Applying an isotope-enabled regional climate model over the Greenland ice sheet: effect of spatial resolution on model bias

• Main Authors: Breil, Marcus, Christner, Emanuel, Cauquoin, Alexandre, Werner, Martin, Karremann, Melanie, Schädler, Gerd
• Format: Article in Journal/Newspaper
• Language: English
• Published: Copernicus Publications 2021
• Subjects: ddc:550; Earth sciences; info:eu-repo/classification/ddc/550; Arctic; Greenland; Greenland ice cores; Ice Sheet
• Online Access: https://publikationen.bibliothek.kit.edu/1000139427; https://publikationen.bibliothek.kit.edu/1000139427/130633862; https://doi.org/10.5445/IR/1000139427

In order to investigate the impact of spatial resolution on the discrepancy between simulated δ$^{18}$O and observed δ$^{18}$O in Greenland ice cores, regional climate simulations are performed with the isotope-enabled regional climate model (RCM) COSMO_iso. For this purpose, isotope-enabled general circulation model (GCM) simulations with the ECHAM5-wiso general circulation model (GCM) under present-day conditions and the MPI-ESM-wiso GCM under mid-Holocene conditions are dynamically downscaled with COSMO_iso for the Arctic region. The capability of COSMO_iso to reproduce observed isotopic ratios in Greenland ice cores for these two periods is investigated by comparing the simulation results to measured δ$^{18}$O ratios from snow pit samples, Global Network of Isotopes in Precipitation (GNIP) stations and ice cores. To our knowledge, this is the first time that a mid-Holocene isotope-enabled RCM simulation is performed for the Arctic region. Under present-day conditions, a dynamical downscaling of ECHAM5-wiso (1.1∘×1.1∘) with COSMO_iso to a spatial resolution of 50 km improves the agreement with the measured δ$^{18}$O ratios for 14 of 19 observational data sets. A further increase in the spatial resolution to 7 km does not yield substantial improvements except for the coastal areas with its complex terrain. For the mid-Holocene, a fully coupled MPI-ESM-wiso time slice simulation is downscaled with COSMO_iso to a spatial resolution of 50 km. In the mid-Holocene, MPI-ESM-wiso already agrees well with observations in Greenland and a downscaling with COSMO_iso does not further improve the model–data agreement. Despite this lack of improvement in model biases, the study shows that in both periods, observed δ$^{18}$O values at measurement sites constitute isotope ratios which are mainly within the subgrid-scale variability of the global ECHAM5-wiso and MPI-ESM-wiso simulation results. The correct δ$^{18}$O ratios are consequently not resolved in the GCM simulation results and need to be extracted by a refinement with ...