Two international studies involving researchers from Imperial College London’s Department of Electrical and Electronic Engineering and the Center for Space and Habitability at the University of Bern in Switzerland, used artificial intelligence (AI) to spot a link between the seismic activity caused by meteoroid impacts and "Marsquakes" recorded by NASA’s InSight mission.
The studies, published in the journal Geophysical Research Letters, looked at all meteoroid impacts on Mars near NASA’s InSight lander between December 2018 to December 2022. AI was used to to identify new impacts in tens of thousands of orbital images data captured by the agency’s Mars Reconnaissance Orbiter (MRO) which the team then cross-referenced against seismic data collected by NASA’s InSight lander. Imperial also provided the silicon microseismometers that are part of the payload that made this seismic detection.
Imperial researcher Dr Constantinos Charalambous said: " This freshly mined data is exciting because it reveals that meteoroid impacts on Mars occur about twice as often as previous orbital image surveys estimated.
"We used to think the energy from most seismic events detected by InSight was trapped travelling in the Martian crust. Now, we understand that seismic waves from some impacts can follow a deeper, faster path - call it a seismic highway - through the crust and deep into the mantle, allowing quakes to reach more distant regions of the planet."
Professor Tom Pike from the Imperial team added "These results show the power of looking deeply into multiple datasets from Mars. Without the seismic data, we would not have known where to look for an impact in the orbital images, and without the orbital images, we would not have been able to locate the source of the seismic energy we detected with InSight.
He continued: "The fact that we have been able to identify an impact in a single pixel of MRO’s low-res orbital camera, intended for daily weather monitoring, shows just how essential it is to combine these large datasets. The power and speed of AI means we have been able to find the proverbial needle in the haystack!"
Now the next step for researchers will be to reassess their models of the composition and structure of Mars’ interior to explain how signals can go that deep.
Machine learning helps spot new impacts
A machine learning algorithm developed at the Jet Propulsion Laboratory at CalTech to detect meteoroid impacts on Mars played a key role in these studies.In a matter of hours, the artificial intelligence tool can sift through tens of thousands of black and white images captured by MRO’s Context Camera to find new impacts. Data from the Thermal Emission Imaging System (THEMIS) onboard NASA’s Odyssey orbiter and from the Mars Express High Resolution Stereo Camera (HRSC) was then used to pin down the time they occurred, and ESA’s Colour and Stereo Surface Imaging System ( CaSSIS)’s data was used to verify and accurately characterise all’identified impacts.
Researchers estimate that if they had completed this whole process using the human eye, it would have taken them years to work through the same volume of images at the same level of accuracy.
The team used AI to search for craters within roughly 1,864 miles (3,000 kilometres) of InSight’s location in a particularly quake-prone region known as Cerberus Fossae, hoping to find some that struck while the lander’s seismometer was recording. By comparing before-and-after images over a range of time, they could determine whether a crater was fresh.
The AI found 123 fresh craters to cross-reference with InSight’s data; 49 of those were a potential match with quakes detected by the lander. Dr Charalambous and other seismologists filtered that pool further, identifying a 71-foot-diameter (21.5 meter) impact crater in Cerberus Fossae.
This crater provided a definitive source location that was nearly twice as far away from InSight as scientists would have thought, based on the quake’s seismic energy. This could only be explained by the seismic energy taking a more direct route through the planet’s mantle.
This image shows the aftermath of a rock from space that slammed into Mars in February 2021, causing seismic waves that reached NASA’s InSight spacecraft, located 1 , 640 km away. The impa left a 21-metre diameter crater and dented an area of about 1 , 400 metres. The blast was captured in this image by ESA’s ExoMars Trace Gas Orbiter using its Colour and Stereo Surface Imaging System ( CaSSIS ) . As it is often the case for CaSSIS imagery, this is a false-colour image created using near-infrared, panchromatic and blue channels, as described here. Credit @ESA
Meterorite impacts on Mars versus Earth
Meteoroid impacts and the resulting ’marsquakes’ are one of the main forces shaping the surface of planets like Mars. According to seismic data from NASA’s InSight mission, Mars experiences around 280 to 360 meteoroids creating craters larger than 8 meters in diameter each year. This is around three times as many impacts as on Earth.This is because the red planet is much closer to the asteroid belt and its thinner atmosphere offers far less protection against falling meteoroids - they do not burn up as much or as quickly as they do when they enter Earth’s denser atmosphere. While Mars is only about half the diameter of Earth, its surface has nearly the same area as Earth’s dry land, meaning the area that could be monitored on each planet is roughly comparable.
A new ’big data’ era of planetary science
The more scientists study InSight’s data, the better they become at distinguishing signals originating inside the planet (caused by rocks cracking under heat and pressure) from ones caused by meteor strikes. The impacts will help scientists further refine how they tell these signals apart.Now that researchers have access to this vast bank of data from space probes like NASA’s Insight lander, the same machine learning principle is already being applied in other ways. For example, to identify other landforms on Mars and the Earth’s Moon and increase our understanding of their geology and history.
This new understanding of the links between impact rates and crater sizes and the dynamics of the Martian surface could also be used by the space scientists and engineers to better evaluate the potential risks to robotic spacecraft, probes and future human exploration on Mars. Potential risks include damage from meteorite impacts, surface disturbances, and changes in local atmospheric conditions, all’of which could affect the safety and longevity of equipment and crew.
Paper 1: New Impacts on Mars: Systematic Identification and Association with InSight Seismic Events
Paper 2: New Impacts on Mars: Unraveling Seismic Propagation Paths through a Cerberus Fossae Impact Detection
This article is based on a
