Can added value be expected in RCM-simulated large scales?

Abstract

Nested Limited-Area Models require driving data to define their lateral boundary conditions (LBC). The optimal choice of domain size and the repercussions of LBC errors on Regional Climate Model (RCM) simulations are important issues in dynamical downscaling work. The main objective of this paper is to investigate the effect of domain size, particularly on the larger scales, and to question whether an RCM, when run over very large domains, can actually improve the large scales compared to those of the driving data. This study is performed with a detailed atmospheric model in its global and regional configurations, using the ''Imperfect Big-Brother'' (IBB) protocol. The ERA-Interim reanalyses and five global simulations are used to drive RCM simulations for five winter seasons, on four domain sizes centred over the North American continent. Three variables are investigated: precipitation, specific humidity and zonal wind component. The results following the IBB protocol show that, when an RCM is driven by perfect LBC, its skill at reproducing the large scales decreases with increasing the domain of integration, but the errors remain small even for very large domains. On the other hand, when driven by LBC that contain errors, RCMs can bring some reduction of errors in large scales when very large domains are used. The improvement is found especially in the amplitude of patterns of both the stationary and the intra-seasonal transient components. When large errors are present in the LBC, however, these are only partly corrected by the RCM.Although results showed that an RCM can have some skill at improving imperfect large scales supplied as driving LBC, the main added value of an RCM is provided by its small scales and its skill to simulate extreme events, particularly for precipitation. Under the IBB protocol all RCM simulations were fairly skilful at reproducing small scales statistics, although the skill decreased with increasing LBC errors. Coarse-resolution model simulations have difficulties in simulating heavy precipitation events, and as a result their precipitation distributions are systematically shifted toward smaller intensity. Under the IBB protocol, all RCM simulations have distributions very similar to the reference field, being little affected by LBC errors, and no significant differences were found between the small scales statistics and the precipitation distributions obtained over different RCM domains.

A review on regional dynamical downscaling in intraseasonal to seasonal simulation/prediction and major factors that affect downscaling ability

“…No data are available to assess the downscaling ability for other atmospheric fields when NARR is used as LBC since the NARR has been considered as the best "real" data with the highest resolution. In a "Big Brother" type of North American regional winter season study focusing on precipitation, specific humidity, and zonal wind (Diaconescu and Laprise, 2013), it is found that if an RCM is driven by a relatively high resolution GCM (Big Brother) with small errors, no improvement is found at the large scales simulated by the RCM. The added value will be solely in the RCM-simulated small scales that are not present in the driving GCM fields.…”

Evaluation of daily precipitation statistics and monsoon onset/retreat over western Sahel in multiple data sets

“…Because the Y-axis presents the total precipitation in a precipitation intensity bin (instead of the more classical representation of percentage of precipitation per bins and their relative contributions to the total precipitation amount per season; e.g. Diaconescu and Laprise 2013;Martynov et al 2013;Šeparović et al 2013), the comparison of diagrams indicates also the bias magnitude or similarity for each precipitation intensity. Note that these distributions should actually be shown as histograms; however, in order to increase the clarity of the comparison between observed and reanalysis/RCMs products, we chose to present them as curves.…”

Climate change projections for CORDEX-Africa with COSMO-CLM regional climate model and differences with the driving global climate models

“…Hong and Kanamitsu 2014); as a result, downscaling is not always able to improve the simulation skills of large-scale GCMs although added value in downscaling GCMs is found especially in the fine scales and in the ability of RCM to simulate extreme events (e.g. Kim et al 2002;Diallo et al 2012;Paeth and Mannig 2012;Diaconescu and Laprise 2013;Crétat et al 2013;Haensler et al 2013;Laprise et al 2013;Lee and Hong 2013;Buontempo et al 2014;Lee et al 2014;Giorgi et al 2014;Dosio et al 2015) In this work we present the results of the application of the COSMO-CLM RCM (CCLM) in the production of climate change projections for the CORDEX-Africa domain. This work builds on two previous studies: Panitz et al (2014) investigated the structural bias of CCLM driven by ERA-Interim (evaluation run), whereas Dosio et al (2015) analyzed the added value of downscaling lowresolution GCMs over the present climate (historical runs).…”