Modifying the G'DAY process-based model to simulate the spatial variability of Eucalyptus plantation growth on deep tropical soils

International audience

anglais. Large differences in productivity have been observed between neighboring Eucalyptus plantations in Brazil, that cannot be explained by climate and are unlikely to be due solely to altered management practices. Current ecophysiological models used by forestry companies to simulate stand development in large plantation zones rely on empirical site fertility indices (representing water and nutrient availability) to capture this spatial variability in growth rates. We propose a model that requires no empirical assessment of site fertility to simulate stand growth over entire rotations. We applied a modified version of the G'DAY model of carbon, nitrogen and water cycling at a daily time step to short-rotation plantations located in Sao Paulo State, including a simple mechanistic description of the effect of water availability on growth. The progressive and rapid root exploration of deep soil layers was modeled in a simple way, by considering that maximum plant available water increased with mean tree height. The model was parameterized using detailed measurements made over the entire rotation of an experimental stand of Eucalyptus grandis, and was subsequently applied to 16 clonal stands managed in a similar way by one company, but with different planting dates and contrasting productivity levels. Stem biomass simulations, driven by daily weather data (maximum and minimum air temperatures, global radiation and rainfall), were strongly correlated with company inventory estimates of stem biomass carried out at different ages. The temporal variation of leaf area index was also adequately simulated, as was shown by comparison with leaf area index derived from satellite data. The model was able to capture more than 95% of the variability of standing stem biomass and more than 85% of the variability of stem growth measured on these stands, provided spatial differences in soil water holding capacity were taken into account.