The cost of day-ahead solar forecasting errors in the United States

As solar energy contributes an increasing share of total electricity generation, forecasting errors become important relative to overall load uncertainty and can add costs systems. We investigated the day-ahead forecast across 667 existing power plants in United States (years 2012 through 2019). Our analysis was based on hourly real-time nodal prices. analyzed two types forecasts: persistence forecasts, a simple approach forecasting, numerical weather prediction forecast, North American Mesoscale Model (NAM), improvement over forecasts public data modelling software. modeled using meteorological plant specific characteristics. Hourly generation debiased with multiple sources records. NAM had relatively low average, at no more than $1/MWh all years except 2016, when rose $1.5/MWh. Even after these error costs, value marginally higher simulating participation markets versus only markets. On premium for participating market, ranged from ?0.5 5.2 $/MWh years. Average were regions penetration (i.e., California New England) compared penetration. However, England similar despite California, indicating that date have been loosely correlated levels.