Solar power forecasting: measurements and numerical weather predictions

This dataset contains hourly power measurements from a 16.32 kW peak PV power plant located on the North of Portugal (Smart Grid and Electric Vehicle Lab – SGEVL, at INESC TEC), divided in the following strings: 3 × 2160 W, 2 × 3360 W and 1 × 3120 W. The available historical period spans from April 28th, 2013 to June 28th, 2016. The missing values in each string were interpolated using the available measured data from the other strings, thus the number of missing values was very small (only 3.82%). The numerical weather predictions (NWP) data was retrieved from MeteoGalicia THREDDS server, which is a publicly available service that provides historical and daily forecasts of several weather variables from the mesoscale Weather Research and Forecasting (WRF) model. The spatial resolution is 4km and the temporal resolution is one hour, each run is initialized at 00h UTC and the time horizon is 96 hours-ahead. The geographical grid is made up by 169 equally distributed points in a 13 x 13 spatial configuration, covering an approximated area of 2400 km2. The raw NWP dataset consisted in 1014 variables for the PV site, namely:

• swflx [W/m2] - surface downwelling shortwave flux

• temp [K] - ambient temperature at 2 meters

• cfl [0,1] - cloud cover at low levels

• cfm [0,1] - cloud cover at mid levels

• cfh [0,1] - cloud cover at high levels

• cft [0,1] - cloud cover at low and mid levels

This data set can used for solar power forecasting with a grid geographically distributed weather forecasts and follows the work described in: J.R. Andrade, R.J. Bessa, “Improving renewable energy forecasting with a grid of numerical weather predictions,” IEEE Transactions on Sustainable Energy, vol. 8, no. 4, pp. 1571-1580, Oct. 2017