1408 Solar Power Forecasts Evaluated with Proprietary Site Generation Data for Various Tracking Geometries, Site History Periods, and under a Variety of Weather and Climate Conditions

Wednesday, 25 January 2017
Stephen D. Jascourt, MacDonald, Dettwiler and Associates (MDA) Information Systems LLC, Gaithersburg, MD; and D. Kirk-Davidoff, C. Cassidy, and T. Hartman

Handout (1.5 MB)

As we advance the MDA Solar Power Forecasting System, each year at this annual meeting we present new lessons learned. At this point our forecast system has gained enough experience with a wide variety of proprietary metered generation data that skill at various forecast lead times overall and in different parts of the diurnal cycle can be compared for different weather conditions, different geographic areas, and different installation types including dual-axis tracking and single-axis tracking with east-west and north-south axes as well as fixed tilt. The geographic areas include mid-latitudes and tropics, dusty and clear environments, moist, stormy, and dry climates, and moist and dry seasons in the same location. Also, skill will be presented for new sites coming online over time as available history goes from nothing to months and compared to skill for nearby longer-established sites. This may be a key topic of interest as continuing exponential growth of installed power means that a large fraction of the total solar power on the grid is always coming from recently commissioned sites. Additionally, methods of incorporating real-time satellite data and proprietary real-time site generation data will be evaluated for their improvement in short-range power forecasts over NWP-based forecasts utilizing a multi-model ensemble which includes the High-Resolution Rapid Refresh model.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner