9:30 AM - 9:45 AM
[HTT19-03] Vegetation indices to express growth heterogeneity in the coastal artificial forest of Japanese black pine on planting bases
Keywords:satellite image, vegetation index, tree planting, embankment materials
The coastal forest of the Japanese black pine in the Sendai Plain was severely damaged by the tsunami caused by the Tohoku-Pacific Ocean Earthquake on March 11, 2011. Reduction in tsunami wave force, the Forestry Agency developed a planting base to manage well growth of planting trees using soil materials brought from the hills of the Sendai Plain. However, poor growth of black pines which has been planted were observed even after several years since their planting. Previous studies have pointed out compaction of the planting base for vertical extension of roots and poor ventilation and drainage as causes of poor growth. However, the growth conditions of the planted Japanese black pine have never been evaluated from the viewpoint of the planting bases. Therefore, it is essential to develop a method to evaluate growth of black pines with high accuracy.
Remote sensing technologies, which can quickly and simultaneously capture spatial images with spectral information covering a wide area, are powerful to evaluate the growth status of planted black pine trees.
In this study, we surveyed 16 plots with different planting years in the coastal forest of Natori City, Miyagi Prefecture, which is one of the target areas of the coastal forest restoration and rehabilitation project. Two multispectral images with different resolutions were applied to obtain NDVI (Normalized Difference Vegetation Index), NDVIre (red-edge Normalized Difference Vegetation Index), and EVI (Enhanced Vegetation Index) followed by calculation to determine the vegetation indices suitable for monitoring the growth of Japanese black pine over a wide area and for each planted tree in the study area.
First, the vegetation indices were calculated and mapped on QGIS3.16.7 using the satellite image of Sentinel-2 taken on June 28, 2021 covering the entire replanted area in the coastal forest of Natori City, Miyagi Prefecture. Then, kernel density estimation was also performed using the calculated values to understand the variation in each vegetation index in the examined plot. In the kernel density distribution, NDVI and EVI showed a similar shape of distribution, but EVI had a larger variance than that of NDVI, indicating relatively high differentiation of vegetation growth by EVI. NDVI showed saturation of values at the points with high canopy cover, while EVI showed no saturation at the same area. NDVIre showed a different distribution with less variances from that of NDVI and EVI. Considering aerosol influences on the satellite imagery, NDVIre of the satellite imagery light scattering by aerosols to reflectance would affect the low response of NDVIre.
Next, we set up five quadrates in the planting plot 2, 11, and 14 in the coastal forest in Natori City, Miyagi Prefecture, and calculated each vegetation index using high-resolution multispectral images taken by a MicaSense multispectral camera RedEdge-M (Cybernetics Corporation) from June 24 to June 25, 2021.
Although NDVI saturated in the area covered with well-grown canopy, the indices of NDVIre showed wide range of values with smaller deviations. In addition, distribution of NDVIre in the target plot represented the forest floor vegetation better than the spatial map using other indices. This is the special advantage of NDVIre using spectral bands relating to high detection potential for vitality of plants in their growing periods. Since EVI has been developed as an index calculated using spectral data of an satellite image by removal of negative effects by aerosol scattering, the calculated EVI using the multi-spectral camera from a low altitude was insufficient even with lower variances. Therefore, multispectral images taken at low altitudes makes the index underestimation.
As a result by comparison between three spectral indices of NDVI, NDVIre, and EVI using multispectral images with different resolutions, there are advantages on EVI when vegetation is expressed using satellite images. On the other hand, multispectral camera images can provide vegetation distribution well by NDVIre, especially for capturing actual growth of Japanese black pine. In the future, we will conduct a field survey to confirm the distribution of vegetation indices and find any controlling factors for low growth rate of planted black pines.
Remote sensing technologies, which can quickly and simultaneously capture spatial images with spectral information covering a wide area, are powerful to evaluate the growth status of planted black pine trees.
In this study, we surveyed 16 plots with different planting years in the coastal forest of Natori City, Miyagi Prefecture, which is one of the target areas of the coastal forest restoration and rehabilitation project. Two multispectral images with different resolutions were applied to obtain NDVI (Normalized Difference Vegetation Index), NDVIre (red-edge Normalized Difference Vegetation Index), and EVI (Enhanced Vegetation Index) followed by calculation to determine the vegetation indices suitable for monitoring the growth of Japanese black pine over a wide area and for each planted tree in the study area.
First, the vegetation indices were calculated and mapped on QGIS3.16.7 using the satellite image of Sentinel-2 taken on June 28, 2021 covering the entire replanted area in the coastal forest of Natori City, Miyagi Prefecture. Then, kernel density estimation was also performed using the calculated values to understand the variation in each vegetation index in the examined plot. In the kernel density distribution, NDVI and EVI showed a similar shape of distribution, but EVI had a larger variance than that of NDVI, indicating relatively high differentiation of vegetation growth by EVI. NDVI showed saturation of values at the points with high canopy cover, while EVI showed no saturation at the same area. NDVIre showed a different distribution with less variances from that of NDVI and EVI. Considering aerosol influences on the satellite imagery, NDVIre of the satellite imagery light scattering by aerosols to reflectance would affect the low response of NDVIre.
Next, we set up five quadrates in the planting plot 2, 11, and 14 in the coastal forest in Natori City, Miyagi Prefecture, and calculated each vegetation index using high-resolution multispectral images taken by a MicaSense multispectral camera RedEdge-M (Cybernetics Corporation) from June 24 to June 25, 2021.
Although NDVI saturated in the area covered with well-grown canopy, the indices of NDVIre showed wide range of values with smaller deviations. In addition, distribution of NDVIre in the target plot represented the forest floor vegetation better than the spatial map using other indices. This is the special advantage of NDVIre using spectral bands relating to high detection potential for vitality of plants in their growing periods. Since EVI has been developed as an index calculated using spectral data of an satellite image by removal of negative effects by aerosol scattering, the calculated EVI using the multi-spectral camera from a low altitude was insufficient even with lower variances. Therefore, multispectral images taken at low altitudes makes the index underestimation.
As a result by comparison between three spectral indices of NDVI, NDVIre, and EVI using multispectral images with different resolutions, there are advantages on EVI when vegetation is expressed using satellite images. On the other hand, multispectral camera images can provide vegetation distribution well by NDVIre, especially for capturing actual growth of Japanese black pine. In the future, we will conduct a field survey to confirm the distribution of vegetation indices and find any controlling factors for low growth rate of planted black pines.