High-Resolution Hybrid Downscaling of CMIP6 Climate Projections Using WRF and AI-Based Models: A Case Study of the Urban Thermal Environment of Nicosia, Cyprus

Global Climate Models (GCMs) provide essential data for assessing potential changes in the climate system under different Shared Socio-economic Pathways (SSPs). Nevertheless, their coarse resolution limits their direct application in climate impact studies. This limitation is particularly critical in regions highly affected by climate change, such as the Eastern Mediterranean and Middle East (EMME) region, a recognized climate change hotspot. To address this, downscaling techniques are employed, including dynamical downscaling (DD), empirical/statistical downscaling (ESD), and hybrid downscaling (HD), which combines both approaches. This study develops an HD method to downscale daily maximum and minimum air temperatures at the local scale. Specifically, the perfect prognosis (PP) framework of the ESD is utilized; for this reason the Weather Research and Forecasting (WRF) model is driven with ERA5 reanalysis data to dynamically downscale predictors to the local scale. This method creates a high-resolution 2D database of the predictand variables, which is necessary to extract the statictical relationships between the regional and the local scale variables. Subsequently, two ESD approaches are implemented: a classical Multiple Linear Regression (MLR) model and an artificial intelligence (AI)-based model using Artificial Neural Networks (ANNs). The methodology is applied at Nicosia, Cyprus, while the performance of these models is evaluated against in-situ measurements from two meteorological stations located in the study area. Finally, the derived statistical relationships are applied to a historical (2008-2012) and a future (2048-2052) period under the SSP2-4.5 scenario of the MPI-ESM1-2-HR model to produce projections of Nicosia's urban thermal environment in fine detail. Results demonstrate  a considerable improvement in the achieved spatial resolution of climate parameters, a fact that supports the detailed development of climate impact studies.