عنوان

The evaluation of multichannel shortwave infrared (SWIR) optical data from the Japanese Earth Resources Satellite (JERS-1), for geological applications

Number

TLets264859

Title Proper

The evaluation of multichannel shortwave infrared (SWIR) optical data from the Japanese Earth Resources Satellite (JERS-1), for geological applications

General Material Designation

[Thesis]

First Statement of Responsibility

Denniss, Anthony Maurice

Name of Publisher, Distributor, etc.

Open University

Date of Publication, Distribution, etc.

1998

Dissertation or thesis details and type of degree

Thesis (Ph.D.)

Text preceding or following the note

1998

Text of Note

The primary objective of this thesis is to evaluate data from the Japanese Earth Resources Satellite (JERS-1), particularly multi-channel shortwave infrared (SWIR) data. These data are to be evaluated in comparison with Landsat Thematic Mapper (TM) data for geological applications, in particular lithological discrimination in an arid environment (Oman) and volcano monitoring (Lascar, Chile). JERS-1 is the first satellite purposely designed for geological remote sensing. Optical data (OPS) acquired by JERS-1 is of a higher spatial and spectral resolution than that from-Landsat TM. Unfortunately the immediate benefits of this higher resolution are lost due to the low quality of all OPS data. All seven OPS channels are affected by along- and across-track striping, image blur at spectral boundaries, poor dynamic ranges, random speckle and inter-channel misregistration. Poor data quality greatly reduces the geological potential of these data, especially as the most severely corrupted channels are OPS 6 and OPS 7, two of the new narrow SWIR channels. However, despite poor data quality the majority of the noise artefacts associated with these data can be either removed or significantly reduced by processing in the frequency domain. A number of frequency domain techmques have been developed which separate noise features from real image data. These techniques greatly improve the quality of most OPS channels, although OPS 6 and OPS 7 are sometimes beyond recovery. The spectral characteristics of the lithologies comprising the Oman ophiolite complex are used to demonstrate the increased discrimination potential of 'cleaned' multichannel OPS data. High resolution lab spectra of the ophiolite lithologies were degraded to TM and OPS spectral resolutions in order to determine the influence of spectral resolution on lithologie discrimination. The results of this comparison indicated that a lot of discrimination information seen at lab resolution is lost at OPS and TM resolution. This results in the reflectance spectra of some ophiolite lithologies appearing very similar, especially at TM resolution. More discrimination information is preserved at OPS resolution due to the sub-division of the wavelength range equivalent to TM 7 into three narrow discrete channels. Analysis of resampled spectra at OPS resolution shows that OPS 8 conrtbutes new spectral information not discernible at TM resolution. Resampled lab spectra at OPS resolution were used to try to develop effective band combinations and image processing techniques capable of discriminating ophiolite lithologies. Tests of OPS band composites indicated that the best composite was one which combined three bands from spectrally distinct parts of the spectrum. OPS 852 proved to be the most effective OPS composite, discriminating the majority of ophilite lithologies. Analysis of both spectral and image data showed that OPS 6 and OPS 7 are highly correlated for ophiolite lithologies. Composites combining both of these channels were very poor, however combining one channel (usually OPS 6 due to quality) with OPS 8 and OPS 2 produced a reasonable composite (OPS 862). OPS 852 is the most informative composite, and would remain so even if OPS 6 and OPS 7 were of a higher quality. Comparisons with the best TM composite (TM 754 decorrelation stretched) showed that the OPS data discriminated more lithologies and often allowed previously mapped units to be sub-divided. To test the potential of OPS data for volcano monitoring a time series of four OPS scenes, straddling the April 1993 eruption of Lascar (Chile), were analysed. This study showed that the higher spatial resolution of the OPS sensor was capable of detecting smaller thermal features than possible with TM. This resulted in four intra-crater maps of thermal anomolies being produced, showing thermal variations associated with periods of dome growth and collapse. These data also demonstrated their usefulness for mapping the extent of pyroclastic deposits emplaced during this eruption, and also their subsequent rapid erosion. Overall OPS data are shown to offer significant improvements with respect to lithological mapping and volcano monitoring despite their greatly reduced quality compared with Landsat TM. Extensive tests of various band combinations and processing techniques show that some of the most informative images derived of the ophiolite lithologies consist simply of cleaned channels of data combined as false colour composites. The effectiveness and possible advantages of advanced image processing techniques are often negligible due to the rapid degradation of the data when subjected to intensive processing. OPS 8 is the channel mainly responsible for contributing additional new valuable spectral information, that is not available from Landsat TM. The slight increases in spatial and spectral resolution offered by OPS data result in significantly more geological information being discernible from an OPS composite than on the equivalent TM composite.

Remote sensing

Denniss, Anthony Maurice

Open University

Electronic name

p

[Thesis]

276903

a

Y