The objective of this study was to obtain the fundamental results for forest tree height estimation at stand level by using visible and near infra red (NIR) spectra analysis in the 400-1000 nm region and tree height data. This study was carried out at the Takakuma Experimental Forest located on the Ohsumi island of the Kagoshima prefecture. The original absorbance spectra and their second derivative spectra in the 400-1000 nm region were obtained; bands around 492, 581, and 673 nm that may be assigned to chlorophyll absorption were observed in the 400-700 nm region, and a small peak around 947 nm may be assigned to the second overtone stretching mode of the intra-molecular hydrogen-bonded OH group. The intensities of the inflection point of bands at 492 and 673 nm observed by the second derivative spectra showed a linear increase with a high correlation corresponding to the increase of tree height. The score of the first component (PC1) obtained by Principal Component Analysis (PCA) also showed a linear increase with tree height change. All of the expected absorptions were observed by the loadings of PC1. For tree height estimation, a simple linear regression, ratio regression, and Partial Least-Squares (PLS) regression using feature bands identified by second derivative were calculated. In the case of a simple linear regression using a band at 673 nm and PLS regression using a band in the 644-699 nm region, a coefficient of determination (R2) of more than 0.8 and a root mean square error (RMSE) of less than 5.1 were obtained, respectively. The results obtained in the present study have demonstrated that the visible and NIR spectral characteristics, especially the chlorophyll absorption band, combined with spectral analysis are powerful tools for the estimation of tree height.
- Chemical band assignment
- Peak intensity
- Spectral analysis
- Tree height
- Visible and near infra red spectra