Using a new local high resolution daily gridded dataset for Attica to statistically downscale climate projections

In this study we present a methodological framework to obtain statistically downscaled high resolution climate projections over the Attica region in Greece. The framework relies on the construction of a local daily gridded dataset for temperature variables (maximum, minimum and mean daily temperatures) and daily precipitation sums. To this aim, a mosaic of data that includes observations derived from ground stations and a high resolution simulation, performed by the Weather Research and Forecasting (WRF) model, for 1 year (1995) are blended using various gridding techniques to produce a 1 km 1 km high resolution daily gridded dataset for the period 1981-2000. The comparison of the gridded dataset against the observations reveals that the produced dataset maintains the observed long term statistical properties over the period 1981-2000 for both temperature and precipitation variables. Consequently, the produced dataset is used to statistically downscale Regional Climate Model simulations from the EURO-CORDEX initiative for the present (1981-2000) and the future climate (2081-2100) under the Representative Concentration Pathway (RCP) 8.5 climate scenario using two different bias adjustment techniques. The results indicate that the selection of the bias adjustment method is important and can affect the simulated climate change signals in a different way. Thus bias adjustment should be performed with caution and examined thoroughly before any such downscaled climate change projection dataset reach decision and policy makers in order to plan climate change related adaptation strategies.

Automated summary

This study presents a methodological framework for obtaining statistically downscaled high resolution climate projections over the Attica region in Greece. The framework involves the construction of a local daily gridded dataset for temperature variables and daily precipitation sums, which is then used to statistically downscale Regional Climate Model simulations. The results highlight the importance of bias adjustment methods in accurately simulating climate change signals.