For northern communities relying on solar photovoltaic (PV) systems to offset the cost of imported diesel fuel, there is an acute need to forecast energy losses due to snow. Two leading snow loss models, developed by Marion et al. and Townsend and Powers, are used widely in the PV industry. In order to test these models, this study compares modeled monthly and annual losses to a best-estimate reference based on PV power output and camera images from an array operating in Quebec, Canada. This study addresses two key areas: the development of an image-analysis algorithm to detect snow on a PV array and the validation of models estimating energy loss. For the PV array in this work, Townsend and Powers is a closer approximation of real annual energy losses than the Marion et al. model. However, to provide communities and PV system developers with accurate snow loss predictions, more work needs to be done to improve these models using arrays at different latitudes. The image analysis algorithm used in this work is 90% accurate in identifying snow. This leads to a difference of about 0.2% points in annual snow loss estimates when using the image analysis algorithm instead of visually inspected photos.

  • All data and free and open-source software used in the study can be acquired from under the GNU General Public License (GPL) 3.0.

See also[edit | edit source]

Discussion[View | Edit]

Cookies help us deliver our services. By using our services, you agree to our use of cookies.