Multidimensional Inconsistency in Forest Ecosystem Representation: An NLP-Assisted Thematic Review

Oluwafemi David Bejide 

Forest ecosystem monitoring increasingly relies on multisensor remote sensing approaches integrating optical imagery, synthetic aperture radar (SAR), and LiDAR observations to assess biomass, degradation, and ecosystem condition. However, these systems frequently generate inconsistent representations of the same ecosystem due to differences in sensor sensitivity, ecological complexity, scale interactions, and recovery dynamics. This study conducted a natural language processing (NLP)-assisted structured thematic review to synthesize the dominant drivers, manifestations, and methodological responses associated with multidimensional inconsistency in forest ecosystem representation. A total of 500 publications were retrieved from the Dimensions database, of which 181 studies were retained following semantic relevance screening, thematic filtering, and manual verification.
Thematic synthesis showed that combined SAR and structural indicators dominated contemporary forest remote sensing literature, followed by ecosystem condition monitoring and multisensor fusion approaches. Ecological heterogeneity and terrain-related structural complexity emerged as the dominant drivers of inconsistency, accounting for 30.2% of all primary causes and occurring in 59.5% of all detected inconsistency pathways. Scale mismatch, sensor saturation, and benchmark instability also represented major sources of representational uncertainty. Biomass and carbon estimation uncertainty emerged as the most widespread manifestation, occurring in 94.0% of the reviewed studies. The review further demonstrated that spectral, structural, and functional ecosystem dimensions frequently become partially decoupled following disturbance, producing misleading representations such as spectrally recovered but structurally degraded “green deserts.” The study proposes a multidimensional inconsistency framework linking ecological, methodological, structural, and functional drivers of representational instability.