Quantifying forest growth and uncertainty across Brazil under potential future climates: combing models and Earth Observation

Forest play a major role in the global carbon cycle storing large amounts of carbon in both living and dead organic matter. Forests can be either a sink or source of carbon depending on the net of far larger fluxes of carbon into (photosynthesis) and out of (mortality, decomposition and disturbance) forest ecosystems. Due to the potential for substantial accumulation of carbon in forests, has led to nationally determined commitments (NDCs) by Governments across the world to protect existing and plant large areas of new forest. However, significant uncertainty remains in our understanding of current forest carbon cycling, especially mortality and decomposition processes, and how carbon cycling will change under climate change. These uncertainties present two connected challenges to effective forest protection and new planting; (i) which existing forests are under the greatest risk to climate change and (ii) where are the most climate safe locations for new forest planting to maximise carbon accumulation.

Here we combine a terrestrial ecosystem model of intermediate complexity (DALEC) with Earth observation (e.g. leaf area, biomass, disturbance) and databased information (soil texture and carbon stocks) within a Bayesian model-data fusion framework (CARDAMOM) to retrieve location specific carbon cycle analyse (i.e. parameter retrievals) across Brazil at 0.5 x 0.5 degree spatial resolution between 2001 and 2015. CARDAMOM allows us to retrieve, independently for each location analysed, an ensemble of parameters for DALEC which are consistent with the location specific observational constraints and their uncertainties. These ensembles give us multiple potential, but observation consistent, realisations of forest carbon cycling and ecosystem traits. We directly quantify our uncertainty in forest carbon cycling and ecosystem traits from these ensembles. The DALEC parameterisations are then simulated into the future under a range of climate scenarios from the CMIP6 model dataset. From these simulations we will, with defined uncertainty, quantify the impact on forest carbon accumulation of existing forest and the potential accumulation of new planting. This information can feed into national planning identifying locations which have the greatest confidence of being a net sink of carbon under climate change highlighting forest areas which are most important to protect and suitable for new planting.