1 Introduction
Trees are possible to live more than hundreds of years old, and accurate their tree-ring analysis of the stem disk provides tree-ring chronological information, such as long-term tree growth environment and weather (e.g.,Fritts (2012)). In ordinary tree ring analysis, core or bar samples of tree trunks have been used to measure tree ring width.In some cases, it is difficult to distinguish tree rings without observing the tree rings on the entire disk, such as false tree rings and incomplete tree rings (Latte et al. 2015). Tree trunks grow radially around the pith and form tree rings, but the ring width is not the same in all directions(Latte et al. 2015; Shi et al. 2015). When trying to evaluate tree growth by stem analysis, it is possible to record the tree ring shape and accurately estimate the amount of tree growth from the tree ring area (Cerda et al. 2007).
It is necessary to record the tree ring shape, and much labor is required. For woods of conifer plantations, tree ring area by image processing have been studied Soille and Misson (2001). However, in the case of natural trees, the tree ring shape is more complicated, and their width are often very narrow (Schweingruber 2012).It is difficult for humans to make automatic recognition at present. It is practically difficult to manually input all the tree ring shapes of natural woods with hundreds annual tree rings. However, the tree ring shape has a certain regularity (Shi et al. 2015), and it is possible to estimate the shape between the tree rings fairly accurately by inputting some representative shapes at a certain interval and complementing the model there is a possibility.
In recent years, the resolution of photographs taken by digital cameras has increased, and a method of capturing the shape of the tree ring of a large disk using a divided photograph, synthesizing the image by geometric correction of the image, and analyzing the tree ring has begun to be studied. Latte st al.(2015) proposed a method of tree-ring analysis of large disks using high-resolution photographs and GIS software using the R program. R is a highly versatile scientific computing software that can handle images and large amounts of data, and has a statistical package specialized in various research fields. Many analysis and database platforms have also been developed and spread in the field of tree ring chronology (Bunn 2008; Hietz 2011).
This packageTreeRingShape (Ishida 2024) is designed for efficiently recording the shape of all tree rings in a tree trunk cross-section. It interpolates the tree rings () between representative tree ring lines () from the tree ring points () and representative tree ring lines inputted on radii measured in Qgis(QGIS Development Team 2009), free GIS software. A refers to the tree rings that characterize the shape of the outer circumference and the cross-section of the disc, requiring fewer rings the closer their shape is to a perfect circle. Although tree rings grow concentrically around the pith, their shape is not perfectly circular. Traditionally, dendrochronology and tree-ring analysis have evaluated tree growth based on the width of the tree rings. However, with the advancement of image analysis technologies, there has been an increasing number of studies evaluating growth based on the area of the tree rings. Most of these studies have focused on conifer plantation samples, where the tree rings are clearly defined. However, this package also enables the recording of tree ring shapes in trees from natural forests, where the ring shapes are more complex.
2 Interpolation model
| Data | Slot | R | Qgis |
|---|---|---|---|
| Input | P | data frame of Tree Ring Points | Points shape, field(id,ring) |
| P_filename | shape file name of tree ring points | *.shp | |
| P_id.tag | column name of id in shape file (P) | default ‘id’ | |
| P_ring.tag | column name of ring no.(ordinaly year,outermost=0) | default ‘ring’ | |
| L | list of Representative Tree Ring Lines | Lines shape, field(ring) | |
| L_filename | shape file name for representative tree ring lines (L) | *.shp | |
| L_ring.tag | column name of ring no.(ordinaly year,outermost=0) in shape file (L) | default ‘ring’ | |
| L2_filename | file name of shape file (L2) for tree ring lines interpolated | *.shp | |
| Append | P00 | x,y coordinates c(px00,py00) of tree ring center point id==0 in (P) | |
| n_id | number of radial measurement points | ||
| YR_P | total number of tree rings | ||
| L_ | data frame of x,y coordinates of representative tree rings | ||
| YR_L | cumulative tree rings number(year) from 0 (cambium layer) | ||
| ln | total number of representative tree rings, length(L) | ||
| n_YR | total number of representative + interpolated tree rings | ||
| L2 | list of All Tree Ring Lines including interpolated lines | Lines shape, field(ring) |
Figure 1: Function execution flow to make whole tree ring shapes. TreeRingShape() includes the procedure from new_classTreeRingShape() to TreeRingInterpolation().
In this study, the orthogonal coordinates with the pith as the origin are called “annulus coordinates”. From the cambium to the medulla, the ring coordinates of the 16 directions of radiation, cracks, and large changes in tree rings were set as XYring, XYp, XYcrack, and XYskew, respectively (Equation 1). The central angle (Equation 2) and the distance {D} from the medulla were calculated for all these ring coordinates XY (Equation \(\ref{eq:XY}\)).
\[\begin{align} XY=\{XY_{RING},XY_{CRACK},XY_P\} \label{eq:XY} \\ {\theta}=atan2(Y,X) \label{eq:atan2} \\ D=\sqrt{X^2+Y^2} \label{eq:sqrt} \\ {\Delta}D= Dring _{i’, j} -Dring _{i’+1, j} \label{eq:D} \\ rWi= (D _{i, j} -D _{i’, j})/{\Delta}D \label{eq:W} \end{align}\]
The shape of the tree rings is not a perfect circle but is distorted, but the tree rings between a certain range of tree rings tend to be parallel (photograph, figure). The relative tree ring width rW (formula) between the two tree rings, with the inner ring near the pith as 0 and the outer ring as 1, was calculated from the central angles of P coordinates. The spline curve was applied to the relationship between the central angle and the relative tree ring width, and the relative tree ring width at an arbitrary central angle was estimated. In applying the spline curve, a margin of 20 degrees was provided at the center angle for smoothing processing of the terminal part, and the margin was set in the range of -200 to 200 degrees. The coordinate value of the complementary tree ring was estimated by adding the difference D width estimated from the relative annulus width to the distance from the pith of the inner tree ring.
3 Procedures
3.1 Processing the Disc
Harvest the tree disc, avoiding areas with branches, rot, or cracks as much as possible.
Polish the disc until the tree rings are clearly visible.
Draw lines from the pith in generally eight directions with a pencil. These lines are referred to as radii measurement lines.
Starting from the cambium layer (the outermost layer of wood touching the bark) as zero, count the tree rings towards the inside and mark them.
Ensure that the number of tree rings matches across all radii measurement lines.
Trace the lines of representative tree rings with a pencil. This improves work efficiency, especially in areas where there are many rings or the rings are unclear.
3.2 Scanning the Entire Disc Image
- Scan the entire disc image at 1200 dpi or higher using a flatbed scanner or a high-resolution digital camera capable of geometric correction.
3.3 Inputting tree Ring Coordinates Using 'Qgis'
Project Properties Set the project’s to ‘none’. Leave the default blank.
Adding Layers
Import the tree rings raster image ().
Create a new shapefile layer for tree ring points () with integer attributes for and .
Create a new shapefile layer for tree ring lines () with an integer attribute for .
Manually Inputting tree Ring Coordinates
- tree Ring Points is the number of the radii measurement line or the correction point group number, and ring is typically the count of the tree rings from the cambium layer inward. Mark all tree rings along the radii measurement lines. After completing the tree ring interpolation, if there is any discrepancy in the rings, add correction points where the rings must pass.
- Representative tree Ring Lines Input the coordinates while tracing the lines of tree rings that characterize the shape of the outer circumference and the disc. The count starts from the cambium layer as zero and goes inward.
4 Examples of analysis
4.1 A tree disk sampled
Figure 2: The sample of tree disk was collected at Mimatsu (1980 m above sea level) in Chubu Sangaku National Park, North Alps in central Japan.This tree was growing just 10m below the Alpen bus road, and fallen in 2018 by typhoon.
Figure 3: Inputting TreeRingPoints and TreeRingLines using the free GIS software Qgis.
The sample of tree disk was collected at Mimatsu (1980 m above sea level) in Chubu Sangaku National Park, North Alps in central Japan.This tree was growing just 10m below the Alpen bus road, and fallen in 2018 by tyhoon(Figure 2).
The function establishes the class classTreeRingShape(Table 1 ) and executes the interpolation of tree rings to complete the tree ring shape data. Shapefiles operated by R package sf (Pebesma 2018).
Figure 4: Inputting TreeRingPoints and TreeRingLines using the free GIS software Qgis.
Figure 5: Inputting TreeRingPoints and TreeRingLines using the free GIS software Qgis.
Figure 6: Inputting TreeRingPoints and TreeRingLines using the free GIS software Qgis.
Figure 7: Inputting TreeRingPoints and TreeRingLines using the free GIS software Qgis.
Figure 8: Inputting TreeRingPoints and TreeRingLines using the free GIS software Qgis.
5 Coclusion
It should be noted, however, that currently, information regarding the shape of the tree rings needs to be manually inputted, which can be labor-intensive.
6 Acknowledgements
I am grateful to the graduates of Gifu University: Yuta Nakano, Shintaro Nakayama, Akari Kitamura, Ryuhei Matsuda, and Tomoko Onodera, for their assistance with tree ring analysis using the ‘TreeRingShape’ scripts. My thanks also extend to Kenji Ogawa, Sio Fukunaga, Yusuke Hanada, Atsuya Yamamoto, Huricha, Moe Miura, Satoko Maeda, as well as Haruki Nakashima and Tetsu Ohmiya from Toyama Prefecture for their help with tree disk sampling. This research was partially funded by the Toyama Prefectural Government and the JSPS KAKENHI Grant Numbers JP23570030 and JP19H04281.