I started my MSc in Geophysics/Planetary Science in September as a member of the Radar Remote Sensing Research Group supervised by Dr. Catherine Neish. This entry is going to be a quick introduction to my research project and how I am getting started.
The goal of my project is to determine the percentage of known impact craters on Earth that can be recognized with synthetic aperture radar (SAR) data. To date, there are 190 confirmed impact structures in the Earth Impact Database.
The aim is to then use this information to infer the number of impact craters on Titan that may be missing. Given that impact cratering is a common process in our Solar System, the surface of Titan is expected to have thousands of craters which we do not observe.
For today, I am going to focus on the first part: mapping craters on Earth using radar data. I am working with data from two satellites, Sentinel-1 and ALOS PALSAR:
Sentinel-1 is part of an Earth observation programme facilitated by the European Space Agency (ESA). This mission is composed of two satellites, Sentinel-1A (launched in April 2014) and Sentinel-1B (launched in April 2016), that carry a C-band (5.6 cm) SAR. It has four operational modes but the main mode over land is the Interferometric Wide Swath (IW). This mode features 5 x 20 m spatial resolution, a 250 km swath, and offers products in single and dual polarization.
The Advanced Land Observation Satellite (ALOS) is part of the Japanese Earth observing satellite program. The Phased Array type L-band SAR (PALSAR), onboard the ALOS, is an L-band (24 cm) sensor with single, dual and full polarization capabilities. I am mainly trying to gather data from its Fine Resolution Mode (9 x 10 m for single polarization and 19 x 10 m for dual polarization).
I have been mostly reading papers and other texts to understand radar basics (and now re-reading some of them), but I did get a chance to play around with some data this week. I am using the Sentinels Application Platform (SNAP) to process the radar data. Here are some results after working with Barringer Crater (Arizona) Sentinel-data.
I start with radiometric calibration of the intensity data (Fig.2). The calibration corrects the image such that the pixel values represent the radar backscatter. I am looking at the VH intensity band, but would like to compare with VV band results. However, I ran into a memory issue when trying to process VV band, so will have to look into that. Next, I apply the deburst operation which combines the burst data into one single image (Fig.3). Then, multilooking averages over range and/or azimuth pixels resulting in less noise and approximate square pixel spacing (Fig.4). I applied 1-by-4 (1 range, 4 azimuth) multilooking and will further investigate different specifications. The image still appears quite noisy so I applied speckle filtering (Fig.5) which reduces the amount of speckle. The drawback of this function is that it blurs features or reduces resolution. I used the default (Refined Lee) setting in this image, and would like to compare it to the other speckle reduction settings in order to determine the best result. The last part is terrain correction which corrects the SAR geometric distortions using a digital elevation model (DEM) and produces a map projected product (Fig.6). The Barringer Crater is visible in the zoomed-in image (Fig.7). My goal for the rest of the week is to explore the different settings for the operations I’m using in SNAP to see which produce the best results. Then, I can apply them to rest of the crater data. I have also started to play around with ALOS-PALSAR data but ran into troubles at the multilooking phase, will have some results for next time!
In other news…
Researchers have discovered a dwarf planet called 2015 TG387 or “The Goblin” at about 80 AU from the Sun (almost as close as it gets), far beyond the orbit of Pluto. Its strange orbit extends to as far away as 2300 AU which supports the idea of a 7x Earth-size planet at the outer edge of the solar system that has yet to be detected.