Estimation of Leaf Area Index Using Downward and Upward Looking Digital Cameras in a Deciduous Broadleaf Forest
Abstract
Monitoring the distribution and changes of leaf area index (LAI) is important for assessing growth of a forest ecosystem. However, it is difficult and time consuming to directly measure LAI. In this study, we suggest an indirect method to calculate the LAI based on the analyses of digital spectral image from the Phenological Eyes Network (PEN) system which consists of Automatic-capturing Digital Fisheye Camera (ADFC) and Hemi-Spherical Spectroradiometer (HSSR). Our main purpose is to develop indirect methods for estimating LAI using either upward or downward ADFC without other ancillary field observation. In developing stage, we used field measured LAI by LAI-2000 plant canopy analyzer (PCA, LI-Cor.), two ADFCs and Hemiview software. The ADFC is a set of Nikon coolpix 4500 camera and FC-E8 fisheye lens and it automatically capture downward and upward canopy in hourly interval. The downward ADFC was used to calculate various vegetation indices through RGB analysis. Meanwhile, the upward ADFC was used to estimate LAI using the Hemiview software. Threshold value of the Hemiview is important to separate the leaves and background such as sky, wood, edge on digital image. In other to decide accurate threshold value of the Hemiview, we performed that comparison of field measured LAI measured and the Hemiview LAI using upward ADFC digital image. Based on the determined threshold value, an objective method to recognize peculiar patterns of RGB histogram around the threshold was developed and applied to estimate LAI from upward ADFC images only. As well, two spectral indices (i.e. G/R ratio and 2G-RB) were calculated from the downward ADFC images. The relations between the spectral indices and LAI time series from the upward ADFC images were investigated and regression models were developed. The regression models were utilized to reconstruct seasonal LAI variation from the downward ADFC images only. Both field-measured and upward ADFC-derived LAIs showed good agreement (R2 = 0.98). Our method on Hemiview parameterization based on RGB histogram and the developed regression models enable to estimate seasonal LAI variation by using only either upward or downward ADFC images. The methodology also could be applied to develop similar detection methods for webcam measurement.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2010
- Bibcode:
- 2010AGUFM.B41C0321C
- Keywords:
-
- 0476 BIOGEOSCIENCES / Plant ecology