Wood Formation Model Intercomparison for Their Suitability in Global Vegetation Models

Abstract:

Wood formation is the irreversible incorporation of carbon (C) into trees and acritical process in forest C sequestration. Despite of its importance, this processhas not been explicitly represented in the current generation of dynamic globalvegetation models (DGVMs). DGVMs’ projections of the carbon sink on landare highly uncertain due to large spatio-temporal disagreements in vegetationresponses to climate and CO2 [Friend et al., 2014, Smith et al., 2016, Klesseet al., 2018]. The lack of an explicit representation of growth processes may beone contributing factor to these uncertainties [Zhang et al., 2018, Fatichi et al.,2019]One important argument to model wood formation is that C incorporationinto plant biomass may be more limited by environmental factors such as wateravailability or temperature than C input through photosynthesis. Therefore,it needs to be tested whether the incorporation of wood formation processescan contribute to better explain vegetation C dynamics in time and space. To-wards this aim, our study investigates the suitability of four wood formationmodels for the incorporation in a global modelling framework. The models havebeen successfully applied in various other disciplines, are of varying complexityand respond to different environmental and internal drivers: The first modelby Deleuze and Houllier [1998] is a parsimonious model that responds to car-bon, water and temperature. The VS-model [Vaganov et al., 2006] responds towater, temperature and daylength and focuses on cell production only. Drewand Downes [2015] ’s model represents C and water dynamics in a complex wayand Friend [2020] considers C and temperature effects. The models are param-eterised using observations of wood formation dynamics and compared againstfinal year ring width and cell numbers at French Vosges mountains [Cuny et al.,2012]. They are evaluated based on their model-data fit, level of complexityand suitability for incorporation into a global modelling framework