Clustering high dimensional meteorological scenarios: Results and performance index

Abstract

The Réseau de Transport d'Electricité (RTE) is the French main electricity network operational manager and dedicates large number of resources and efforts towards understanding climate time series data for the purpose of energy optimization. A key challenge at the core of understanding the climate time series data is being able to detect common patterns between temperatures time series, and to choose representative scenarios for simulations, which in turn can be used for energy optimization. We addressed this challenge using climate time series provided by RTE, which is comprised of 200 different possible scenarios on a grid of geographical locations in France. We first show that the choice of the distance used for the clustering has a strong impact on the meaning of the results. Depending on the type of distance used, either spatial or temporal patterns prevail. Later we discuss the difficulty of fine-tuning distances with a dimension reduction procedure and we propose a methodology based on a carefully designed index.