Using biomass distributions to determine probability and intensity of tropical forest disturbance

Background: Tropical forest biomass is not at steady state, being determined by a balance between deterministic growth and stochastic disturbances. Understanding tropical carbon sources and sinks therefore requires better characterisation of the incidence and intensity of disturbance at critical scales.Aims: We determine if information from remotely sensed biomass maps (ALOS-PALSAR) can constrain estimates of miombo woodland biomass dynamics and disturbance.Methods: We analyse biomass maps created over Mozambican woodlands undergoing varied disturbances. We use a simple ensemble model of biomass dynamics to test the hypothesis that biomass distributions can diagnose disturbance processes in specified areas, and use the model to explore the sensitivity of biomass to disturbance parameters.Results: Ensemble runs can reproduce qualitatively similar biomass patterns to those observed in miombo, through varying two parameters that determine frequency and intensity of biomass loss. Using sensitivity analyses, we show for a synthetic case that these two disturbance parameters can be retrieved from satellite observations.Conclusions: Biomass distributions provide enough information to constrain the two critical parameters of the disturbance model, the local probability of disturbance, and its intensity (fraction of biomass lost). These results provide a proof of concept for assimilating biomass maps into models of carbon cycling.