Application of Photogrammetric Software for Digital Canopy Height Modelling from Old Aerial Photographs

Abstract

Accurate digital canopy height models (DCHMs) derived from historical aerial photographs are essential for reconstructing long-term forest structural dynamics; however, the influence of photogrammetric software on DCHM quality and reliability remains insufficiently evaluated. This study compared the performance of two structure-from-motion (SfM) photogrammetric platforms, Metashape and Pix4Dmatic, for processing old aerial photographs and generating DCHMs in Ishikawa prefecture. Software performance was assessed using image processing efficiency, geometric accuracy based on root mean square error (RMSE), and correlation between derived DCHMs and National Forest Inventory (NFI) measurements. The results revealed that Metashape required shorter image processing times for the digital surface model generation and produced denser point clouds with broader spatial coverage. By contrast, Pix4Dmatic achieved higher geometric accuracy, with RMSE values of 0.571 m, 0.870 m, and 2.120 m in the X, Y, and Z directions, respectively. The Metashape-derived DCHM showed a higher mean value (15.267 ± 5.882 m) than Pix4Dmatic (14.749 ± 5.834 m), but Pix4Dmatic-generated DCHMs showed a closer relationship (r = 0.880) with NFI data (15.322 ± 5.451 m). These findings demonstrate that photogrammetric software selection substantially influences three-dimensional reconstruction from old aerial imagery and affects the reliability of DCHM generation. This study provides practical guidance for selecting SfM software for forest structural analysis and long-term forest monitoring.

Plant phenotyping relevance

森林樹冠高という植物群落の構造形質を対象に、SfMソフトウェアを比較評価し、DCHM生成の精度・信頼性を検証しているため、方法検証が中心である。

Code and data availability

The supplied blocks describe photogrammetric DCHM generation from old aerial photographs and comparison of Metashape vs. Pix4Dmatic, but contain no public phenotype/trait dataset, image deposit, author analysis code, or trained model with an availability statement or URL. Input data (aerial films, NFI plots, GSI DTM)都是

No evidence-backed public reproduction asset is currently recorded.