Sensitivity analysis of predicted change in soil carbon following afforestation

Abstract

A credible and cost-effective methodology is needed to support the use of new tree plantations to offset greenhouse gas emissions, and ultimately to form part of an emissions trading scheme. A number of validated models of forest growth are available. However, there has been relatively little validation of models to predict changes in pools of C in litter and soil, and thus suitable for C accounting. A modelling approach is needed to track changes in soil C because direct measurements are currently cost-prohibitive. Modelling approaches also allow for scenario analyses that can be useful for planning purposes. We used a complete C accounting model for forests, GRC3, to simulate patterns of change in soil C following afforestation under four test cases representing typical conditions in Australia. Soil C was predicted to initially decrease (usually during the first 10 years) before a gradual recovery and accumulation of soil C occurred. Sensitivity analyses were used to determine which parameters and inputs potentially cause the greatest uncertainty in calculated change in soil C using GRC3. Taking into account the Uncertainties in the values of parameters and inputs, initial (0-10 years) decrease in soil C was predicted to be 0.96-2.35% per year (or 4.16-14.8 t C ha-1) with a standard deviation between 0.10 and 0.43% per year among case studies, whereas the predicted increase in soil C (10-40 years) was predicted to be between 0.49 and 1.80% per year (or 7.57-24.4 t C ha-1) with a standard deviation between 0.18 and 0.69% per year. Results indicated that uncertainty could be greatly reduced by calibration of the fraction of above-ground litter transferred to soil C (i.e. humification), fraction of C lost by respiration during decomposition of litter, dead roots and soil C, and decomposition rates of the soil C pools. It was also important to obtain accurate input data for initial soil C content (including inert soil C), climatic conditions and allocation of net primary production to various tree components.