A New National Database for Solar Resource Assessment Commpiled from Ground-Based Observations
Abstract
A new solar radiation resource with complete coverage of the United States has been created from ground-based observations of global irradiance from 2005 to the present. Daily and hourly data have been collected from over 7000 professionally maintained weather stations. These sites belong to nearly 100 different federal, state and university networks. The average distance between sites in the continental US is about 37 km - somewhat farther in the great plains and closer in more populated areas. They have been combined into a “network of networks” so the accuracy of each individual site can be judged by comparing to nearby neighbors, often from other networks. Advanced filtering and sensor fusion techniques common in artificial intelligence are used to identify sensors that are not performing correctly and filter errors from the data stream. Validation tests have shown that the daily and hourly measurements of global irradiance in this ground-based resource have roughly half the uncertainty, half the observation error and half the bias error compared to the satellite based observvations in the National Solar Radiation Database (NSRDB-PSM). Accuracy, bias and uncertainty at the monthly and annual levels are comparable. This new solar resource can be combined with existing data, or used as a stand-alone database, to enhance site planning, production forecasts and on-going monitoring of solar projects.
Keywords
Solar radiation resource, Solar radiation database, Ground-based solar measurements, National solar radiation database, Nsrdb