5:15 PM - 6:45 PM
[AHW17-P07] Estimation of forest light environment using drone LiDAR data in an artificial forest and its effect on forest floor evapotranspiration
Keywords:drone, LiDAR, plantation forest, light environment, Forest floor evapotranspiration
The forested area accounts for 67% of Japan's total land area, of which planted forests account for 40%.
However, due to the decline of the forestry industry in Japan, the lack of forest management in planted forests has led to a deterioration of the forest environment due to canopy closure, which may affect the understory vegetation by blocking precipitation and solar radiation, and reduce groundwater recharge due to increased canopy evaporation. In some cases, intense thinning may reduce canopy evapotranspiration but not increase groundwater recharge due to the relative increase in evapotranspiration on the forest floor. Against this background, it is necessary to quantify the light environment in the forest in order to determine the appropriate amount of thinning and to predict the impact of thinning on the understory vegetation.
Therefore, this study aims to create a model estimating the amount of solar radiation in the forest by utilizing 3D point cloud data measured from above the forest by a drone, and to focus on the effects of solar radiation interception by the canopy and evapotranspiration of the understory vegetation, so that the drone data can be used for forest management and evaluation of the forest environment.
The study site was a cypress plantation plot in FM Karasawa, a TUAT training forest in Sano City, Tochigi Prefecture. The plot was thinned in rows at a thinning rate of 50% in 2011 on a south-facing slope of approximately 30 degrees. Solar radiation meters were installed at 25 grid points at 1-meter intervals in the forest, and measured the actual solar radiation in the forest for approximately one year from June 2022.
In conventional methods, the forest environment has been evaluated by image analysis of all-sky photographs taken with a fisheye lens to determine the amount of solar radiation in the forest and the degree of canopy openness. However, low resolution and lens distortion problems prevent accurate representation of the actual canopy structure, limiting estimation accuracy, requiring a great deal of labor in forest work, and limiting the number of representative points makes comparison across multiple points in the forest difficult.
However, by using drone data, data can be measured simply by flying over the area, eliminating the need for work in the forest. The LiDAR data used in this study is point cloud data with three-dimensional coordinates, which can reproduce the canopy structure itself with high precision.
To estimate solar radiation in the forest, LiDAR data measured by a drone was transformed into polar coordinates and synthesized into an image that looks as if it is looking up from the forest. In order to create the composite image, multiple patterns of sky images were created by adjusting the perspective of the point cloud data, making the points closer to the forest floor larger than the distance from the forest floor, and making the points farther away, such as the top of the tree canopy, smaller. By superimposing the open sky portion of the canopy and the trajectory of the sun on these images, the magnitude of solar radiation in the forest was estimated, and the accuracy of the model was evaluated by comparing it with measured solar radiation in the forest.
As a result, the accuracy of the solar radiation estimation varied depending on the parameter settings of the composite image. In particular, to determine the appropriate parameter settings, the perspective and the spacing between point cloud data were adjusted in multiple ways to reproduce a canopy structure close to the actual one. Based on the composite image, we calculated the daily total amount of solar radiation, and by taking the relative relationship with the solar radiation meter outside the forest, we considered the amount of solar radiation in the forest intercepted by the canopy and the relationship with the evapotranspiration of the understory vegetation by the lysimeter.
The above results show that it is possible to describe the forest environment from drone data, and as a result, it is possible to quantify the light environment to improve the efficiency of forest management.
However, due to the decline of the forestry industry in Japan, the lack of forest management in planted forests has led to a deterioration of the forest environment due to canopy closure, which may affect the understory vegetation by blocking precipitation and solar radiation, and reduce groundwater recharge due to increased canopy evaporation. In some cases, intense thinning may reduce canopy evapotranspiration but not increase groundwater recharge due to the relative increase in evapotranspiration on the forest floor. Against this background, it is necessary to quantify the light environment in the forest in order to determine the appropriate amount of thinning and to predict the impact of thinning on the understory vegetation.
Therefore, this study aims to create a model estimating the amount of solar radiation in the forest by utilizing 3D point cloud data measured from above the forest by a drone, and to focus on the effects of solar radiation interception by the canopy and evapotranspiration of the understory vegetation, so that the drone data can be used for forest management and evaluation of the forest environment.
The study site was a cypress plantation plot in FM Karasawa, a TUAT training forest in Sano City, Tochigi Prefecture. The plot was thinned in rows at a thinning rate of 50% in 2011 on a south-facing slope of approximately 30 degrees. Solar radiation meters were installed at 25 grid points at 1-meter intervals in the forest, and measured the actual solar radiation in the forest for approximately one year from June 2022.
In conventional methods, the forest environment has been evaluated by image analysis of all-sky photographs taken with a fisheye lens to determine the amount of solar radiation in the forest and the degree of canopy openness. However, low resolution and lens distortion problems prevent accurate representation of the actual canopy structure, limiting estimation accuracy, requiring a great deal of labor in forest work, and limiting the number of representative points makes comparison across multiple points in the forest difficult.
However, by using drone data, data can be measured simply by flying over the area, eliminating the need for work in the forest. The LiDAR data used in this study is point cloud data with three-dimensional coordinates, which can reproduce the canopy structure itself with high precision.
To estimate solar radiation in the forest, LiDAR data measured by a drone was transformed into polar coordinates and synthesized into an image that looks as if it is looking up from the forest. In order to create the composite image, multiple patterns of sky images were created by adjusting the perspective of the point cloud data, making the points closer to the forest floor larger than the distance from the forest floor, and making the points farther away, such as the top of the tree canopy, smaller. By superimposing the open sky portion of the canopy and the trajectory of the sun on these images, the magnitude of solar radiation in the forest was estimated, and the accuracy of the model was evaluated by comparing it with measured solar radiation in the forest.
As a result, the accuracy of the solar radiation estimation varied depending on the parameter settings of the composite image. In particular, to determine the appropriate parameter settings, the perspective and the spacing between point cloud data were adjusted in multiple ways to reproduce a canopy structure close to the actual one. Based on the composite image, we calculated the daily total amount of solar radiation, and by taking the relative relationship with the solar radiation meter outside the forest, we considered the amount of solar radiation in the forest intercepted by the canopy and the relationship with the evapotranspiration of the understory vegetation by the lysimeter.
The above results show that it is possible to describe the forest environment from drone data, and as a result, it is possible to quantify the light environment to improve the efficiency of forest management.