Distinguishing vegetation types with airborne waveform lidar data in a tropical forest-savanna mosaic: A case study in Lopé National Park, Gabon

Abstract

Tropical forest vegetation structure is highly variable, both vertically and horizontally, and provides habitat to a large diversity of species. The forest-savanna mosaic in the northern part of Lopé National Park, Gabon, has a large and complex variation in vegetation structure along a successional gradient. The goal of this research is to assess whether large footprint full-waveform lidar data can be used to distinguish successional vegetation types based on their vertical structure in this area. Eleven vegetation metrics were derived from the lidar waveforms: canopy height, canopy fractional cover, total Plant Area Index (PAI) and vertical profile of PAI. The PAI profiles from airborne waveform lidar showed good agreement with those from Terrestrial Laser Scanning, sampled at eight field plots across different vegetation types (r² = 0.95, RMSE = 0.63, bias = 0.41). The agreement further strengthened our confidence that lidar waveforms can be used to distinguish between the five vegetation types, within the limits of the sampled structure, because TLS was known to provide distinct PAI profiles for these vegetation types. We then employed a Random Forest model, trained with 193 locations of known vegetation type, to classify the entire study area into five successional vegetation types (classification accuracy = 81.3%). The resulting predictive map revealed the overall spatial pattern of vegetation types across the study area. Our results suggest that lidar-derived vegetation profiles can provide valuable information on vegetation type and successional stage. This, in turn, can further help to improve habitat and biodiversity conservation and forest management activities.

Plant phenotyping relevance

航空波形LiDARから樹冠高、被覆率、PAI、垂直構造を抽出し、TLSとの一致度を検証したうえで植生構造・遷移段階を分類しており、植物状態の取得手法が中心です。

Code and data availability

公開本文の所在を確認できませんでした。非公開または購読が必要な可能性があります。

No evidence-backed public reproduction asset is currently recorded.