Modelling the response of forests to climate change : the impact of drought : The case of Walloon forests simulated with the HETEROFOR model

Details

Supervisors

Goosse, Hugues ; Jonard, Mathieu

Faculty

Faculté des bioingénieurs

Degree label

Master [120] : bioingénieur en sciences et technologies de l'environnement, à finalité spécialisée

Abstract

enWalloon forests cover nearly a third of the region's territory and are increasingly exposed to climate change. However, its precise impact on tree productivity, water stress and vitality remains poorly quantified at the stand scale. This is further complicated by the fact that climate models tend to underestimate the rate of recent warming observed in Europe, which raises questions about the reliability of forest projections based on them. This master thesis addresses both issues: it quantifies how climate change affects Walloon forests using a process-based model, and assesses how far biases in the climate data used propagate into these projections. The study uses the HETEROFOR model to simulate five stands dominated by common beech and sessile oak across three sites (Baileux, Chimay, Louvain-la-Neuve), chosen to represent the pedo-climatic diversity of Wallonia. Simulations cover the historical period (1961-2024, forced by observed IRM data) and the future period (2025-2100, forced by the GFDL-ESM4 model under scenario SSP3-7.0). Eight simulation configurations were designed to isolate the effects of CO₂ fertilisation and of a hydraulic cavitation module. An additional set of historical simulations forced by GFDL-ESM4 was used to quantify how the biases of this climate model propagate through HETEROFOR. Water stress is projected to roughly double by 2100, shifting from occasional episodes to a nearpermanent condition that leaves trees little time to recover between stress periods. Net primary productivity follows a three-phase trajectory: a moderate historical increase, a peak around 2055–2060, and a subsequent decline that erases most of the gains accumulated over the 20th century, as rising temperature switches from a stimulating to a limiting factor once combined with intensifying water deficit. Sessile oak, owing to a higher hydraulic conductivity, maintains markedly more stable productivity than common beech, whose relationship with temperature deteriorates earlier under stress. Regarding climate model biases, GFDL-ESM4 was found to overestimate absolute temperature levels (+0.9 to +2.3°C depending on the site) while simultaneously underestimating the recent warming trend relative to IRM observations. These two biases pull the projections in opposite directions, and their relative weight varies by site and indicator, so no single correction can be applied uniformly across the results. Overall, this work shows that Walloon forests face a strong intensification of water stress, and that the window to adapt forest composition before productivity gains are reversed is limited. Sessile oak emerges as a more resilient species than beech under these conditions, supporting its promotion, together with broader species diversification, as an adaptation strategy. The analysis of climate model biases also shows that they do not simply make the projections too pessimistic or too optimistic overall: their effect depends on the site and indicator considered and cannot be corrected with a single adjustment.