Modeling and monitoring CO2 fluxes and stocks in forest ecosystems

Although there is no shortage of data and models to assess stand biomass, there remains the task of examining the dynamics of above- and below-ground biomass by way of a generic modelling approach. To be of general use, such an approach should use numerically expressed parameters, such as stand compartments (leaf biomass, above-ground woody biomass, roots, litter, dead wood, and soil organic matter), transition flows, rates of production, and decay processes of biomass in relation to the particular conditions determined by site and species composition. In turn, the assessment of these conditions should allow for efficient and realistic representations and simulations of whole-stand biomass dynamics (plant and soil) as a function of stand dynamics, silvicultural interventions, and other factors that may influence stand development. A modelling approach that has been used widely to explain nutrient dynamics, such as the carbon cycle, is to subdivide the forest ecosystem in more or less homogeneous compartments. This permits to simulate mass flows between the system compartments inside the system, and between those and the world outside the system. For example, the CO2FIX model uses a stand-based approach as the starting point. In this model, estimations of living carbon pool dynamics are driven by stem wood increment, with ratio allocations to leaves, branches, and roots. The dynamics of the rest of the system are based on estimated or published transfer coefficients of mass between the various compartments. Knowing or estimating these coefficients and incorporating variations in the value of various parameters in an uncertainty analysis, allows for analytical simulation of biomass fluxes with statistically robust confidence intervals for different site and species combinations. In the example that is used to demonstrate the model, each forest cover class is simulated with a separate set of initial compartment state and transfer coefficients and C fluxes estimated with a 95% confidence interval, using randomly selected values of the parameters from a normally distributed population around the default value. Each class associated to a particular biomass group was assumed to behave similarly in terms of biomass dynamics, such that total variation within a group was expected not to exceed the limits of the confidence interval. Monitoring biomass dynamics of a selection of plots within each biomass group for a number of years is underway to test this latter hypothesis. Standard monitoring procedures have been developed to track the changes over time of those compartments that are prone to rapid change during stand development.