Influence of forest age on parameter estimation in a temperate forest ecosystem using model-data assimilation

Model-data assimilation techniques have recently received considerable interest in forging novel understanding in carbon, water and energy exchanges between forests and the atmosphere. Assimilation approaches are used to parameterize, calibrate, evaluate and validate ecosystem process models with ground based observation of mass and energy fluxes, as well as remote sensing fields pertaining to climatic and biospheric processes. Model-data assimilation studies have generally focused on mid- to old-age forests which undergo minimal inter-annual changes in vegetation structure compared to young actively regenerating forests. Large inter-annual vegetation structure changes, especially in young forest stands influenced by suitable climatic, edaphic and biological (e.g., seed abundance) conditions can lead to large inter-annual variations in ecosystem functioning (e.g., carbon, water and energy fluxes). We explore how the constrained parameters controlling these fluxes vary yearly in a recent clear cut stand, and compare these to a mature forest stand in Harvard forest. A soil-vegetation-atmosphere transfer (SVAT) scheme is used to constrain parameters pertaining to photosynthesis-conductance, multi-layer light profile, turbulent fluxes, two source energy and water fluxes, and multi-layer soil heat and water transfer using the Markov chain Monte Carlo (MCMC) assimilation approach.