Can Tree Rings Help to Refine Vegetation Modelling?: Fused Empirical Data for Benchmarking Forest Biomass Estimates

Accurate estimations on forest above-ground biomass (AGB) are essential for improving our ability to simulate vegetation response to climate and assess the future role of forests as carbon sinks. In this context, Dynamic Global Vegetation Models (DGVMs) are the most important and rapidly evolving tool to estimate forest carbon dynamics and its potential trajectories in a warmer future at regional to global scales. However, DGVMs remain rather unprecise in their estimation of biomass components due to the lack of representation of growth processes within the model. Thus, enhancing our forest modelling skills at regional and global scales does not only depend on improving the photosynthetic or ecophysiological module (carbon uptake) but to correctly account for allocation into biomass (carbon storage). In this study, we aim to provide a refined, annually resolved empirical AGB estimates that serves as an accurate benchmark to assess the reliability of DGVMs biomass simulations.

To achieve this, we integrate a network of 230 National Forest Inventory (NFI) plots with tree-ring width data collected from the same locations at the Iberian Peninsula, using both frequentist and Bayesian approaches. The NFI data offer detailed forest structure information at the tree and stand levels, typically recorded at 10-year intervals, while tree-ring data provide a reliable measure of annual tree growth. We retrospectively interpolate annual estimates of diameter at breast height (DBH) in trees from NFI plots based on tree-ring width measurements, the climate drivers of tree growth and stand variables. These estimated DBH values are then used to calculate AGB.

Based on our estimated AGB, we assessed the annual net biomass change (NBC) for the last three decades, which allowed us to infer the impact of interannual climate variability and extreme climate events on forest biomass change.  We compared the estimated NBC with Net Primary Production (NPP) outputs from a selected set of DGVMs included in the TRENDY initiative. Our results revealed a general discrepancy between the simulated NPP and the NBC estimates, particularly evident when analyzing the biomass response to extreme climate events. During years marked by extreme summer droughts, such as 1994, 1995, 2003, and 2012, the spatial patterns of NPP anomalies were inconsistent with those observed in the NBC estimates. This discrepancy became even more pronounced during consecutive extreme climate events. During consecutive events, the simulated NPP showed a marked decline during the first year, whereas NBC estimates revealed that drought-induced biomass reduction became more pronounced in the following year due to the legacy effects. These results reflect the source-driven structural deficiency in DGVMs. Incorporating detailed growth dynamics and recovery trajectories into DGVMs is essential for improving their accuracy in a changing climate. In this context, the fusion of tree-ring and NFI data represents a significant advancement, not only for benchmarking DGVMs but also for improving data assimilation procedures.