Biogeosciences (Apr 2021)

Plant phenology evaluation of CRESCENDO land surface models – Part 1: Start and end of the growing season

  • D. Peano,
  • D. Hemming,
  • S. Materia,
  • C. Delire,
  • Y. Fan,
  • Y. Fan,
  • E. Joetzjer,
  • H. Lee,
  • J. E. M. S. Nabel,
  • T. Park,
  • T. Park,
  • P. Peylin,
  • D. Wårlind,
  • A. Wiltshire,
  • A. Wiltshire,
  • S. Zaehle

DOI
https://doi.org/10.5194/bg-18-2405-2021
Journal volume & issue
Vol. 18
pp. 2405 – 2428

Abstract

Read online

Plant phenology plays a fundamental role in land–atmosphere interactions, and its variability and variations are an indicator of climate and environmental changes. For this reason, current land surface models include phenology parameterizations and related biophysical and biogeochemical processes. In this work, the climatology of the beginning and end of the growing season, simulated by the land component of seven state-of-the-art European Earth system models participating in the CMIP6, is evaluated globally against satellite observations. The assessment is performed using the vegetation metric leaf area index and a recently developed approach, named four growing season types. On average, the land surface models show a 0.6-month delay in the growing season start, while they are about 0.5 months earlier in the growing season end. The difference with observation tends to be higher in the Southern Hemisphere compared to the Northern Hemisphere. High agreement between land surface models and observations is exhibited in areas dominated by broadleaf deciduous trees, while high variability is noted in regions dominated by broadleaf deciduous shrubs. Generally, the timing of the growing season end is accurately simulated in about 25 % of global land grid points versus 16 % in the timing of growing season start. The refinement of phenology parameterization can lead to better representation of vegetation-related energy, water, and carbon cycles in land surface models, but plant phenology is also affected by plant physiology and soil hydrology processes. Consequently, phenology representation and, in general, vegetation modelling is a complex task, which still needs further improvement, evaluation, and multi-model comparison.