Recent study has pointed out that the wetlands are the major emission source of methane, which is one of the green house gases. To produce the wetland vegetation distribution map with high spatial resolution satellite images are urgent necessity for estimating the amount of methane emission. As an initial classification step, optimal observing seasons and combinations must be selected to improve the accuracy of classification and reducing the data volume. In this study, effective combinations of observing season and band number were selected to produce a wetland distribution map with multispectral and panchromatic satellite sensor data. In the process of determine, test data were classified by maximum likelihood method by changing every combinations of spectral bands from every 6 season's and combined two seasonal data. The best classification accuracy was attained by using 10 bands of combined TM data from July and August which are middle and final growth stages of wetland vegetation.
In this paper, we described an approach for 3D object reconstruction using simulated laser range finder (LRF) data. For the purpose of automatic registration of multiple views of LRF data without apriori knowledge of the views, such as location and direction, planar faces are extracted from the LRF data, and by automatically identifying the corresponding pairs of planar faces, the spatial relationships (the rotation matrixes and the shift vectors) between the views are estimated. To extract planar faces reliably from erroneous LRF data, the MDL (minimum description length) principle is used. In registrating multiple views, two steps-local and global registration-are used to find transformations to avoid the accumulation of registration errors between all the views.
This paper describes an applicability of morphological operations for slope failure detection using low resolution remote sensing imageries. The distribution of remote sensing data on the failure area is formulated using diffusion equation based on the distribution of collapsed soils. The equation is presented by gray-scale morphological operations. The method is applied to the detection of slope failures triggered by the Hyogo-ken Nambu Earthquake. As the result, the reliable sizes, shapes and morphological types of the failures are computed with the accuracy of 75%. By using the parameters which have used in above case study, the method is also applied to failures caused by heavy rainfall after the earthquake. The results show the applicability of the method for time series monitoring using satellite imageries.
Remotely sensed data is the most useful tool for observing vegetation globally and temporally. It is needed to estimate vegetation amount quantitatively from satellite data by building the indirect model using spectral information and biomass data measured in the field. It is necessary to develop the method to process the field data which is satisfied with condition of the simultaneity and spatial scale for satellite data. The aim of this study is to develop and verify the procedure using field data obtained by moving observation we developed in order to measure the quantitative vegetation amount globally in grassland using satellite data. The conducted study area is Mongolian grassland which is very flat and homogeneous. By using moving observation system, we could obtain so many video images and spectral reflectance at the same time. We analyzed about 300 spectral information, about 300 video images and about 10 biomass data by applying RGB Pattern method we proposed for estimating relationship between vegetation coverage ratio and 4 vegetation indices (NDVI, SAVI (0.5), MSAVI, APVI) . We calculated these vegetation indices considering NOAA AVHRR sensor Ch. 1 and Ch. 2. Finally, we verified relationship between vegetation coverage ratio and vegetation indices, which is estimated by linear regression analysis. The summary of results is shown below 1. we could obtain many field data at the satellite scale. 2. we developed the method which calculate VCR automatically from many video images together.
In this paper, we have made a polarization analysis of the airborne POLDER image data over lands, by introducing the combined model with the atmosphere and ground surface. Then we assume that the surface reflection matrix consists of the diffuse and specular components for a natural surface, using the mixing ratio (α) of specular to diffuse component. We find a simplified Rondeaux-Herman model for “Forest 1” and “Rice field 1”, and a modified Bréon model for “River 1”, are able to explain observed polarization variations against zenith viewing angles in the principal plane. Our conclusions are as follows; 1) the isotropic assumption of LIDF seems to be valid at both 550 nm and 650 nm, 2) and the values of α for “Forest 1” are estimated to be α-0.5 and 0.2<α<0.4 at 550 nm and 650 nm, respectively. Those for “Rice Field 1” are to be α-0.5 and α-0.4 at 550 nm and 650 nm, respectively, and for “River 1” is to be α-0.4 at both 550nm and 650 nm.
To support ground truth observation for remote sensing, a navigation system was developed by using the Real Time Kinematic/On The Fly (RTK/OTF) GPS technique with PC-based GIS. The position of an observer is displayed graphically over an digitized air photograph on a notebook type PC in real time. The distance and direction to the destination are also indicated on the PC. We used this system at Kushiro mire where there are very few landmarks to determine the position before the GPS observation. As a result a lot of observation points were determined in a short time. After observation we can use this system for various type of data management by using database function of GIS.
With the expansion of GPS users, various types of GPS have appeared in the market. Accuracy is very much required for precise surveying, whereas cost performance is more important for many purposes such as car navigation system, field check investigation, etc. The accuracy of “Germin GPS-38”, a handy GPS marketed for recreational use, is examined in this paper. The error ranges for X (longitude), Y (latitude) and Z (altitude) direction are respectively-38m to +52m, -130m to +74m, and -289m to +71m by single GPS. In order to improve the accuracy, we found that the pseudo-differential method is effective. The error ranges of X, Y, Z direction changed to -40m to +49m, -31m to +33m and -66m to +46m respectively. By averaging 10 minutes data at the same point with pseudo-differential method, 14m distance error in X-Y plain is expected, which is acceptable for ground truth investigation for Remote Sensing data.