Modelling past and future impacts of droughts on tree mortality and carbon storage in Norway spruce stands in Germany

The consecutive drought events between 2018 and 2020 caused an unprecedented increase in Norway spruce (Picea abies) tree mortality across Germany. Despite the observable forest dieback, process-based vegetation models have difficulty reproducing it. This gap between observed and simulated forest dynamics underscores the pressing need for more advanced modeling approaches to accurately capture drought-induced tree mortality.

In our study, we adopted a data-driven statistical approach to enhance the representation of drought-induced Norway spruce tree mortality in the process-based vegetation model LPJ-GUESS. Using Norway spruce mortality data from the German Crown Condition Survey (Waldzustandserhebung, WZE), as well as climate and weather anomaly data, we developed logistic regression models to predict drought-induced tree mortality, which were then integrated into LPJ-GUESS.

This enhanced modeling framework successfully reproduced the general temporal and spatial patterns of historical Norway spruce mortality rates (1998–2020). Future simulations (2021–2070) under the RCP2.6 and RCP8.5 climate scenarios show periodic increases in Norway spruce mortality, comparable to or even exceeding the high rates observed in 2020. Although the drought-mortality models effectively replicate past dynamics, they diverge in predicting the timing and magnitude of future drought-induced mortality events.

The vegetation model also enabled us to quantify the impacts of mortality on forest productivity. Our projections indicate a drought-driven reduction in aboveground biomass of 18% under RCP2.6 and 36% under RCP8.5 (mean across all simulations). Moreover, we observed a significant decline in potential spruce timber harvests in Germany between 2021 and 2070, with cumulative losses amounting to 310 million Mg of C under RCP2.6 and 447 million Mg of C under RCP8.5. These impacts vary depending on the chosen climate scenario and the statistical mortality model applied.

Our study highlights the severe risk of large-scale future dieback in Norway spruce forests across Germany. However, the prediction of the timing and magnitude of such events remains highly uncertain. Nevertheless, the effects of droughts should be considered in predictive modeling studies, as they could have significant impacts on forest carbon cycling and timber harvests.