Stochastic simulation of predictive space-time scenarios of wind speed using observations and physical model

In this work, we build stochastic scenarios of wind speed prediction. These scenarios are used in power grid applications to account for the uncertainty associated with renewable energies. The modeling aims to improve wind forecast using measurement data and deterministic model simulations and to quantify the Numerical Weather Prediction (NWP) forecasts errors in wind speed. A stochastic space-time process is calibrated on the augmented dataset of NWP outputs and measurements. Space-time interactions between the two datasets are accounted in a bivariate Gaussian framework. The process is written in a conditional way to avoid the specification cross-variances. First results indicate an improvement of the predictions from the proposed model in comparison with forecasts from NWP.