The dominant species are Rhizophora stylosa Griff. and Bruguiera gymnorrhiza (L.) Lamk in the mangrove forest in Okinawa. Measurement of seven stand parameters, tree hight, diameter at breast hight, stem volume, crown area, stand volume, mixture ratio of R. stylosa and stand density for mangrove forests was carried out at fourteen plots in Ishigaki Island and Iriomote Island. A matrix of correlation coefficient among these stand parameters and CCT count of each band of Thematic Mapper (TM) data of Landsat 5 was obtained. For band 1, 2 and 3, the correlation coefficients were not so high although negative correlation was found among each mean CCT count and the mean for five parameters to express the size or volume of a tree. For band 4, there was significantly high correlation among the mean of same parameters mentioned above and mean CCT count, and negative correlation was found for the mixture ratio of R. stylosa and stand density. For band 5, the similar correlation among the parameters and CCT count was regarded as apparent correlation although slightly high correlation coefficients were obtained. In this area, it is observed that R. stylosa is a dominant species on the seaward side and B. gymnorrhiza is a dominant on the landward side in the typical natural mangrove forest. R. stylosa is a kind of pioneer and B. gymnorrhiza is a maxmum species, so the former forms relatively young and poor stand and the latter does old and rich stand. The other hand, the seaward stand on the low level sediments shows small value of CCT count and the landward stand on the high level sediments does larger value. It was considered that this correspondence was a reason of the apparent correlation. Although the stand parameters had not significantly high correlation with the data of this band, correlation coefficients were risen with separation of the data between two islands. One reason of this matter is considered that the tide in Iriomote Island was lower 9 cm for the level above the mean sea level than Ishigaki Island. For band 7, any correlation was not found among the parameters and the CCT count. It was made clear that good estimations of mangrove forest stand parameters were possible with CCT count of band 4. For the general estimation, it is remained to normalize CCT counts of this band with location, season and observation time.
In this study, we have analyzed a subset of the airborne POLDER data over lands obtained in the socalled La Crau campaign conducted by Centre National d' Etudes Spatiales (CNES), France in 1990. The airborne POLDER image data was converted into the surface albedo image data by using our atmospheric correction method, in which the multiple scattering were taken into account for. Then the accurate Bidirectional Reflectance Distribution Function (BRDF) for selected surface covers were estimated. The estimated BRDF for “Forest1”, that was one of vegetated surface covers, shows Lambert's law at 550 nm and 650 nm. But at 850 nm, it is shown that an anisotropic reflection law was valid for the surface and fitting parameters for the BRDF are presented. And we calculated the normalized vegetation index (NDVI) for “Forest1” based on the estimated surface albedos and we found that the NDVI varies significantly with the change of viewing angles in the principal plane.
The studies on merging satellite images has been done and some new methods were suggested. Especially, images merged multispectral images such as Landsat/TM data with SPOT-PA data are available for remote sensing application for example land cover classification and researching land use. Most popular method for merging satellite images is a technique by using HSI transformation. We actually attempted to merge images by using this method, it caused some problems. In this paper, we pay attention to intensity of satellite images to solve these problems. Two methods, which were improving methods for merging images and named “Intensity Preserved Method” and “Intensity Adjusted Method”, were compared and described, we have obtained an effective result for both methods.
An effective automated building extraction method by integrating both the feature-based and area-based stereo matching is addressed in this paper. This new technique aims to improve the efficiency and reliability of the building extraction method introduced in vol.34, no.5 of this magazine writen by the authors, in which 3D regions based building extraction model was proposed. In this paper, three other building extraction models based on regions, 3D lines and the results of area-based stereo matching are put forward besides former 3D region based model.