Modeling Phenological Responses to Climate Change: Uncertainty and Missing Drivers

Friday, 19 December 2014: 3:10 PM
Mirco Migliavacca, Max Planck Institute for Biogeochemistry, Jena, Germany, Andrew D Richardson, Harvard University, Cambridge, MA, United States, Alessandro Cescatti, Joint Research Center Ispra, Ispra, Italy, Edoardo Cremonese, ARPA Vda, Aosta, Italy, Trevor F Keenan, Macquarie University, Sydney, Australia and Oliver Sonnentag, University of Montreal, Montreal, QC, Canada
Phenology, the timing of recurring life cycle events, controls numerous land surface feedbacks to the climate system through the regulation of exchanges of carbon, water and energy between the biosphere and atmosphere. Vegetation phenology is affected by climate change and weather variability. It is widely established that warmer temperatures lead to earlier spring onset and might lead to delayed autumn senescence in temperate and boreal ecosystems, while water controls the phenology patterns in arid and Mediterranean ecosystems.

Despite increasing understanding of the main drivers of phenological events, terrestrial biosphere models are known to have systematic errors in the simulation of spring phenology, which potentially could propagate to uncertainty in modeled responses to future climate change.

This is mainly due to misparameterization, model structural errors, and missing processes in phenology models..

Here, we use different data streams collected over spatial gradients in North America and Europe to investigate and characterize the sources of model uncertainty. By using a model-data fusion approach, we combined information from phenological observations of different woody species, with 12 leaf bud-burst models that varied in complexity.

We discuss the uncertainty associated with model parameterization and with the climate drivers used to forecast phenology. Moreover, we discuss the differences in model selection (forcing, chilling, and photoperiod) and parameterizations when the models are applied to the same woody species along the gradients.

Finally we evaluate other potential biological controls (e.g. foliar stoichiometry, maximum rates of photosynthesis at saturating irradiance at ecosystem level) on plant phenology by evaluating the model residuals against a variety of plant traits.