Using a reduced-order model to predict short-term future evolution of fair-weather cumuli

What is it about?

This work attempts to model some of the dynamics of fair-weather cumuli, so short-term forecasts can be made, not only of their future locations, but also their future shapes. We use a novel linear regression method that presents time evolution as a superposition of complex exponentials.

Featured Image

Photo by Dallas Reedy on Unsplash

Photo by Dallas Reedy on Unsplash

Why is it important?

Cloud shadows cause most of the intra-day variabilities in the outputs of solar photovoltaic generation facilities, and summertime fair-weather cumuli may routinely cause deep and repeated ramps over intervals of just a few minutes. Almost all short-term irradiance forecasting methods that address shadows of these shallow cumuli employ a form of "frozen-cloud advection" wherein a cloud is translated forward in time to a future position, estimating whether and to what degree it will block sunlight from the PV site. This work is important because it partially accounts for how these clouds change shape, grow, and dissipate.