Comparison of Measured and Modeled Snow Losses for Photovoltaic Systems in Colorado

Abstract

We quantify measured and modeled snow losses at four utility-scale single-axis tracking photovoltaic (PV) power plants in Colorado. Across 50 site-months of data collected over three winters, the Marion and Townsend snow loss models exhibited similar absolute bias errors, although the Townsend model had lower monthly root mean square error. Based on our results, we recommend that, for PV systems in Colorado and similar climates, the Townsend and Marion model predictions be averaged together to generate the most accurate snow loss predictions for monofacial tracking PV facilities, and that solely the Townsend model be used for bifacial tracking PV facilities.