The weather roulette: seasonal wind speed predictions vs. climatology

Introduction

Predicting the future variability of the wind resource over the coming weeks or seasons can be relevant to quantify operation and maintenance logistic costs or to inform energy trading decision with potential cost savings and/or economic benefits. Applied research on climate services has already shown that probabilistic climate predictions can improve current prediction practices, which are based on the use of climatology and the assumption that what happened in the past is the best estimation of future conditions.

Approach

To demonstrate the value of climate predictions and how they can outperform climatology in areas where the model is skilful we have used the weather roulette framework of Hagedorn and Smith (2009). In this “roulette game” the player places a bet using a forecast system, and tries to beat the opponent’s forecast (seasonal predictions vs. climatology in this case). The actual wind speed observed and the hit rate of the forecasts translate the statistical concept of forecast system skill into those of “interest rate” and “return of investment” concepts, which are commonplace in the corporate world.

Main body of abstract

This framework sets a diagnostic tool that informs in a playful and more intuitive way about the skill and usefulness of a forecast in the decision making process, by providing an economic and financial oriented assessment of the benefits of using a particular forecast system instead of climatological information. We present the results in a mobile app, allowing stakeholders to play the weather roulette game for individual wind farms or selected portfolios of wind farms and visualise the overall prediction success of seasonal wind speed predictions compared to climatology.

Conclusion

This app aims to bridge the gap between state-of-the-art research in climate predictions and the wind energy sector, and seeks the feedback and engagement of wind energy users with climate services research community.


Learning objectives
This demonstration exercise will illustrate the value of probabilistic predictions in economic terms to improve user's understanding the benefits of seasonal climate predictions.