Improving tree diameter measurements above irregularities in Central African forests: a Close-Range Photogrammetric approach

Abstract

Accurate measurement of tree diameter in forests is essential for sustainable management of forest resources, ecological assessment, and scientific research. However, most trees in tropical forests have irregularities at the base of the trunk, making it challenging to measure the trunk diameter above them with a tape measure. To meet the increasing demand for data accuracy and reliability, approaches using three-dimensional (3D) point clouds offer a valuable new source of data for tree measurements. This study examines the accuracy of diameter measurements above irregularities using the Close-Range Photogrammetric approach, with diameter tape serving as the reference. A total of 212 trees measured in the north of the Republic of Congo were reconstructed in three dimensions (3D), including 128 trees in semi-deciduous forest and 84 trees in evergreen forest. Comparisons were made in terms of dependence (simple linear regression), correlation (Pearson, Kendall, and Spearman tests), agreement (Bland and Altman method), and difference (Mean Absolute Error - MAE, Root Mean Square Error - RMSE, bias - BIAS, and coefficient of variation - CV). In addition to a near perfect match, a strong association of diameter measurements and a good degree of agreement, the results indicated the presence of differences between diameter measurement approaches in semi-deciduous forest (MAE = 9.25 cm, RMSE = 16.95 cm, BIAS = 7.45 cm) and evergreen forest (MAE = 3.88 cm, RMSE = 8.47 cm, BIAS = 2.37 cm). These differences are minor in the evergreen forest. The magnitude of the differences found is mostly due to the size of the large-diameter classes. In addition, the coefficients of variation (CV) of diameter obtained from the Close-Range Photogrammetric approach were lower than those obtained from the classic conventional approach in both forests, indicating the higher accuracy of the former approach. Further studies could use larger data samples to provide more accurate estimates and verify the limits of these applications’ measurement capabilities.

Code and data availability

The paper describes photogrammetric tree diameter measurements (212 trees, Republic of Congo) and R-based cross-section analysis, but no public phenotype dataset, image repository, or author analysis code is deposited. Supplementary material (Tab. S1-S4, Fig. S1-S6) is only statistical tables/figures, not raw data or a

No evidence-backed public reproduction asset is currently recorded.