Perspectives of regional paleoclimate modeling

Abstract

Regional climate modeling bridges the gap between the coarse resolution of current global climate models and the regional-to-local scales, where the impacts of climate change are of primary interest. Here, we present a review of the added value of the regional climate modeling approach within the scope of paleoclimate research and discuss the current major challenges and perspectives. Two time periods serve as an example: the Holocene, including the Last Millennium, and the Last Glacial Maximum. Reviewing the existing literature reveals the benefits of regional paleo climate modeling, particularly over areas with complex terrain. However, this depends largely on the variable of interest, as the added value of regional modeling arises from a more realistic representation of physical processes and climate feedbacks compared to global climate models, and this affects different climate variables in various ways. In particular, hydrological processes have been shown to be better represented in regional models, and they can deliver more realistic meteorological data to drive ice sheet and glacier modeling. Thus, regional climate models provide a clear benefit to answer fundamental paleoclimate research questions and may be key to advance a meaningful joint interpretation of climate model and proxy data.

Keywords: paleoclimate; proxy data; regional climate modeling.

Figure 1

From global to regional scales: a schematic of a typical nested domains setup in dynamical downscaling exercises. In this case, up to four domains have to be nested to downscale the GCM fields from a spatial resolution of 1° to the final target of 2 km over the Alpine region. In each step of the simulation, the RCM iteratively resolves the equations for each outer domain and uses this information to provide the boundary conditions to the interior one.

Figure 2

Forcing used for the Last Millennium simulations. From top to bottom: (A) Different total solar irradiance (TSI) reconstructions;107 (B) example of a volcanic forcing as total volcanic aerosol mass; (C) radiative forcing (RF, calculated according to IPCC) from the greenhouse gases CO₂, CH₄, and N₂O;108 (D) major changes in land cover (as fraction of global land area).107

Figure 3

Thirty‐one‐year running mean of surface air temperature (K) (SAT, left) and precipitation (mm/month) (right) anomaly series averaged for the nine subregions defined in the map on top. In all cases, light and dark colors represent the model output and the reconstruction, respectively. All series depict anomalies respect to the preindustrial period (1500–1850).

Figure 4

(A) Orbital parameters over the past 800 ky and atmospheric concentration of CO₂ from Antarctic ice cores (adapted from IPCC 2013, fig. 5.3).1 (B) RCM domain adapted to LGM surface boundary conditions (land‐sea‐mask and ice‐sheets based on PMIP3 21Ka experimental design).72

Figure 5

Simulated annual average GCM/RCM precipitation amount for LGM conditions and comparison of precipitation differences between LGM and preindustrial climate conditions based on proxy data: simulated precipitation (mm y⁻¹) (A) for MPI‐ESM‐P and (B) for WRF. Precipitation differences (Δmm y⁻¹) between corresponding LGM and PI simulations (C) for MPI‐ESM‐P and (D) for WRF. (E) Pollen‐based precipitation difference (Δmm y⁻¹) between LGM and PI.75 (F) and (G) as (C) and (D) but precipitation differences interpolated on proxy data grid (adapted from Ludwig et al.).30

Figure 6

Schematic of GCM/RCM—proxy data comparison for temperature in the Pyrenees: (A) part of the GCM model domain (orography shaded), black box marks RCM domain, red cross marks grid point for time series data in (D); (B) RCM model domain (orography shaded), blue box marks area averaged over the Pyrenees for RCM data in (D), black arrow illustrates location of (C) tree ring used as proxy. (D) Synopsis of GCM, RCM, and proxy data time series.