Extraction of remote sensing-based forest management units in tropical forests

► We segmented maps of stem volume and basal area are predicted using empirical models. ► Models were established for ALS, Airborne CIR, ALOS AVNIR-2 and their hybrids. ► We propose an objective method using AICvar index to assess delineation quality. ► We propose a criterion to compare segmentation results for different data sources. ► The quality of segmentation reasonably reflected that of the empirical models.