MUSCATELLA: Multiscale Spatiotemporal Forest Landscape Modeling

A forest landscape model, in particular its migration module, is upscaled to the continental scale.

Topics: Landscape dynamics, landscape ecology, forest dynamics, upscaling, ecological modelling

Frame

This project is supported by the Swiss National Science Foundation. It belongs to the group of projects aiming to improve the TreeMig forest landscape model for climate change impact assessments on regional to continental scales.

Project members and partners

• Heike Lischke
  

• NN, WSL
  

Rationale

Spatiotemporal vegetation dynamics, including   plant species migration, play an important role for carbon sequestration, climate-biosphere-feedbacks and biodiversity, particularly in the changing climate. To assess future vegetation dynamics, large scale, spatiotemporal vegetation models simulating plant migration are indispensable. The models currently available simulate vegetation in an oversimplified way, neglect spatial processes, or are too complex and computationally inefficient to be applied on a large, e.g. continental scale.  Particularly the simulation of the spatial interactions, i.e. seed dispersal and migration, is encumbered by discretization errors and long computing times.

Research goals

The principal goal of the project is to scale a spatiotemporal forest model from the landscape to the continental scale and to obtain a model which…

• is reliable, i.e. reproduces the aggregated small scale results
• is computationally efficient,
• and can be operated from local to continental scales with changing resolution

Methods

• We evaluate and adapt multiscale methods addressing the spatial and the involved hierarchical upscaling of small scale models. Such methods are
  
  • multiresolution simulations,
    • which adapt the resolution depending on the actual gradients in the simulation
  • or coarse graininig,
    • e.g. by combining simulations on a subset of fine-scale grid cells within each coarse grid-cell and interpolation of the results to the coarse cell,
    • or by deriving simpler models (metamodels) based on simulation results of

      the base model  to assess within-cell patterns and migration
      

• The two most promising methods of the evaluation will be selected and implemented into the spatiotemporal forest landscape model TreeMig.
• The validity and efficiency of the chosen methods will be assessed in a comparative study on different scales.
• The multi-scale model will then be tested in Swiss to continental scale applications (cf. project BBS)