The transferability and cross-use of airborne laser scanning-based leaf-off and leaf-on biomass models

Abstract

Airborne laser scanning data (ALS) can be acquired during leaf-on or leaf-off conditions. Thus, the relationship between ALS metrics and vegetation is different, especially in forests dominated by deciduous tree species. We studied the application of leaf-off and leaf-on ALS data in two boreal forest inventory areas in Finland and modelled above ground biomass (AGB). The relative RMSE was typically approximately 20%. In both study areas, we also cross-used the models, i.e. leaf-off model was applied with leaf-on data and vice versa. This increased RMSE% and caused over- and underestimates, especially in plots dominated by deciduous species. However, calibration by empirical ratio estimator (mean between cross-used and correct estimates) removed the over- and underestimates and decreased the RMSE%. When the models were transferred to other study areas and applied with their intended ALS data type, the RMSE% values increased, but only slightly. When the models were transferred to other study areas and cross-used with the wrong ALS data type, the increase in RMSE and over- or underestimation was the largest. However, also the empirical ratio estimator from the other inventory areas could be transferred, and the calibration improved the correctly transferred and cross-used AGB estimates in most cases.