Handling Remote Sensing Data Using Toolboxes and Open-Source Platforms
Remote sensing applications pan across various domains, like disaster management, climate studies, socio-economic analysis, geology, astronomy and astrophysics, and many more. A far greater number of satellites are orbiting the Earth than ever. These satellites inspect many different areas of the Earth, as a result of which a large amount of scientific data is downlinked on the ground stations every day. Scientists, researchers, and agencies are always trying to make full use of this data. But this analysis is quite challenging due to the high quantity of data.
Nowadays, remote sensing data is captured in various wavelengths (or electromagnetic spectrum), and it is made freely available to the end-user. As a result, the use of remote sensing data has increased more than ever. Using remote sensing data, we can complete surveys that pan over a large area quickly. Remote sensing data is traditionally analyzed using software developed by various space agencies and developers. Some remote sensing data processing toolboxes have been enlisted below:
- SNAP - Brockmann Consult, SkyWatch and C-S and ESA (Download)
- QGIS - Gary Sherman, currently maintained by volunteer developers (Download)
- SAGA GIS - J.Böhner, O.Conrad, R.Köthe and A.Ringeler (Website) (Download)
- GRASS - GRASS Development Team (Download)
- PolSARPro - The Institute of Electronics and Telecommunications of Rennes (France) and ESA (Download)
- Whitebox GAT - John Lindsay and Anthony Francioni (Website)
Active and passive remote sensing provides complementary information to each other, filling the necessary gap, enabling broad applicability of the remote sensing data. Here, we're going to discuss a few applications where the active remote sensing data provide valuable insights.
Land deformation: Synthetic Aperture Radar (SAR)is a technique wherein the high resolution (in a few meters) earth observation data is captured by synthesizing a long antenna using a small physical antenna element. This fleet is achieved with indigenous signal processing and computing. Interferometric SAR (InSAR) uses two images for land deformation mapping, and with consecutive temporal pairs, the deformation in cm/year is estimated. This technique is called Differential ISAR (DInSAR).
The information provided by DInSAR is crucial for monitoring natural hazards like earthquakes, volcanoes, landslides and subsidence. The InSAR can be only implemented on a single look complex (SLC) image; it consists of amplitude (strength of radar backscatter) and phase (fraction of one sine wave cycle) measurement. Phase is a function of the distance between the sensor (onboard satellite) and ground targets.
The interferogram is generated by taking the difference between the phases of the two coregistered SAR imagery, which correlates to the terrain's topography. In DInSAR, the topographies effect is removed with the digital elevation model (DEM), and the phase change then represents any surface movement between the two dates.
For this case study, two SLC Sentinel-1 images acquired over Mexico City are considered for deformation mapping between 6 June 2016 and 10 September 2016. Figure 1 shows the satellite footprint. The standard InSAR flow for Sentinel-1 IW data consists of coregistration, interferogram formation (wrapped phase and coherence), TOPS deburst, Goldstein phase filtering, phase unwrapping (snaphu), phase to displacement, and terrain correction. A detailed description of the processing chain is available here.
Figure 1. Sentinel-1 IW footprint over Mexico City (More information on Sentinel-1 data specification is available here.
The output of the DInSAR processing chain is shown in Figure 2. The surface with no movement is shown in white, land subsidence is shown in red and land uplift is depicted by blue. Detecting changes at a cm-scale is a pretty impressive fleet in remote sensing for a satellite orbiting at an altitude of 693 km. The color bar in Figure 2 shows surface movement in meters. If converted to cm scale, the land displacement map now shows maximum surface displacement of 4.2 cm and 8.1 cm, towards and away from the sensor, respectively.
Figure 2. Land displacement (in meters) created using SNAP toolbox. Positive values (blue) show land upliftment, and negative values (red) show land deformation.
Mapping hidden routes: InSAR coherence is a measure of similarity between the pixels of two images. Between the two SAR images (captured some time apart), any change related to roughness, dielectric properties, and movement of the target/surface is captured as a decorrelation in InSAR coherence. Generally, normalized coherence (0 to 1) is used for change detection, where 0 represents no similarity between the scenes (change in target properties) and 1 depicts a stable target with no change. Natural surfaces usually have coherences between 0 and 1 and can be classified or mapped using InSAR coherence.
Figure 3. The animation of true-color composite image (A), InSAR coherence (B), and overlayed hidden routes (C) highlighted by yellow and magenta lines.
If you like our technical articles on Remote Sensing and would like to collaborate with our team to publish your articles on such topics, feel free to write to us at email@example.com.