Geoscientific Model Development (Nov 2020)

Description and evaluation of the process-based forest model 4C v2.2 at four European forest sites

  • P. Lasch-Born,
  • F. Suckow,
  • C. P. O. Reyer,
  • M. Gutsch,
  • C. Kollas,
  • C. Kollas,
  • F.-W. Badeck,
  • H. K. M. Bugmann,
  • R. Grote,
  • C. Fürstenau,
  • M. Lindner,
  • J. Schaber

DOI
https://doi.org/10.5194/gmd-13-5311-2020
Journal volume & issue
Vol. 13
pp. 5311 – 5343

Abstract

Read online

The process-based model 4C (FORESEE) has been developed over the past 20 years to study climate impacts on forests and is now freely available as an open-source tool. The objective of this paper is to provide a comprehensive description of this 4C version (v2.2) for scientific users of the model and to present an evaluation of 4C at four different forest sites across Europe. The evaluation focuses on forest growth as well as carbon (net ecosystem exchange, gross primary production), water (actual evapotranspiration, soil water content), and heat fluxes (soil temperature) using data from the PROFOUND database. We applied different evaluation metrics and compared the daily, monthly, and annual variability of observed and simulated values. The ability to reproduce forest growth (stem diameter and biomass) differs from site to site and is best for a pine stand in Germany (Peitz, model efficiency ME=0.98). 4C is able to reproduce soil temperature at different depths in Sorø and Hyytiälä with good accuracy (for all soil depths ME > 0.8). The dynamics in simulating carbon and water fluxes are well captured on daily and monthly timescales (0.51 < ME < 0.983) but less so on an annual timescale (ME < 0). This model–data mismatch is possibly due to the accumulation of errors because of processes that are missing or represented in a very general way in 4C but not with enough specific detail to cover strong, site-specific dependencies such as ground vegetation growth. These processes need to be further elaborated to improve the projections of climate change on forests. We conclude that, despite shortcomings, 4C is widely applicable, reliable, and therefore ready to be released to the scientific community to use and further develop the model.