Spatially distributed Added Value Index and Climate Change Downscaling Signal for convection permitting scale simulations

Abstract

<p>A new metric that quantifies Added Value (AV) was developed that compares the difference within the entire probability distribution functions (PDFs) of the Regional Climate Model (RCM) and its driving General Circulation Model (GCM) with a high-resolution observation source, at every grid point, to obtain a spatial distribution of AV. This is important to assess the validity of the computationally expensive process of downscaling, especially for Convection Permitting Models (CPMs). The method can be adapted to focus on the tail-end of the distribution, since GCMs struggle to resolve precipitation extremes. To achieve this, the threshold value of the percentile of interest (for example, the 95th percentile) is obtained from the observation source and then applied to the PDF data as a filter, after which the corresponding AV can be obtained. This metric can also be adapted to assess the Climate Change Downscaling Signal (CCDS) of climate projections, by comparing to the corresponding historical data-set instead of an observation source.</p><p>This method is now being adapted to CPM simulations using a multi-model approach. The analysis is focused on both daily and hourly data from a 14-model ensemble of the ALP-3 domain using 5 high-resolution observation sources (GRIPHO for Italy; EURO4M for large alpine area; COMEPHORE for France; RADKLIM for Germany; and RdisaggH for Switzerland). The primary objective is to assess the added value of the CPM with the driving RCM, but a comparison to the GCM is also included. Preliminary results show that the CPM runs add value over the RCM, with possible emphasis in models/regions of lower RCM AV (requires confirmation by comparing RCMs to the driving GCMs). The analysis is will also focus on the CCDS metric of the near- and far-future simulations of the CPM, and the historical analysis is being replicated using hourly precipitation instead of daily.</p>