Exploring Secrets Beneath the Ice Layers

Drone-mounted radar explores future climate changes.

I stand on a meter-thick ice surface, watching a drone dart over the Slakbreen Glacier in the Svalbard archipelago of Norway, over 600 kilometers from the mainland. I am part of the “Peregrine” test team, a fixed-wing unmanned aerial vehicle equipped with a small ice-penetrating radar that can image the ice and probe down to the bedrock beneath the glacier.

The current temperature is minus 27°C, and the biting wind chill brings it down to minus 40°C, far below the operational temperatures of most commercial equipment we brought for this expedition. Our phones, laptops, and cameras quickly stop working. The last computer still functioning is placed on a small heating pad inside a tent.

The weather here is already severe, and we need the “Peregrine” to operate under even harsher conditions, regularly surveying the ice sheets of Antarctica and Greenland. If these massive glaciers were to completely melt, the vast amount of water they hold could raise global sea levels by 65 meters. Although these two ice sheets won’t melt completely anytime soon, their immense scale will have significant implications for the future of our planet. The data collected by the “Peregrine” will help scientists understand how these critical areas will respond to climate change.

Exploring Secrets Beneath the Ice Layers

For a long time, scientists have relied on satellite laser altimeters to collect data and observe changes in the surface elevation of ice sheets. Most of this data comes from the ICESat satellite launched in 2003 and its successor ICESat-2 launched in 2018. Scientists measure changes in elevation based on the information provided by these NASA satellites, inferring the net impact of snowfall, melting, and other processes on ice surface changes, as well as the rate at which icebergs are released from the ice sheets into the ocean.

These measurements are certainly important, but laser altimetry cannot directly provide information about what is happening beneath the ice surface, including how the ice deforms and how it slides on the bedrock.

When we try to understand how ice sheets respond to new extreme climates, these processes are critical. How do temperature changes affect the rate of ice deformation under its own weight? To what extent does the presence of liquid water at the base of glaciers lubricate the ice bed, causing the ice to slide more rapidly into the ocean?

To find answers to these questions, we need to understand the conditions beneath the ice surface. Ice-penetrating radar (IPR) is a technology that uses radio waves to image the inner layers of glaciers and their underlying bedrock. Unlike other labor-intensive methods (such as drilling or setting up arrays of geophones to collect seismic data), ice-penetrating radar systems have been airborne since their inception.

In the 1960s, as part of an international collaboration, the U.S. Navy modified a Lockheed C-130 Hercules transport aircraft to serve as an ice-penetrating radar data collection platform. This project (which I will discuss in more detail later) demonstrated that these types of data could be rapidly collected even in the most remote areas of Antarctica. Since then, ice-penetrating radar instruments have improved significantly, and data analysis methods have also advanced, allowing for predictions of future sea level rise.

Meanwhile, the aircraft used for data collection have changed relatively little. Modern instruments are often mounted on the DHC-6 Twin Otter twin-engine turboprop aircraft from Canada or the Basler BT-67, which is a modified Douglas DC-3. (Some Basler aircraft currently operating in Antarctica previously served during World War II.) While the level of support for such operations varies by country, the demand for new data has outstripped the capacity of manned aircraft to collect it, at least in terms of cost, only the most well-funded missions can afford it.

Collecting such data should not be that difficult today.

That’s why Dustin Schrader and I, along with other students at Stanford’s Radio Glaciology Laboratory, began developing several new types of ice-penetrating radar systems, and the “Peregrine” is one of them.

The “Peregrine,” as a modified drone, carries our small ice-penetrating radar based on a software-defined radio design. The radar system weighs less than 1 kilogram, which is much lighter than traditional ice-penetrating radar systems that often take up the entire equipment rack of a manned aircraft. The entire setup (the drone plus the radar system) costs only a few thousand dollars and is packed into a robust box about the size of a large checked suitcase.

But to truly understand why we feel the need to bring the “Peregrine” into the world, you need to understand the history of ice-penetrating radar data collection.

The first large-scale traditional ice-penetrating radar surveys in Antarctica began in the late 1960s, when a group of earth scientists from the U.S., the UK, and Denmark installed a radar antenna under the wing of a C-130. At that time, GPS had not yet been invented, and the project used an internal navigation system and known ground waypoints to record the flight path. The system recorded radar echoes using a modified cathode ray tube, scanning a roll of optical film, which researchers supplemented with handwritten notes. This survey produced hundreds of rolls of film and piles of notebooks.

The project ended in 1979, after which various countries and regions launched projects to conduct regional surveys of Antarctica and Greenland. Although the initial survey scope was limited, these projects continued to evolve and, most importantly, they began to collect digital data with GPS coordinates.

At the end of the first decade of the 21st century, ice-penetrating radar surveys received an unexpected boost. In 2003, after just 36 days of data collection, ICESat lost one of its laser altimeters, and by the end of 2009, all of the satellite’s laser equipment ceased operation. The issues with laser altimetry seemed unrelated to airborne ice-penetrating radar surveys. But with a few years until the launch of ICESat-2 and a favorable political climate for public Earth science funding in the U.S., NASA organized Operation IceBridge, a large-scale airborne survey effort aimed at filling the data gaps in laser altimetry for Greenland and Antarctica.

Although the primary goal was to collect laser altimetry data, using aircraft instead of satellites meant that additional instruments could be easily added. At that time, the University of Texas Institute for Geophysics and the Center for Remote Sensing and Integrated Systems (CReSIS) at the University of Kansas were both developing improved versions of ice-penetrating radar instruments, which were ready for deployment.

From 2009 to 2019, the aircraft involved in Operation IceBridge flew over 350,000 kilometers in Antarctica, collecting ice-penetrating radar data. During the same period, the National Science Foundation’s Antarctic Central Plate Ice Sheet Evolution Survey (ICECAP) project funded over 250,000 kilometers of additional Antarctic ice-penetrating radar data collection projects.

Operation IceBridge significantly increased the volume of global ice-penetrating radar data collected. While other organizations around the world, particularly the British Antarctic Survey and the Alfred Wegener Institute, also collected and continue to collect ice-penetrating radar data, the U.S.-led data collection efforts during Operation IceBridge transformed what had been an almost negligible state for many years into a primary data source.

In 2018, the launch of IceSat-2 marked the end of Operation IceBridge. While some ice-penetrating radar measurements continue, the pace of data collection has noticeably lagged behind the scientific demand for observations since 2018.

Recently, scientists have noted that the important types of ice-penetrating radar data have changed, increasing the demand for better ice monitoring tools. Historically, this type of radar measurement has been used to determine the thickness of glaciers over rock or sediment layers.

With few exceptions, the bedrock beneath the ice does not change on human-relevant time scales. Therefore, collecting this type of ice-penetrating radar data is usually a one-time effort, or at least not frequently collected. Once enough data is collected, sufficiently detailed maps of the bedrock beneath glaciers or ice sheets can be established.

But the depth from the glacier to the bedrock is not the only important information hidden beneath the ice surface. For example, ice-penetrating radar data reveal internal layering within glaciers caused by variations in snowfall. The shapes of these internal layers provide clues about the flow of ice in the present and past.

Scientists can also observe the reflectivity of the bedrock, which can indicate the presence of liquid water. The presence of water can indicate the temperature of the surrounding ice. The presence of water plays a crucial role in the speed of glacier movement, as it can lubricate the base of the glacier, leading to faster sliding and greater mass loss.

All of this can change on a yearly or even seasonal basis, requiring dynamic observations. Conducting radar surveys every few years is not enough.

Relying solely on manned flights to collect data more frequently is challenging—it’s costly, logistically difficult, and the harsh environment puts people at risk. The main question in replacing manned aircraft is which direction to take: up (satellite constellations) or down (drone fleets)?

A few satellites can provide years of global coverage and frequent repeat measurements, but this is not an ideal platform for ice-penetrating radar. To achieve the same unit area power on the ice surface as a 1-watt transmitter on a drone at 100 meters altitude, a satellite in a 400-kilometer orbit would need approximately 15 megawatts of power, which is more than three times the maximum power allowed by the U.S. Federal Communications Commission for SpaceX’s Starlink satellites.

Interference is another challenge. Suppose you have an antenna that emits power primarily in a 10-degree conical range. You want to observe the bottom of the ice sheet at a depth of 1.5 kilometers, but there is a mountain 35 kilometers away. From over 400 kilometers up, your antenna will also pick up that mountain, and the energy reflected back from the mountain will be much stronger than the energy reflected back from the bottom of the ice sheet, which is attenuated by passing through 1.5 kilometers of ice.

Another option is drones, which can fly closer to the ice than manned aircraft. Researchers have been interested in the potential of drone-mounted radar systems for ice imaging for at least a decade. In 2014, CReSIS equipped a 5-meter wingspan radio-controlled aircraft with a miniaturized version of the ice-penetrating radar system. This design cleverly utilized the geometry of the wing as a low-frequency antenna, but the bandwidth was limited, affecting data quality.

This demonstration was a proof of concept, and most subsequent research has shifted focus to high-frequency systems, sometimes referred to as snow radar, for near-surface imaging, which can better understand snow accumulation on mountain slopes, snow on sea ice, and layering structures in the top few meters of the ice sheet. CReSIS has tested its snow radar on a small unmanned helicopter; recently, it collaborated with NASA and Vanella Drone Company to equip a large drone with a wingspan of 11 meters with snow radar, which can stay airborne for days at a time.

Nonetheless, we still need ice-penetrating radar imaging that can penetrate the ice sheet and has sufficient bandwidth to distinguish internal layers, all at a cost that is widely accessible.

This is precisely where the “Peregrine” comes into play. The project, initiated in 2020, aims to build a smaller, more economical system than ever before, made possible by advancements in fixed-wing drones and miniaturized electronics.

We knew we couldn’t use existing systems to do ice-penetrating radar. We had to start from scratch to develop a small and lightweight system suitable for inexpensive drones.

We decided to use software-defined radio technology in the radar because the radio frequency transmitters and receivers are highly customizable, and much of the system complexity is shifted from hardware to software. Using software-defined radio, the entire radar system can be built on a few small circuit boards.

From the outset, we focused on going beyond the first project, developing a software application interface for Ettus-based USRP hardware that could work with various software-defined radios, with costs ranging from $1,000 to $30,000 and weights ranging from tens of grams to several kilograms.

We added a Raspberry Pi single-board computer to control the software-defined radio. The Raspberry Pi also connects to a series of temperature sensors, ensuring the system does not get too hot or too cold.

The software-defined radio itself is bi-directional, one side for transmitting radar signals and the other for receiving echoes, each connected to our custom antenna through amplifiers and filters. The entire system weighs just under 1 kilogram.

The design of these antennas is tricky. Ice-penetrating radar antennas need relatively low frequencies (because higher frequencies experience more attenuation through the ice) and relatively wide bandwidths (to achieve sufficient range resolution). Typically, this means a large antenna, but our small drones cannot carry large and heavy antennas.

I started with a standard bowtie antenna, commonly used in ground radar systems. The initial design was too large, and our small drones couldn’t fit even one antenna, let alone two. So, I adjusted the geometry of the antenna using a digital model, finding an acceptable compromise between size and performance based on the simulation software I used.

I also built several prototypes to understand how the real antenna performance differed from the simulations. The first was made by cutting copper tape and sticking it onto a plastic sheet. Later versions and the final version were made as printed circuit boards. After several iterations, I had a working antenna that could be mounted beneath each wing of the small aircraft.

For the drone, we began using the X-UAV Talon radio-controlled aircraft kit, which includes a foam fuselage, tail assembly, and wings. Every conductive material on the aircraft affects the performance of the antenna and can produce adverse effects. Testing showed that the carbon fiber wing spar and the wires connecting the servos on the wings created problematic inductive pathways with the antenna, so we replaced the carbon fiber spar with a fiberglass spar and added ferrite beads to the servo lines as low-pass filters.

I thought everything was ready by this point. But when we took the drone to a site near the lab, we found that we couldn’t GPS-locate the drone when the radar system was powered on. After some initial confusion, we discovered the source of the interference: the USB 3.0 interface of the system. To solve this, I designed a plastic box to house the Raspberry Pi and software-defined radio, 3D printed it, and wrapped it in a thin layer of copper tape. This shielded the troublesome USB circuitry, preventing it from interfering with the rest of the system.

Finally, we launched the micro-radar drone over a dry lake bed at Stanford University. While our system couldn’t image through the soil, we were able to obtain strong reflected waves from the ground, at which point we knew we had a working prototype.

Six months later, we conducted our first real-world tests on the Vatnajökull glacier in Iceland, thanks to the help and generosity of local partners at the University of Iceland and funding from NASA. This location was excellent because the nearby volcano occasionally erupts, covering the ice sheet’s surface with volcanic material (ash). This ash is then buried under new snow and forms layers beneath the ice surface. We believe these geological layers can serve as good analogs for the internal layering found in Greenland and Antarctica. Although there is a lot of liquid water in the relatively warm ice layers of Vatnajökull, our system could not detect the ice beneath several meters, but the volcanic ash layers were still very apparent in the radar detection.

However, these initial tests were not without their challenges. After one test flight, I found that the data collected was almost entirely noise. We tested every component and cable until I discovered that one coaxial cable’s shielding had cracked, leaving only intermittent connections. After replacing it with a spare cable and a lot of hot glue, we completed the remaining tests.

In the next round of testing, our goal is to image the bedrock beneath glaciers, not just the internal structure of the glaciers. That’s why we came to the coldest regions of Norway’s glaciers in March 2023, where the presence of liquid water within the ice is less likely to interfere with measurements. There, we were able to capture images of the bedrock located 150 meters beneath the ice surface. Crucially, we are confident that our system will function properly in the harsh conditions of Antarctica and Greenland.

Currently, our system is relatively small. It is designed to be economical and portable, allowing research teams to carry it to remote locations. But we also hope it serves as a testing platform for a larger drone-mounted ice-penetrating radar system, which will have an operational range of about 800 kilometers and a low enough cost to be permanently deployed at Antarctic research stations. With the existing 11 research stations as bases, at least one member of a drone fleet could survey every coastal area of Antarctica. The next generation of drones, while larger and more expensive than our initial “Peregrine,” will still be much cheaper and easier to operate than manned systems.

It is impossible for a few PhD students to operate a large drone, let alone a fleet of drones, so we are collaborating with Stanford University, the Scripps Institution of Oceanography, and Lane Community College in Eugene, Oregon, to bring this new platform to fruition. If all goes well, we hope to conduct ice-penetrating radar-drone surveys of the Antarctic and Greenland ice sheets within three years. This will undoubtedly help scientists study how Earth’s ice sheets respond to climate change. By using permanently deployed drones covering most active research areas, requests for new data could be fulfilled within days. Surveys could be dynamically and regularly repeated in different areas. When rapid and unpredictable events occur, such as ice shelf collapses, drones could be deployed to collect real-time radar data.

Today, such observations remain impossible. But the “Peregrine” and its successors could make it possible. Having the capability to collect such radar data will help glaciologists address fundamental uncertainties in ice sheet physics, improve sea level rise predictions, and aid in better decision-making to mitigate and adapt to future climate changes on Earth.

Author: Thomas Teisberg

IEEE Spectrum

Technology Review

Official WeChat Public Platform

PreviousRecommendations

Fire Detection Drone

Drones That Withstand Strong Winds

Smart Eye: Silicon-Based Optical Phased Array Solid-State Lidar

Leave a Comment Cancel reply