTGCV, with RTGCV being a more conservative estimate that includes error from noise in the underlying data. For annual means, RTGCV ranges between 0.5% and 5.3% of the mean depending on the PV array orientation, while RTMSE ranges between 0.2% and 2.4% of the mean.
Lifetime average annual PV potential values for most systems should be within about 10% of the values presented here. Well-functioning PV systems typically have annual performance ratios in the first years of operation in the range of roughly 0.75 to 0.9 and degrade at a rate of about 0.6% per year (see e.g. Jordan et al. 2016) over an expected lifetime of 25 years or more. Since our estimates represent lifetime averages, initial PV system performance should typically compare favorably to these. Regarding monthly values, these are only indicative and do not take into account monthly performance variations linked to changes in temperature, snowfall, interannual variability in the solar resource, etc.
For an overview of the methodology used and the model uncertainty, see
Pelland, S., McKenney, D. W., Poissant, Y., Morris, R., Lawrence, K., Campbell, K. and Papadopol, P. 2006. The Development of Photovoltaic Resource Maps for Canada, In Proceedings of the Annual Conference of the Solar Energy Society of Canada (SESCI) 2006.
A more detailed scientific article is also available:
McKenney D. W., Pelland S., Poissant Y., Morris R., Hutchinson M, Papadopol P., Lawrence K. and Campbell K., 2008. Spatial insolation models for photovoltaic energy in Canada, Solar Energy 82, pp. 1049–1061.
- Date modified: