Near-Surface Characterization of Mars InSight Data by Polarization Analysis

Abstract

On the 25th of November 2018, NASA’s InSight Mars-lander landed at Elysium Planitia on Mars (NASA, 2022a). The lander includes the SEIS instrument, which measures the seismic signals on the planet (NASA, 2022b). Because seismic waves are altered by the medium they travel through from the source to the receiver, they contain information on the interior structure of the planet. This information about the structure of Mars can be extracted by applying different data processing and evaluation techniques.
One of the proposed methods is the polarization analysis of the first-arriving P-wave, which is expected to contain information about the near-surface S-wave velocity structure. The technique uses the polarization of an incoming P-wave in horizontal and radial direction to estimate the apparent P-wave incidence angle (Knapmeyer-Endrun et al., 2018). The measured P-wave incidence angle can further be used to estimate a frequency-dependent S-wave velocity from the first break, given that the ray parameter is known. An inversion of such frequency-dependent S-wave velocity curves can then potentially lead to velocity-depth profiles of the near-surface zone.
A first pre-testing of the polarization analysis on a synthetic model should help to understand the process of frequency-dependent S-wave velocity extraction and to work out the limitations of the method. The synthetic data testing showed, that there are a range of important parameters, which influence the velocity estimation. It is highly frequency-dependent and later arrivals, as well as noise, can affect the accuracy. One of the indicators that can be used to evaluate whether or not the extracted S-wave velocities can be trusted is the ellipticity of the polarized wave, which should not be above a threshold of 0.2. The time delay between the Pand the mode-converted PS-arrival (defined as ∆t_app) could be used to define the areas where the estimated velocities change from their expected value. It was found, that ∆tapp has to be greater than 1.4×1e-4 - 2×1e-4 s, that a velocity estimation of the first layer is possible. This is valid for frequencies between 0.08 Hz up to 6 Hz. The second layer velocity could only be calculated for frequencies lower than 0.6 - 2 Hz for the tested models. Furthermore, a proper velocity extraction is not possible for certain ∆tapp and frequency combinations because the angle estimation in this area can not be calculated properly, so far. However, synthetic tests showed that the velocities of a two-layer model could be extracted for specific first layer thicknesses in a given frequency range and a following basic inversion enabled the extraction of the interface depth between these layers.
The application of the polarization analysis to two Martian events allowed a frequency-dependent Vs-velocity extraction for the event S1222a, as 2451 m/s and 2665 m/s for the frequencies of 0.21 Hz and 1.43 Hz respectively. These values could be further used to compare with velocity extractions from other methods. For the second event S0235b, the ellipticity of the first break was above the 0.2 threshold and did therefore not lead to trustworthy outcomes. The polarization analysis applied to real data showed, that a velocity calculation is challenging but possible under certain conditions. Many parameters like the source and its frequency content should be known to actually invert the data, which is why a further analysis with receiver functions is recommended. For further processing, an extended synthetic data testing for other parameters should be performed.