GC43A-0687:
Developing and testing solar irradiance forecasting techniques in the Hawaiian Islands region

Thursday, 18 December 2014
Dax Kristopher Matthews1, Joao Marcos Souza2 and Karl Stein1, (1)University of Hawaii at Manoa, Hawaii Natural Energy Institute, Honolulu, HI, United States, (2)University of Hawaii at Manoa, Oceanography, Honolulu, HI, United States
Abstract:
Irradiance variability, primarily driven by cloud formation and advection, can be problematic in the state of Hawaiʻi, because of the high penetration of distributed solar and the small scale of the island electrical grids. The Hawaiʻi Natural Energy Institute (HNEI) is developing an operational system in order to research and test new techniques to generate solar forecasts for the Hawaiian Islands. The operational system comprises the following three components.

(i) A ground-observation-based advection model, using sky imagers and a ceilometer located at the University of Hawaiʻi at Mānoa. Every 10 minutes (during daylight hours), this component generates a high-resolution 1 hour Global Horizontal Irradiance (GHI) prediction for a region that is within ~15 km of the instrumentation.

(ii) A satellite-image-based advection model, using Geostationary Operational Environmental Satellite (GOES) imagery and the Heliosat-II method. Every 30 minutes (during daylight hours), this component generates a 1 km resolution, 6 hour GHI prediction for the entire Hawaiian Archipelago.

(iii) A coupled ocean-atmosphere model, using the Regional Ocean Modeling System (ROMS) model and the Weather Research and Forecasting (WRF) model, including newly available microphysics, shallow convection parameterization, and radiative transfer model options. Nightly, this component generates 48 hour GHI, Direct Normal Irradiance (DNI), and Diffuse Horizontal Irradiance (DHI) predictions for (a) a 10 km resolution domain covering the full Hawaiian Archipelago and (b) a nested 2 km resolution domain covering the islands of Maui, Oʻahu, and Hawaiʻi.

We discuss the development and validation of the system, and the scales of forecasting accuracy for each component. We also examine the impact of the coupled model on the simulations of surface flux processeses and ocean-atmosphere feedbacks, both of which influence the prediction of regional cloud properties.