Aboveground biomass relationship with canopy cover and vegetation to improve carbon change monitoring in rangelands

Abstract

Rangelands cover vast areas of the global land surface and are important to the terrestrial carbon budget. However, carbon accounting in rangeland systems is often limited by the lack of transparent and systematic methods for assessing changes in aboveground biomass (B_AG). Although relationships between B_AG and canopy cover, C, have been investigated at site and regional scales, there are few studies across regions where the impact of a range of vegetation types and site conditions has been assessed. Here, results were compiled from extensive field measurements across 431 Australian rangeland sites (covering an area of ~6 million km²) to develop empirical relationships to predict B_AG from C and other structural variables. A boosted-regression-tree model was trained to identify the relative importance of predictor variables. Then, based on these results, a stepwise empirical log-linear relationship was developed to estimate B_AG. About 70% of the B_AG could be described using C, the percentage of large trees (stem diameter >50 cm), and height. Because such detailed information is not yet available at sufficient spatial and temporal resolution, classifications based on existing maps of structural vegetation classes, using C as the single predictor variable, were explored as an alternative approach to estimate B_AG. For most structural vegetation classes assessed, estimates of B_AG from C were statistically significant, with Lin's concordance coefficients of 0.67–0.79 and proportional error of <36% relative to the B_AG across all the classes. There was generally little improvement in model performance with the inclusion of additional explanatory variables. Overall, this study has improved our understanding of relationships between C and B_AG across rangeland systems. Additionally, combining remotely sensed woody cover data with these relationships may offer a transparent and accurate approach to monitor changes in biomass carbon stocks in these ecosystems at a large spatial scale.