Continental-scale spatial optimisation of a solar irradiance monitoring network

• A solar monitoring network is designed for the Australian continent.
• A genetic algorithm is applied to gridded satellite-derived daily irradiance.
• Spatial coverage is enhanced substantially with the addition of 5–10 stations.
• Projected station density tends to be highest in areas of moderate variability.

Ground-based measurements of solar irradiance are the ‘gold standard’ for accurate insolation reference data, and enhance the precision of satellite-based estimates, yet the ground network in Australia remains sparse. The aim of this work is to determine the optimum locations for augmenting an existing network of ground stations, and to assess those locations in relationship with the spatial patterns in irradiance variability. Using gridded satellite-derived daily irradiance as a model, a genetic algorithm is used to guide the locations of up to 10 stations on the Australian mainland in addition to the existing 15 fixed locations. The optimisation criterion seeks to minimise the area where the mean spatial prediction error, based on linear regression, is larger than a fixed threshold. This threshold is chosen based on estimates of satellite accuracy with respect to ground stations. It is found that the coverage of the solar monitoring network over the Australian continent may be substantially enhanced with the addition of 5–10 stations. The projected station density tends to be highest in areas of moderate variability. The area to the east of the Great Dividing Range remains problematic, even for a large network. Even so, this method is effective at identifying the main gaps in the existing network, and can be used when planning a new network at a continental scale.