Non-destructive monitoring of annual trunk increments by terrestrial structure from motion photogrammetry.

Abstract

Annual trunk increments are essential for short-term analyses of the response of trees to various factors. For instance, based on annual trunk increments, it is possible to develop and calibrate forest growth models. We investigated the possibility of estimating annual trunk increments from the terrestrial structure from motion (SfM) photogrammetry. Obtaining the annual trunk increments of mature trees is challenging due to the relatively small growth of trunks within one year. In our experiment, annual trunk increments were obtained by two conventional methods: measuring tape (perimeter increment) at heights of 0.8, 1.3, and 1.8 m on the trunk and increment borer (diameter increment) at a height of 1.3 m on the trunk. The following tree species were investigated: Fagus sylvatica L. (beech), Quercus petraea (Matt.) Liebl. (oak), Picea abies (L.) H. Karst (spruce), and Abies alba Mill (fir). The annual trunk increments ranged from 0.9 cm to 2.4 cm (tape/perimeter) and from 0.7 mm to 3.1 mm (borer/diameter). The data were collected before- and after-vegetation season, besides the data collection increment borer. When the estimated perimeters from the terrestrial SfM photogrammetry were compared to those obtained using the measuring tape, the root mean square error (RMSE) was 0.25-1.33 cm. The relative RMSE did not exceed 1% for all tree species. No statistically significant differences were found between the annual trunk increments obtained using the measuring tape and terrestrial SfM photogrammetry for beech, spruce, and fir. Only in the case of oak, the difference was statistically significant. Furthermore, the correlation coefficient between the annual trunk increments collected using the increment borer and those derived from terrestrial SfM photogrammetry was positive and equal to 0.6501. Terrestrial SfM photogrammetry is a hardware low-demanding technique that provides accurate three-dimensional data that can, based on our results, even detect small temporal tree trunk changes.