Democratizing 3D ecology: Mobile neural radiance field for scalable ecosystem mapping in change detection

Abstract

High-resolution, three-dimensional monitoring is increasingly essential for capturing ecological dynamics, yet conventional approaches such as terrestrial laser scanning (TLS) and photogrammetry remain limited by cost, accessibility, and technical barriers. Here, we introduce and evaluate the application of mobile neural radiance field (NeRF) methods for ecological research. Leveraging consumer-grade smartphones and open-source platforms (e.g. Luma AI), we demonstrate that mobile NeRFs can reconstruct detailed 3D structures of vegetation with accuracy comparable to TLS in open-canopy environments. We assess the strengths and limitations of NeRFs across habitat types, showing that while performance declines under occlusion (e.g. dense canopies), these methods excel at capturing understory complexity, making them particularly valuable for savannas, grasslands, and urban systems. We further explore the potential of radiance fields to integrate hyperspectral and robotic data streams, expanding their utility for dynamic ecosystem monitoring. By reducing hardware requirements and broadening participation, mobile NeRFs offer a promising avenue for democratising ecological data collection and advancing scalable environmental surveillance.

Plant phenotyping relevance

モバイルNeRFによる植生の3D構造取得を中心に開発・評価し、TLSとの精度比較や遮蔽条件での性能評価を行っているため、植生構造の画像ベース表現型計測法として収載対象。

Code and data availability

The article describes NeRF/TLS phenotyping measurements (DBH, height, crown area, point-density profiles) but contains no data availability statement, no public dataset deposit, and no author code/scripts/workflows with a public URL. The only URL present (GSMA report) is a cited reference about smartphone access, not a

No evidence-backed public reproduction asset is currently recorded.