Journal metrics

Journal metrics

  • IF value: 4.252 IF 4.252
  • IF 5-year value: 4.890 IF 5-year 4.890
  • CiteScore value: 4.49 CiteScore 4.49
  • SNIP value: 1.539 SNIP 1.539
  • SJR value: 2.404 SJR 2.404
  • IPP value: 4.28 IPP 4.28
  • h5-index value: 40 h5-index 40
  • Scimago H index value: 51 Scimago H index 51
Volume 11, issue 7 | Copyright
Geosci. Model Dev., 11, 2563-2579, 2018
© Author(s) 2018. This work is distributed under
the Creative Commons Attribution 4.0 License.

Model description paper 03 Jul 2018

Model description paper | 03 Jul 2018

Comparison of spatial downscaling methods of general circulation model results to study climate variability during the Last Glacial Maximum

Guillaume Latombe1,2, Ariane Burke3, Mathieu Vrac4, Guillaume Levavasseur5, Christophe Dumas4, Masa Kageyama4, and Gilles Ramstein4 Guillaume Latombe et al.
  • 1Centre for Invasion Biology, Department of Mathematical Sciences, Stellenbosch University, Stellenbosch, South Africa
  • 2School of Biological Sciences, Monash University, Melbourne, Australia
  • 3Département d'Anthropologie, Université de Montréal, Montréal, QC, Canada
  • 4Laboratoire des Sciences du Climat et de l'Environnement/Institut Pierre-Simon Laplace, Université Paris-Saclay, CE Saclay, l'Orme des Merisiers, Bât.701, Gif-sur-Yvette, France
  • 5Institut Pierre Simon Laplace (IPSL), Pôle de modélisation du climat, UPMC, Paris, France

Abstract. The extent to which climate conditions influenced the spatial distribution of hominin populations in the past is highly debated. General circulation models (GCMs) and archaeological data have been used to address this issue. Most GCMs are not currently capable of simulating past surface climate conditions with sufficiently detailed spatial resolution to distinguish areas of potential hominin habitat, however. In this paper, we propose a statistical downscaling method (SDM) for increasing the resolution of climate model outputs in a computationally efficient way. Our method uses a generalised additive model (GAM), calibrated over present-day climatology data, to statistically downscale temperature and precipitation time series from the outputs of a GCM simulating the climate of the Last Glacial Maximum (19000–23000BP) over western Europe. Once the SDM is calibrated, we first interpolate the coarse-scale GCM outputs to the final resolution and then use the GAM to compute surface air temperature and precipitation levels using these interpolated GCM outputs and fine-resolution geographical variables such as topography and distance from an ocean. The GAM acts as a transfer function, capturing non-linear relationships between variables at different spatial scales and correcting for the GCM biases. We tested three different techniques for the first interpolation of GCM output: bilinear, bicubic and kriging. The resulting SDMs were evaluated by comparing downscaled temperature and precipitation at local sites with paleoclimate reconstructions based on paleoclimate archives (archaeozoological and palynological data) and the impact of the interpolation technique on patterns of variability was explored. The SDM based on kriging interpolation, providing the best accuracy, was then validated on present-day data outside of the calibration period. Our results show that the downscaled temperature and precipitation values are in good agreement with paleoclimate reconstructions at local sites, and that our method for producing fine-grained paleoclimate simulations is therefore suitable for conducting paleo-anthropological research. It is nonetheless important to calibrate the GAM on a range of data encompassing the data to be downscaled. Otherwise, the SDM is likely to overcorrect the coarse-grain data. In addition, the bilinear and bicubic interpolation techniques were shown to distort either the temporal variability or the values of the response variables, while the kriging method offered the best compromise. Since climate variability is an aspect of the environment to which human populations may have responded in the past, the choice of interpolation technique is therefore an important consideration.

Download & links
Publications Copernicus
Short summary
It is still unclear how climate conditions, and especially climate variability, influenced the spatial distribution of past human populations. Global climate models (GCMs) cannot simulate climate at sufficiently fine scale for this purpose. We propose a statistical method to obtain fine-scale climate projections for 15 000 years ago from coarse-scale GCM outputs. Our method agrees with local reconstructions from fossil and pollen data, and generates sensible climate variability maps over Europe.
It is still unclear how climate conditions, and especially climate variability, influenced the...