Hyperspectral reflectance is an important data source for remote sensing of vegetation and ecosystems, especially in assessment of ecophysiological and physiochemical properties. Nevertheless, major advantages of hyperspectra such as waveband richness, sharpness of wavebands, and spectral continuity have not been well utilized to date. Here, we investigated the ability of two methods 1) normalized difference spectral index (NDSI) using thorough combinations of two wavebands (NDSI [i, j]=[Rj-Ri]/[Rj+Ri] using reflectance values Ri and Rj at i and j nm wavelengths), and 2) partial least squares regression (PLS) with waveband selection based on a case study in rice canopies. Three important rice variables (grain protein content : GPC, chlorophyll content : CHL, and above-ground biomass : AGB) were analyzed with airborne and ground-based hyperspcetral data. Results showed that the predictive-ability map of NDSI was useful for extracting effective wavebands and bandwidth for specific variables. NDSI [970, 570], NDSI [710, 550], and NDSI [710, 630] were selected as the best NDSIs for predicting GPC, CHL and AGB, respectively. The predictive ability of the best NDSIs was much higher than that of NDVI, especially in GPC and AGB, and comparable with those of the multiple linear regression (MLR) using four selected bands or PLS using all bands. The PLS with waveband selection used only 20-50% of all wavebands, but showed much higher predictive ability than the other methods. Number of wavebands selected commonly for the three variables was only 2% and 10-20% for any pairs among the three, whereas 18% was never used. Results suggest that NDSI map and IPLS are useful for advanced use of hyperspectral data, especially in parallel assessment of multiple vegetation/ecosystem variables.

View full abstract

Individual tree crown parameters and tree height are desirable in forest inventory and ecological studies to estimate forest carbon stocks and volume acculately. In this study, a new method to detect model-based individual conifer tree crown and estimate tree height using small footprint Light Detecting And Ranging (LiDAR) raw data is developed. The model-based conifer tree crown has solid geometry form, and can be expressed by a geometric equation which is a function of crown radius, crown height, crown curvature and 3-dimensional tree top position. To estimate crown parameters, LiDAR point clouds which represent tree crown are extracted. Then, tree crown parameters are estimated by hill-climbing method using extracted LiDAR point clouds. Hill-climbing method searches the best fit parameters by changing tree crown parameters iteratively. From the estimated crown parameters, crown region and tree height are estimated. The developed method is applied to a Japanese cedar (Cryptomeria japonica D. Don) plantation. Detected tree crown parameters are reconstructed on 3-dimensional scene. Detected trees are validated with field measurements which are number of detected trees, tree height, position and projected crown area. In total, 83 percent of the field measured trees are correctly detected. However, 17 percent are not detected due to the suppression or proximity. Tree height derived by LiDAR is estimated with root mean square error (RMSE) of 1.37m. Underestimation of tree height is approximately decreased by 1m, because the hill-climbing method estimates 3-dimensional tree top position higher than “nearest” tree top pulse. Projected crown area derived by LiDAR corresponds with field measurement. However, the underestimation can be seen and RMSE is 7.12m². These results show that the developed method is appropriate for detecting tree crown and estimating tree height.

View full abstract

Seaweed are essential primarily producer in coastal seashore. However the seaweed vegetation has decreased in Japan and changing the seashore environment was expected to the barren sea. This study was focused on developing estimating method for the biomass of seaweed in the habitat using remotely sensed data fundamentally.
The spectral reflectance of seaweed was measured with handheld spectral radiometer on varying the depth, i.e. distance from sea surface to the top of seaweed. At the same time, spectral images were taken by the video camera mounted with several bandpass filters. The several wavelengths, corresponded with depth (special wavelength), were identified by differentiating the reflectance curves with wavelength. An index for identifying the seaweed with several depths was estimated in this study. In these calculations, the effects of the surface reflection of the sea were neglected by estimating the uniformly scattering using the Lambert-Beer law.
Studying the relationships between the digital numbers of the composed image by the applied the index for the spectral images, the heights of seaweed were estimated. These algorithms and procedures show the possibility of monitoring seaweed bed in the coastal region.

View full abstract

Toward the reliable estimation of leaf area index (LAI) and fraction of absorbed photosynthetically active radiation (FAPAR), the relationship between LAI/FAPAR and bidirectional reflectance factor (BRF) at the top of canopy should be accurately modeled by the radiation transfer models. These relationships vary with the forest landscape due to its horizontal heterogeneity and needles clumping within shoot. In this study, the effect of the forest heterogeneity on the relationships between BRF and LAI, and NDVI and LAI/FAPAR were examined through the three-dimensional radiative transfer simulation, and were compared with the results from one-dimensional radiative transfer simulation. In addition to the simulation, limitation of one-dimensional radiative transfer simulation was evaluated. The results showed that BRF at red and near infrared, and NDVI had large variations with different forest landscape under the same LAI conditions. However the relationship between NDVI and LAI, and NDVI and FAPAR derived from dense canopy condition were quite similar to the results from one-dimensional model. If we add the shoot clumping effect in one dimensional radiative transfer model as a universal parameter for three-dimensional effect of the forest, one dimensional radiative transfer model can work well for the BRF simulation in spatially heterogeneous landscape except higher LAI conditions.

View full abstract