Why applying remote sensing?
As remote sensing sensor can obtain hundreds of spectrum, majority of the ground features can be detected. It is sufficient to recognize different phenomena on topography. Modeling is compulsory to this domain as well. In some extent, remote sensing data can provide significant factors to this modeling procedure. Not only predictive model can be built, but also the model after hazard occurrence.
Basic aspects of spectral data
The fundamental color of a remote sensing image are red, green and blue. Majority of the images are organized in these three colors, and it is named true color image. Hyperspectral image divides the spectrum into more bands. Different hyperspectral sensors can provide divers bands from ten to hundreds. With more bands reflection, more specific ground features can be attained. Hyperspectral remote sensing collect the information consists spatial data, electromagnetic data and spectral data. In this way, it widely applied in oceanic domain, vegetation analysis, agriculture, atmospheric research and geological observation, etc.
Tectonic motion is the main factor of geological hazard. As lithosphere composed by diversity of rock, various of imaging characteristics would present on spectral image. Applied hyperspectral remote sensing technique to such issue, separated of bands represent different rock can be achieved. In this way, an effective detection on tectonic motion would be realized.
Current hyperspectral sensor recognition
So far, hyperspectral image can be gained on various number of platforms. Airborne hyperspectral sensor can collect a high resolution image in both ground resolution and spectral resolution. On the other hand, the expense ratio is high. In contrast, spaceborne hyperspectral sensor can measure high resolution image but in lower expense. Some well known examples of hyperspectral sensor are LANDSAT MSS and ALOS/PALSAR.
A typical hyperspectral sensor consists seven elements. All the elements are displayed as the table below:
|Spectral regions||VIS, NIR, SWIR, MWIR, TIR|
|Number of channels||100 – 200|
|Spectral bandwidths||10 – 20 nm|
|Spatial resolution||2 – 10 m|
|Swath width||60 – 70 degrees|
|Signal to noise ratio (30 degrees SZA, 50% reflectance)||>500:1|
|Operational altitude||2000 – 5000 m AGL|
PALSAR(Phased Array type L-band Synthetic Aperture Radar) is a L band sensor. Due to the large wavelength, it achieves cloud free and all weather observation. The resolutions of this sensor are 2.5 meter in panchromatic imaging and 10 meter in multispectral imaging. The orbit altitude is about 692 kilometer. Moreover, about 98.2 degrees of inclination. All the characteristics meet the measuring requirement in an earthquake disaster. Majority of deformation in earthquake are can be detected on PALSAR image.
Case study of interferogram
The case study is on 12 May 2008, Wenchuan earthquake. The earthquake area coverage is over thousands square kilometer. For the reason, L band ALOS/PALSAR data of six ascending tracks 471 to 476 has been applied to the case. Moreover, ALOS/PALSAR has high sensitivity on horizontal ground displacement. ALOS/PALSAR incidence angle is about 38.7 degree in vertical direction. The decorrelation regions in the interferogram are masked out.
As interferogram displays, based on interferometric analysis, such interferometric image is generated. In the image, the rainbow fringe represents the displacement during and after the earthquake. The green star in the image is the epicenter. From the scale bar of the rainbow fringe and the density of fringe line, a significant deformation of ground is revealed.
Consequently, interferometric data is efficient on analyzing earthquake. It provides realistic ground situation from the interferometric image. Researching on the before-after image some reasonable rescuing activity can be implemented. On the other hand, there are some disadvantages of DInSAR application. One is that interferometric procedure requires a considerable amount of analysis. Data processing duration is long and complicated. Hyperspectral data is organized by large load of layers, an index might be useful to select the proper band of layer. The other one is that the displacement measurement is selective to the interferometric sensor. Only the displacement along the track of satellite can be measured. For this reason, a completed image is based on several times of satellite measurement. As time limitation in earthquake data processing, the duration is delayed.