Optimal interpolation of satellite and ground data for irradiance nowcasting at city scales

Abstract

We use a Bayesian method, optimal interpolation, to improve satellite derived irradiance estimates at cityscales using ground sensor data. Optimal interpolation requires error covariances in the satellite estimates and ground data, which define how information from the sensor locations is distributed across a large area. We describe three methods to choose such covariances, including a covariance parameterization that depends on the relative cloudiness between locations. Results are computed with ground data from 22 sensors over a 75 × 80 km area centered on Tucson, AZ, using two satellite derived irradiance models. The improvements in standard error metrics for both satellite models indicate that our approach is applicable to additional satellite derived irradiance models. We also show that optimal interpolation can nearly eliminate mean bias error and improve the root mean squared error by 50%.

Broad-Time-Horizon Solar Power Prediction and PV Performance Degradation Research at the University of Arizona

“…In past work, UA's PFG has led the development of technology and methods to improve satellite-derived irradiance forecasts using a sparse set of readily deployed sensor elements and distributed PV installations. Using a Bayesian technique (optimal interpolation (OI)), significant improvement in GHI nowcasts can be achieved under a variety of weather conditions [4,7]. As shown in Figure 6, the use of sparse ground data and OI methodology (associated with "sensor" legend items in Figure 6) resulted in a significant improvement in GHI prediction compared with a prediction that did not involve OI (labelled "Background" in the Figure).…”

“…The UA Power Forecasting Group (UA-PFG) generates forecasts of weather, solar power, and wind power on time horizons from minutes to 7 days. A combination of real-time solar power data from utility-scale plants and rooftop PV, geostationary satellite imagery, and a 1.8 km resolution configuration of the Weather Research and Forecasting model is used to achieve better forecast accuracy at different time horizons [2][3][4].…”

“…Recent efforts have focused on the development of a new data assimilation methodology using a distributed sensor and PV generation network to increase local irradiance forecasting accuracy [4].…”

An unsupervised method for estimating the global horizontal irradiance from photovoltaic power measurements

“…The active control of distribution networks, of which some of the critical sections can take up a small area, requires a more accurate and fast estimate of GHI. Satellite-based methods could also profit from an increased availability of on-ground GHI measurements, as they could be used for calibration [14,15,16], a technique also known as site adaptation. In this study, we investigate the possibility of using local PV power measurements to estimate GHI with a high temporal and spatial accuracy.…”

Comparison of Data-Driven Techniques for Nowcasting Applied to an Industrial-Scale Photovoltaic Plant

“…Although these phenomena could be modeled from a physical point of view, their complex and turbulent behavior is a challenge that can be tackled effectively with data-driven techniques. Some approaches make use of imaging techniques, either from ground-based [6] or satellite [7] cameras, while others only make use of endogenous data [8] (i.e., the power output is used as the only input feature to the forecasting). The advantages of a data-driven approach for solar power forecasting lies in its ability to learn, from historical data, the complex patterns that are difficult to model (for instance shading phenomena).…”