Abstract
Spatial interpolation errors at 2816 grid points covering Canada
are determined using 10- year averages of the solar irradiances for
130 stations unevenly distributed throughout the same area. Three
specific interpolation schemes are evaluated (2 and 3 station inverse
distance weighted interpolation and single station extrapolation)
though the data set and algorithm (based on kriging) are sufficiently
general that other strategies could also be evaluated. In southern
Canada relative values of the extrapolation error for ten-year means
are typically between 1 and 2% and decrease to less than 1% if values
are interpolated using data from three stations. Errors are higher
at coastal locations and in the Canadian Arctic due to generally
greater interpolation distances and increased spatial variability.
Lower actual irradiances also influences the relative errors. There
is little advantage to including additional data from stations in
adjacent areas of the U.S. since interpolation and extrapolation
errors are generally already less than the errors associated with
measurement and modelling procedures and with temporal sampling (i.e.,
the long-term representativeness of a 10-year mean).
Users
Please
log in to take part in the discussion (add own reviews or comments).