This page outlines the PV modelling methodology and specifications of Solcast's Rooftop PV Model, through which users can access forecasts and modelled actuals with global-coverage across Live (-7 days to present moment), and Forecast (present moment to +14 days) time periods.
The Solcast Rooftop PV Model, which produces the data parameter Rooftop PV Power Output (kW) is an empirical PV model, designed for modelling the production from rooftop PV systems (particularly fleets) with limited system specifications available. This model is limited to four system specifications: system azimuth and tilt, AC capacity, and a bulk loss factor. Because of this limited specification, there is no need for the creation of sites as with the Advanced PV model. Users simply make their API requests by referencing the latitude and longitude and capacity at a minimum, and also the azimuth tilt and loss factor if known. Some users cluster nearby systems (<5km) together, using known or assumed average values of these specifications.
The Solcast Rooftop PV Model is used by Solcast in the production of its behind-the-meter Grid Aggregations data product, which provides production estimated actual and forecast data for network assets or grid regions of any size from neighbourhood to continental scale. For more information on this product, refer to the Grid Aggregations Product Guide and Specifications document.
The Rooftop PV model uses Solcast's irradiance and weather data as its time-dynamic input. The Solcast data parameters used by the model include irradiance parameters Direct Normal Irradiance (DNI) and Diffuse Horizontal Irradiance (DHI), ambient Air Temperature, and a dynamic site-specific Snow Soiling Loss – Rooftop based on irradiance, temperature and precipitation. For more information about this input data, please refer to Solcast Irradiance and Weather Data Inputs and Algorithms documentation.
Model development and applications
The Solcast Rooftop PV model is designed to estimate production from a fleet of PV systems where system specifications, shading and other losses are poorly known at individual system level, and where system output measurement data is incomplete and/or not available in real time. The model can also be used to estimate system geometry and losses from measurement data. The original version of the model is described in Killinger et al., 2016 ("Projection of power generation between differently-oriented PV systems." Solar Energy 136: 153-165). During 2016 to 2019 the model was refined at the Australian National University in a $2.6M ARENA industry research project, and thereafter was licenced to Solcast. The model is now in operational use globally by a range of TSOs, Utilities and load forecasters in Australia, Taiwan, Korea, US, UK and Germany. Each deployment of the model has included local calibration and validation in concert with users.
The model is empirical, consisting of a fitted quadratic function of the plane-of-array irradiance and air temperature (and their interactions), and the loss factor. The quadratic coefficients and the loss factor are fitted to measured inverter output (AC) data from a range of PV systems targeted at performance for specific countries.
The model output is in terms of inverter output (AC) power. In cases where installed capacity data is in terms of module (DC) capacity, the model loss factors can be calibrated using inverter output (AC) power measurements. This approach does not invalidate the shape of model's power curves so long as very high inverter loading ratios (ILRs) (e.g. >1.25) are not so dominant amongst the fleet so as to overwhelm shading, soiling and other loss factors that drive the shape of the power curve.
The Solcast API Toolkit
Solcast takes on the many challenges of producing live and forecast solar data, so that you don’t have to. And that means making the data as easy to access, validate and integrate as possible, which is made possible through our API Toolkit. We provide instant access to live and forecast data products via this web interface, which is free to try. These include direct estimates of global, direct and diffuse solar radiation, as well as PV power output.