Ensemble Sensitivity Analysis of Wind Ramps

Wind ramps, generally defined as a sharp increase or decrease in wind speed over a short period of time, have received recent attention due to their application to the wind energy industry. Despite this focus, forecasting wind ramps remains a challenge. This study uses ensemble sensitivity analysis (ESA) to identify atmospheric features that are relevant to wind ramp forecasting. Previous uses of ensemble sensitivity have related state variables at a single forecast time (called the response function) to variables in the initial state or at earlier forecast times. This work uses response functions that are not fixed in time, but are based on characteristics of the event such as the ramp slope, maximum wind speed and timing of the ramp in order to diagnose the relevant features of the flow, and whether these features vary for different ramp characteristics. Furthermore, such sensitivity can be used with ensemble observation targeting techniques to learn where additional observations should be taken to most improve the forecast. The results of many cases can be used to identify recurring locations of sensitivity and this could be used to design observational networks to increase the predictability of wind ramps at specific wind farms.