Wet atmospheric propagation path delay is one of the most important measurement error factors in repeat-pass Interferometric SAR (INSAR) technique for wet regions such as Japan. Its influence can not be neglected for the measurement accuracy of several meters in elevation and several centimeters in surface deformation. The theoretical analysis of height and displacement error by atmospheric phase change for JERS-1/SAR data has showed some severe error even for slight atmospheric inhomogeneity. The simultaneous observation by water vapor radiometer (WVR) at GPS stations in Izu district for about a year has been carried out. The comparison between GPS and WVR wet delay has induced that correlation coefficient was 0.99 and r. m. s. wet delay was about 1.1cm except for rainy day. It was showed that the wet delay information of GPS network would be useful and efficient in the quantitative evaluation of wet atmospheric delay effect. In order to evaluate wet delay effect appeared on INSAR data, we have applied GPS network data to actual phase of two pairs of differential SAR interferogram. The result has concluded that INSAR residual phase change was successfully extracted by using high coherent interferogram but was not very accurately evaluated by GPS delay due to the serious noise factors.
There are two parameter tuning algorithms, time update and measurement update algorithms for parameter estimation of Kalman filter. Two learning methods for parameter estimation of Kalman filter are proposed based on RLS (Recursive Least Square) method. One is the method without measurement update algorithm (RLS1) . The other one is the method without both time and measurement update algorithms (RLS2) . The methods are applied to the time series data of DMSP/SSM/I data with a plenty of missing data. It is found that the proposed RLS2 method shows smooth and fast convergence in learing process in comparison to the RLS1.
This paper describes a new method to remove high frequency fluctuations in timeseries normalized difference vegetation index (NDVI) of NOAA AVHRR. Time series NDVI of AVHRR are frequently used for global or continental land cover monitoring. However, NDVI values have undesirable noises due to cloud or atmospheric effect and signal noises. These effects cause high frequency fluctuations in time series NDVI. In order to remove these fluctuations, several methods such as Maximum Value Composite (MVC) method, Cloud screening method, J. Cihlar's method, and Maximum Value Interpolated (MVI) method have been developed. Since none of them provides satisfactory result, authors developed a new method, the Temporal Window Operation (TWO) method, main part of which is the removal of the effect of clouds to make a smooth temporal NDVI change. The TWO method also includes the removal of high value noises.
This study examines some potential applications for an urban area interpretation using multisensor data. We can get different satellite images for a single test site. The study area was investigated using remote sensed imagery, such as DD-5, IRS-1C, and Landsat TM data. The objective of this study is to propose data mining and data fusion technique for the interpretation of urban area. The methods are used IHS transform and proposed methods (3S and 3H) . The spectral information of Landsat TM is extracted by the proposed data mining techniques. Then, this extracted spectral information is fused, by Brovey transform with high resolution remotely sensed data such as IRS-1C and DD 5. The conclusion is that data mining and data fusion has great potential to improve the performance of detailed urban area analysis using multisensor and multiresolution imagery. Fused images allowed for continual monitoring of infrastructure needs. The finer spatial resolution of space imagery will greatly enhance monitoring of urban area.