EGU General Assembly 2020
© Author(s) 2020. This work is distributed under
the Creative Commons Attribution 4.0 License.

Model intercomparison of idealized global deforestation experiments

Victor Brovkin1, Lena Boysen1, Julia Pongratz2, Nicolas Vuichard3, Philippe Peylin3, and David Lawrence4
Victor Brovkin et al.
  • 1Max Planck Institute for Meteorology, The Land in the Earth System, Hamburg, Germany (
  • 2LMU, Department of Geography, Munich, Germany
  • 3LSCE/IPSL, Gif sur Yvette, France
  • 4NCAR, Boulder, CO, USA

We present first results of idealized deforestation experiment designed within the Land Use Model Intercomparison Project (LUMIP). In order to obtain a robust signal-to-noise ratio and to harmonize deforestation implementation across participating ESMs, global forest extent is linearly decreased by 20 million km2 for the 30% of most forested grid cells over a period of 50 years starting from pre-industrial climate conditions. This experimental setup is in favor of predominantly tropical deforestation patterns, however, there is also substantial boreal deforestation. In this experiment, atmospheric and oceanic physical processes respond to large-scale deforestation while other forcings such as atmospheric CO2 concentration and aerosol load are kept constant at the pre-industrial level.

First analysis of results from ESMs participating in the LUMIP experiments reveal a general cooling trend in response to deforestation, although a spread in an amplitude of response is substantial. In boreal region there is significant cooling effect, presumably due to an increase in surface albedo, while tropical deforestation results in a regional warming in most of models. A sensitivity of temperature change per forest fraction change on a grid cell level, ∂T/∂F, likely could be used as a generic response for any forest change scenario, although it is complicated by mixing together local and non-local effects. We also quantified so-called “zero effect latitude” at which forest cover change does not have pronounced biogeophysical effect. It is located in northern subtropics in most models.

Analyses of ensemble-members of three models (MPI-ESM1.2-LR, IPSL-CM6A-LR, and CESM2) indicate that the “time of emergence” of climate response, when signal becomes larger than a noise, is quite different among the models. However, when we compare the “deforested fraction of emergence”, the model responses become much more coherent. Biomass and soil carbon storages are decreasing with time, and their “time of emergence” is much shorter comparing to the temperature and precipitation. More results of biogeophysical and biogeochemical responses to deforestation will be presented.

How to cite: Brovkin, V., Boysen, L., Pongratz, J., Vuichard, N., Peylin, P., and Lawrence, D.: Model intercomparison of idealized global deforestation experiments, EGU General Assembly 2020, Online, 4–8 May 2020, EGU2020-10295,, 2020

Comments on the presentation

AC: Author Comment | CC: Community Comment | Report abuse

Presentation version 1 – uploaded on 03 May 2020
  • CC1: Comment on EGU2020-10295, Andrea Alessandri, 07 May 2020

    Dear Victor and All, 
    Very interesting work
    It seems to be quite considerable non-local effects: to what extent you think the differences in response among models is due to differences in the representation of the non-local effects? E.g. some models have strong signal of non-local positive feedback from sea-ice (large cooling over Arctic for CanESM and UKESM); divergence (warming or cooling) over tropics also may point to possible effects in large-scale Circulation…
    These non-local effects are very interesting to me actually; have you done or have you planned any analysis on this?

    I’ll also try to have a look at these simulations if not done already…

    Kind Regards

    • AC1: Reply to CC1, Lena Boysen, 07 May 2020

      Hi Andrea,
      thanks for your question! Yes, the manuscript is finalizing and in that we account for the importance of non-local effects. Unfortunately, we cannot separate them explicitly but it is obvious that models are not equally sensitive to the deforestation forcing. In the tropics, changes in circulation or simply the parametrization of grasses (eg higher ET in EC-Earth or CESM2) cause the observed signals. A further manuscript is currently being set up on the changes in atmospheric circulation.