Correlated spatiotemporal downscaling of Euro-CORDEX climatic data for infrastructure resilience assessment

Abstract

A methodology is presented for downscaling the Euro-CORDEX climatic projections in order to derive spatially and temporally correlated weather fields that can be used for risk and resilience assessment of large-scale asset portfolios or interconnected infrastructure. The temporal resolution of the Euro-CORDEX data is downscaled to a 10 min basis by employing a modified analogue-type approach that utilizes the k-NN algorithm along with measurements from weather stations. The aim is to generate composite “Frankenstein” days comprising 144 jigsaw pieces of observed 10 min timeseries that are scaled and/or shifted, and matched together to form a continuous daily record. These point-estimates, valid only at the locations of the weather stations, are expanded spatially by employing high-fidelity weather intensity measure fields that provide variable yet synchronous patterns of weather parameters at all locations of interest. As a case study, the Euro-CORDEX projections for wind, temperature, and precipitation are downscaled for the Metsovo-Panagia segment of Egnatia Odos highway in Greece, by employing high-fidelity Computational Fluid Dynamic simulations that account for the topography of the site to simulate turbulent wind flows. These are combined with measurements of two local weather stations to generate the Frankenstein timeseries and corresponding weather fields that can be used for estimating operability, recovery and direct/indirect loss statistics on an event-by-event basis for an ensemble of interconnected highway assets.