Developing a Parameterization for Solar Variability in WRF-Solar Version 2
Abstract
Solar forecasting can help predict the future solar resource and make solar energy a more reliable and cost-effective part of utility grids. Numerical weather models are the most useful tool for day ahead solar forecasts and often for forecasts 1-6 hours in advance. However, standard numerical weather models are not optimized to give the information most needed by the solar energy community. A version of the Weather Research and Forecasting (WRF) model was released that has been better optimized for solar forecasting (WRF-Solar) including outputting higher temporal resolution solar irradiance, outputting diffuse and direct components of radiation, and improving surface irradiance forecasts for aspects such as aerosol impacts on clear sky irradiance. This presentation will discuss current work on further improvements to version 2 of WRF-Solar including adding a parameterization of solar variability under partially cloudy skies. Variability is one of the biggest challenges for including high penetrations of solar power in electrical grids, as utilities need reliable, consistent power. Coincident ground-based observations of cloud properties and solar irradiance have been used to develop statistical relationships of irradiance variability under different cloud cover conditions for defining a surface solar irradiance variability parameterization. This is tested at multiple locations in the continental US, using a machine learning algorithm to define cloud types consistently at multiple locations from ground-based observations. Using a variability parameterization is a computationally fast way to reproduce ramp rate statistics at sub-grid temporal and spatial scales.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2019
- Bibcode:
- 2019AGUFMGC52C..02R
- Keywords:
-
- 3322 Land/atmosphere interactions;
- ATMOSPHERIC PROCESSES;
- 0480 Remote sensing;
- BIOGEOSCIENCES;
- 1640 Remote sensing;
- GLOBAL CHANGE;
- 1952 Modeling;
- INFORMATICS