Perspectives for forest modeling to improve the representation of extensive tree mortality after extreme events

We are currently observing increased tree mortality following multi-year drought events, particularly in low mountain ranges like the Harz Mountains in Germany, where over 70% of spruce stands have died. Forest models are useful tools for understanding the long-term effects of climate change on forest ecosystems, yet struggle to reproduce this massive dieback. In this study, we simulated spruce mortality in the Harz Mountains using five forest models (ForClim, FORMIND, 3-PG-Hydro, LPJ-GUESS, GOTILWA+) of very different complexity. Estimated from the crown condition survey, spruce mortality in the Harz region increased to values above 30% during recent drought years (2018-2020). We found that most models failed to capture these observed high mortality rates, although they showed a clear signal in reduced forest productivity during drought. This discrepancy between the observed high spruce mortality and simulated forest dynamics highlights the need for improved modelling approaches to accurately represent tree mortality processes during and after extreme drought events. We discuss several perspectives for enhancing dynamic forest models by integrating missing processes prospectively. This includes novel (i) process-based drought mortality approaches, (ii) enhanced description of eco-physiological processes like plant hydraulics, (iii) data-driven and AI approaches, and (iv) improved representation of biotic damaging agents (i.e., insects and pathogens). Incorporating these perspectives into forest models has the potential to improve their ability to simulate forest dynamics under extreme drought, ultimately contributing to the assessment of forest resilience and informing adaptive management strategies in Germany and beyond.