Implementing new hydraulic architecture in ORCHIDEE-CAN to model tree mortality risk in response to drought

Abstract:

Extreme drought events affecting the Amazonia region are expected to become more frequent and more intense along the evolution of climate change. To be able to anticipate future drought effects on Amazonia vegetation, estimate for tree vulnerability in coping with such adverse environments is required. However, most state-of-the-art dynamic global vegetation models are relatively poor in mechanistic description of physiological processes like hydraulic failure, which challenges their abilities to accurately capture the plant ecophysiological responses to drought stress. Here, we implemented a mechanistic plant hydraulic module into the ORCHIDEE-CAN land surface model to produce dynamic water potential at leaf, stem and root levels and percentage loss of conductivity (PLC). The model is well evaluated against experimentally observed seasonal variability in stand-scaled sap flux, soil moisture and productivity under both control and throughfall exclusion experiments in Caxiuana site during 2001 to 2008. Relationship between PLC and tree mortality is also built with assemblages of two empirical parameters, cumulated drought exposure days to trigger mortality and mortality fraction in each day that meets the requirements. Our model captures the large biomass drop in year 2005, and produces comparable annual tree mortality rates with observation from throughfall exclusion experiments over the study period. Our new hydraulic architecture provides important and promising avenues for future research in assimilating the experimental data into drought-induced xylem dysfunction – mortality routine. We also highlight that species-based (isohydric or anisohydric) hydraulic traits should be further tested to generalize the model performance in predicting the drought risks.