Photogrammetric image analysis: Papers accepted on the basis of peer-reviewed full manuscripts

Stilla, Uwe

Bibliographic data

Multivolume work

Persistent identifier:: 856685461

Author:: Stilla, Uwe

Title:: Photogrammetric image analysis

Sub title:: PIA 07 ; Munich, Germany, September 19 - 21, 2007

Year of publication:: 2007

Place of publication:: Lemmer

Publisher of the original:: GITC

Identifier (digital):: 856685461

Language:: English

Document type:: Multivolume work

Volume

Persistent identifier:: 856685739

Author:: Stilla, Uwe

Title:: Papers accepted on the basis of peer-reviewed full manuscripts

Sub title:: PIA 07; Munich, Germany, September 19 - 21, 2007

Scope:: X, 224 Seiten

Year of publication:: 2007

Place of publication:: Lemmer

Publisher of the original:: GITC

Identifier (digital):: 856685739

Illustration:: Illustrationen, Diagramme

Language:: English

Usage licence:: Attribution 4.0 International (CC BY 4.0)

Publisher of the digital copy:: Technische Informationsbibliothek Hannover

Place of publication of the digital copy:: Hannover

Year of publication of the original:: 2016

Document type:: Volume

Collection:: Earth sciences

Chapter

Title:: ASSESSING THE 3D STRUCTURE OF THE SINGLE CROWNS IN MIXED ALPINE FORESTS. A. Barilotti, F. Sepie, E. Abramo, F. Crosilla

Document type:: Multivolume work

Structure type:: Chapter

In: Stilla U et al (Eds) PIA07. International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, 36 (3/W29A) 133 extracted trees varies meaningfully depending on the structure of the different forestry plots examined. Juvenile forests, with a high population density and a high percentage of small diameters, highlight the difficulty of using laser technology to characterize the population well. In these cases, underestimation is evident in terms of “dominated” trees. On the contrary, the results seem to improve significantly when the forestry plot is mature and mono-storey structured (even-aged). In this case, the percentage of extracted trees reaches high values in coniferous forests (80-92%) as well in broad-leaved forests (83%), meaning that the most interesting part of the forest (from an above ground biomass content point of view) is extracted anyway. The tree height value is calculated using the maximum height of the laser points (apexes). As far as this parameter is concerned, the method does not introduce relevant underestimations, which are possible using different approaches based on rasterized data. 4.2 Crown delineation Table 4 reports the difference between the crown base height values measured on site and those extracted from the laser data. As far as this parameter is concerned, the correlations correspond to the well-extracted trees. In these cases, we isolated and analyzed the base height values connected to the different species within each plot, as the table shows. ID PlotID Species Cnt Min Max Ave SD 1 FOAB Beech 4 -12,77 -0,07 -4,10 5,86 2 FO_A_B Spruce 41 -13,27 11,00 -0,25 5,12 3 MB_A_B Fir 8 -2,29 6,50 2,05 2,96 4 MB_A_B Beech 4 -20,19 10,86 -5,52 12,7 5 MB_A_B Spruce 21 -19,80 8,59 -0,74 7,43 6 MB_C Spruce 17 -4,10 0,41 -1,07 1,28 7 MB_D Fir 14 -16,00 9,15 -2,53 6,49 8 MB_D Spruce 55 -18,57 4,32 -2,84 5,46 9 PR_B_C Spruce 56 -15,93 1,71 -3,70 3,58 10 SAA Beech 65 -9,97 12,03 -0,47 4,51 11 TUABC Beech 79 -19,37 15,50 -5,54 8,65 12 VBA Spruce 31 -15,84 11,32 -0,62 6,93 Table 4 — Summary of the crown base height analysis. The results are reported in terms of different species surveyed within each transect. Excluding from the analysis process those sub-plots where the number of trees per species (Cnt) is insufficient to perform statistical analysis (row 1, 3 and 4 in Table 4), the result show: • The minimum and maximum differences of base height for each forestry transect reach high values, respectively as a result of the difficulty of the laser beam in penetrating the canopy (negative values) and of the presence of outlayers (positive values); • The average values (ave) are always negative, suggesting the tendency of the laser to overestimate the base heights (it follows that the crown depth and the volume are underestimated); • The worse valuation concern those plots characterized by juvenile broad-leaved forests (ave of TU_A_B_C = -5,54), while there is an improvement in mature broad-leaved transect (ave of SAA = -0,47). However, this last area was surveyed in the absence of leaf cover; • The best results correspond to the coniferous transect and don’t seem to be affected by the age of the forest. The average and the standard deviation values for those plots which are statistically significant are shown in graph form in Figure 4. The graph highlights that the average values of crown base height have a small overestimation (negative values) in most transects while the standard deviation has a high range of values. Figure 4 - Graphic visualization of the average and standard deviation of the values reported in Table 4 for those plots where the species number is significant. The same analysis (average difference and standard deviation) was considered from a tree species point of view, without considering the plot characteristics. The values reported in Table 5 underline what can be expected in terms of crown estimation if a vast area of coniferous or broad-leaved forests is surveyed. While coniferous areas (spruce and fir) show little underestimation of the base depth (- 1,95 m < Ave < - 0,87 m), the base depth for beech trees is underestimate more (Ave < - 4 m). ID SPECIES cnt Min Max Ave SD 1 Spruce 227 -19,80 11,32 -1,95 5,34 2 Beech 146 -20,19 10,86 -4,89 5,72 3 Fir 22 -16,00 9,15 -0,87 5,83 4 Pine 5 -13,93 11,51 2,62 9,85 5 Larix 4 -13,04 8,61 -1,05 10,32 6 Map 1 e 3 0,83 8,69 4,01 4,14 7 Ash 2 0,79 2,04 1,42 0,88 Table 5 - Comparison of the crown base values between field surveyed and laser extracted data. The differences are summarized considering the different tree species. Moreover, the application of the methods to multiple pulse surveyed data (cfr Table 2) doesn’t seem to give better results, compared to the first & last data. The application of the re clustering method previously mentioned would help to remove those return pulses due to the presence of low vegetation, like the example in Figure 5. The quality of the crown area values depends substantially on the validity of tree extraction method. The qualitative comparison between the crown values measured on site and the clustered laser data correspond, especially in the case of the dominant vegetation layer. The reliability of the crown parameter estimates generally improves when laser point

Access restriction

Copyright

fullscreen: Papers accepted on the basis of peer-reviewed full manuscripts (Part A)

Multivolume work

Volume

Chapter

Table of contents

Cite and reuse

Volume

Chapter

Image

Image fragment

Citation links

Citation recommendation