As one of the effective techniques of CO2 sequestration, a systematic afforestation method at the project level in an arid area has been proposed, and Australia was chosen as the trial site: Appropriate carbon sequestration estimation technique has not been determined in this arid area. Thus, in this research, the stand biomass estimation combined with vegetation classification was examined. A decision tree method was chosen as the vegetation classification method. Vegetation indices (NDVI, SAVI, MSAVI1, MSAVI2, OSAVI) were used for the stand biomass estimation method. The overall accuracy of the vegetation classification result was 89%, and the Khat statistics was 0.86. Therefore, the vegetation classification accuracy was considered to be reliable. Strong correlations (R2=0.88-0.96) between the vegetation indices except MSAVI1 and the stand biomass were observed. Thus, all the vegetation indices except MSAVI1 were considered to be applicable to this research area. However, the stand biomass estimation combined with vegetation classification increased the difference between estimated values and actual values. This was considered to be attributable to the vegetation classification error. Considering the specific circumstances that more than 95% of this research area is bare ground or Acacia aneura woodland, the estimation error was decreased by omitting other classification items.
Unsupervised contextual image classification of land-cover categories is discussed by taking mixel information into account. From the knowledge that most of mixels locate in boundaries of land-cover categories, we first detect edge pixels and remove them from the image to reduce influence of mixels. Then, we make classes by clustering spectrum observed at the remaining pixels which are considered as pure pixels. We here introduce a new measure of spatial adjacency of the classes. The classes are aggregated into categories by the use of the adjacency measure. Further, class-labeling of the pure pixels are updated by a novel technique based on the Markov-Random-Field model of the image. Finally, the mixels are allocated to the categories, and the category-labeling of the mixels are updated similarly. Thus, all the pixels are assigned to one of the land-cover categories. We apply the proposed method to LANDSAT TM images. The method shows an excellent performance compared with the noncontextual maximum likelihood method.
An analyzing method of quasi-monochromatic waves is proposed. Some periods are selected within the band width and wavelet transformations are operated on objective time sequential data using the fixed scale parameters corresponding to the selected periods. Transformations are repeatedly operated to emphasize the distributions of signal intensities every selected periods on time axis. Availability of this method is confirmed by test data and this method is actually applied into GMS VISSR IR time sequential data to know the details of climatic behavior regarding the waves around the equator. The results show that the method is useful and succeeds to emphasize the distributions of the objective periods out of quasi-monochromatic waves.
The photochemical reflectance index (PRI) derived from narrow band reflectance centered at 531 and 570 nm is related to the light use efficiency (LUE) of photosynthesis by terrestrial vegetation. Thus, there is a possibility that LUE can be estimated from the remotely sensed data via PRI. However, the satellite sensor that has these two spectral bands has not existed so far. Therefore, it is necessary to substitute PRI using spectral bands of current satellite sensors. In this study, first we investigated the effectiveness of alternative indices of PRI by the band combination and the multiple regression analyses that use the MODIS land and ocean band reflectances simulated by ground observation data. In the band combination analysis, it was possible to substitute PRI when each growth stage of vegetation was separately analyzed. But, it was difficult to substitute PRI by a single equation in all growth stages. In the multiple regression analysis, it was possible to substitute PRI using the logarithm of MODIS land and ocean band reflectances even when all growth stages were analyzed at the same time. Second, we verified accuracy of alternative index by the multiple regression analysis by applying ground data acquired in different years and sites. It was found that the error of PRI was 0.0112 (RMSE) and this corresponds to 10% error of LUE.
Sea ice monitoring in Lutzow-Holm Bay by using European Remote sensing Satellite-2 (ERS-2) data was conducted during the wintering period of the 43rd. Japanese Antarctic Research Expedition. ERS-2 Synthetic Aperture Radar (SAR) data were directly received and processed at the Syowa Station. Ground truth data were acquired almost simultaneously with ERS-2 observations. This paper describes the time-dependent change of backscattering coefficients for various sea-ice types found in Lützow-Holm Bay by using ERS-2 SAR data acquired in 2002 and 2003. The mechanisms which cause various backscattering change of sea-ice were investigated from the ground truth data.
In this study, spatio-temporal patterns of continuous paddy fields were examined using the patterns observed in metrics calculated for five years of MODIS over East Asia. Firstly, discrimination scores were computed to determine optimal images among seven types of MODIS derived metrics in channel 1, 2 and NDVI supplemented by ASTER classification result. The given scores were stable over five years to distinguish paddy from other land cover category. Then fractional coverage of paddy field were mapped over five years and they resulted in the consistency with the training data derived from ASTER. The derived metrics and paddy fields fractional coverage maps were not sensitive to year to year or the annual cycle and can limit the inclusion of atmospheric contamination. Finally our MODIS product were compared with the past efiorts on continental-scale land cover monitoring with AVHRR and MODIS sensors, and statistics by IRRI showed that the others' estimation except our product overestimate the paddy fields covers. The un-mixinng technique with difierent spatial resolution sensors played a crucial role in depicting the paddy fields cover.