With the number of ocean stations is far less than 100, most of sea surface temperature (SST) data of ocean interior come from satellites. Polar orbiting meteorological satellites like NOAA satellites of USA and Fengyun-1 satellites of China, can view any place of the Earth’s surface for at least twice per day if it is cloud free. Geostationary satellites like GOES, GMS and Fengyun-2 monitor most Earth’s surface hourly, or any time at our command. Compared to SST that is observed by many surface buoys, the SST derived from GOES satellites has root mean square difference (RMSD) about 1K, slightly inferior to the 0.7K of SST from NOAA satellites. After stringent filtering to remove spikes and potentially contaminated data, the RMSD of GOES SST may also be improved to 0.7K. The remaining errors are from the skin effect that is controlled by the surface wind speed W and insolation Q. The wind induced error in SST is up to 1.2K when W < 2 m/s. QuikScat data show that the fraction of sea surface where W 2m/s gradually approaches to a constant and it is accounted for in the regression equation that converts satellite-measured radiation into SST. Our study shows that Q-induced SST error in the low latitude is up to 0.6K. This error makes many researchers using nighttime GOES SST only. The GOES SST error may be reduced if the location-dependent bias and the diurnal variation of 0.3K are included in correcting GOES SST. These corrections will greatly enhance the usefulness of daytime GOES SST.

The purpose of this investigation is to generate true orthophotos from high resolution satellite images. The major work includes 4 parts:(1)determinate the transformation between object space and image space,(2)generate the traditional orthophotos using terrain model,(3)correction of relief displacement, and(4)compensate the hidden areas. We use Rational Polynomial Coefficients (RPCs) to determinate the transformation between object space and image space. In the generation of traditional orthophotos, we employ RPCs and Digital Terrain Model (DTM) to rectify tilt displacements and relief displacements for terrain. We, then, compute relief displacements for buildings with Digital Building Model (DBM). In order to avoid double mapping, we detect hidden areas caused by buildings. Due to the satellite’s small field of view, an efficient method for the detection of hidden areas and building rectification will be proposed in this paper. Test areas cover the city of Hsin-chu in northern Taiwan. The test images are from the QuickBird satellite.

Recently, the spatial resolution of earth observation satellites is significantly increased to a few meters. However, it is more difficult to develop automated image algorithms for automated image feature extraction and pattern recognition. In this study, we propose a scheme to extract road information from high resolution satellite images based on watershed segmentation. First, the multi-spectral and panchromatic images are fused by simplified image fusion technique and the mathematical morphology erosion is applied next to decrease the complexity of fused image. Second, the segmentation result is created by applying watershed segmentation on the infrared image generated in previous step. The iteratively patches merging procedure is also performed to prevent from over segmentation. Third, we take multi-spectral information for each segment by averaging the pixel values of corresponding segment area of multi-spectral image. The ISODATA classification algorithm is followed to classify the suitable road patches from the spectral information of each segment. Finally, the mathematical morphology close operation and area threshold are used as a final refinement. In this study, we use panchromatic and multi-spectral images of QuickBird satellite as test dataset. The experiment result shows that the accuracy of generated road objects by proposed scheme is about 70%.

The study is aimed to derive the regional evapotranspiration scheme by using MODIS images based on both Penman-Monteith and surface energy balance methods(SEBAL and S-SEBI). The observations of meteorological stations in Western Taiwan are used to verify with the estimations to suggest the proper method. Then, the influences of evapotranspiration from the meteorological factors are discussed. We have concluded the Penman-Monteith method has the best performance (ME=1.07mm/day, RMSE=1.28mm/day), and is affected by the net radiation. The SEBAL model (ME=1.61mm/day, RMSE=1.91mm/day) is slightly less performance than Penman-Monteith method and is affected by the net radiation. The accuracy of S-SEBI mode is the worst(ME=2.07mm/day, RMSE=2.68mm/day). It is due to the mixed-pixel of the low resolution in MODIS images and leads to the inaccuracy of evaporative fraction. Therefore, we can suggest that Penman-Monteith method is the best method to estimate evapotranspiration with MODIS images. However, we should improve the estimation of the net radiation and adopt high resolution images in the future.

The study area is located in the 1999 Ji-Ji Earthquake disaster area of central TAIWAN. Two times of airborne laser scans were conducted in 2002 and 2005 and, thus, the change of terrain in this period is examined. There have been 3 typhoon events and one torrential rainfall event in this period of time. When the DEM derived by 2002 LiDAR data was compared to the photogrammetric data in 1987, a change of 34.0 m in height and 3,284.76*106 m³in quantity were observed in the Gio-Fen-Erh-San Slide due to the earthquake event. The change of height from 2002 to 2003 is not obvious, which could be due to erosion process is slowly. After Min-Do-Lle typhoon in June, 2004. A significant slide process is observed at the south-west and south areas of Gio-Fen-Erh-San slide, where the maximum difference in depletion is about 20 m. The height difference of accumulation is circa 4~8 meters and an earth volume change of 279,177 cubic meters were estimated at the eastern riverbed.

In Taiwan, the government surveys the rice paddy area from the aerial photography every year.This data will be used in agricultural food policy making and farmer reparation management.However, to make this information (GIS Paddy map) need many manpower and waste many time.Therefore, this research used very high-resolution satellite images (QuickBird) as a source to extract the rice pattern for rice management.In order to reaching this purpose, this study used the Semivariogram texture method to improve classification accuracy for rice category from Quickbird Pan band. On the other hand, we also get paddy edge line by Sobel and Canny algorithms from Pan data. These edge extraction methods can take complete information for rice pattern map, and we developed the image reparation function for edge line data. After that, we integrated these area and edge line information of re-process data to Rice Pattern Extraction Model (RPEM). Finally, the image data has been independently classified using MSS gray levels, MSS add textural information, and MSS add textural information and combined edge data, which showed significantly improved classification accuracy with the “MSS add textural information and combined edge data”. The result showed using the study process can successfully scheme for rice pattern extraction from very high spatial resolution satellite images.

The purpose of this study is to propose an adaptive Kalman filtering algorithm for singular point detection using airborne LIDAR data. Dr. Ping Wang has developed a Kalman filtering algorithm in 2001.To make the algorithm adaptive, a least-square fitting approach is utilized in this paper to automatically judge the surface type of a local DSM, so that the predicting model can be selected automatically and adaptively. The feasibility of the proposed algorithm is tested by using some airborne LIDAR data with different types of topography. The experiments offer some analyses on the detection ability of singular points.

A major earthquake (M L =7.3) occurred near the small town Chi-Chi in the Nantou County, Taiwan, at 1:47 a.m. local time on September 21, 1999 (17:47 GMT on 20 September). The epicenter was 12.5 km northwest the Sun-Moon lake. The focal depth is 8 km. The death toll had exceeded 2100 and was mounting during the first week after the main shock. In this study, we mainly focus on the numerous aftershocks; some of which reach magnitudes of M L = 6.8. In this paper we used 8 scenes (1998/01/01, 1999/01/21, 1999/05/06, 1999/06/10, 1999/07/15, 1999/09/23, 1999/10/28, and 2000/01/06) of ERS2 images and an external DEM data, by the means of two passes SAR differential interferometry, to detect the deformations before and after the Chi-Chi earthquake. The results based on the available image sets show that prior to the earthquake, it has no clear deformation signature in the interferogram and there are slight displacements in the east of Chelungpu fault; and the displacement areas of the two pairs of DINSAR (1999/10/28 – 1999/09/23; 2000/01/06 – 1999/10/28) were at the same place. The former had about 5.6 cm deformation in slant range, while the latter was about 3 cm. These post-seismic deformations may be due to gravity effects, or the theory of elastic rebound.