The guest on today’s show is Lauren Guy, the CTO and founder of Asterra. He has a background in Geophysics and discovered the technology on which Asterra is built during his masters. Lauren was involved in a project that used radar sensors orbiting Mars to search for water on the planet. The Synthetic Aperture Radar (SAR) signals used penetrate the ground to give a clearer view of what lies beneath and whether there is any indication of the presence of water.
How Bad Are Water Leakages?
Most of our drinking water is transported from far-off locations like reservoirs or even desalination plants near the oceans.
About 30-40% of the water that is moved around the world is lost through leaky pipes.
It is not only water that is lost in leakages – a huge amount of energy (used to pump the water) is lost as well. Considering the growing impacts of climate change, and looming future water scarcity issues, this is incredibly significant.
How Does Synthetic Aperture Radar (SAR) Find Water Leakages on Earth?
Different materials have varying dielectric constants and therefore, electrical conductivity. These properties cause materials to reflect SAR signals differently. Treating drinking water gives it a distinct salinity level from other makeups of water.
Since salinity affects conductivity, it reflects a distinct SAR signal that differentiates drinking water from other kinds of water sources in the ground.
The assumption made is that there is no other source of drinking water in the ground – it has to come from pipes. This makes it possible to use SAR to create an underground map illustrating leakages.
Even with the capability to accurately isolate drinking water from other kinds of water, there are still possibilities for false positives as the same water is used for a lot of different uses like watering lawns, gardens, or filling swimming pools. The isolation has to go a notch higher to be able to distinguish the drinking water that is coming from pipes.
One way to do this is by calibrating the algorithm to only show moisture in the ground that has been accumulating for more than 48 hours. The thinking behind this is that most people do not water their lawn for more than 48 hours at once. This helps to avoid wasting time on false triggers.
How Far Can SAR Penetrate into the Ground?
SAR can only penetrate a few meters into the ground, depending on the soil type, and the top covering (i.e. asphalt, pavement, etc.). Generally, the depth of SAR penetration is about 2m in cities, 5m in more rural locations, and up to 10m in very sandy soils. The penetration depth of SAR is suitable for this application since water pipes are usually laid at a depth of 1m.
Georeferencing SAR Images
Georeferencing is a critical part of working with SAR images. The images need to be georeferenced in order to figure out where they are on the surface. Finding the exact location where there is a leak as shown in a SAR image is very important to avoid sending a crew to the wrong place.
Since SAR sensors are usually pointed at Earth at an oblique angle, georeferencing SAR images can be difficult. Georeferencing algorithms have to undergo a robust training phase in order to achieve the level of accuracy required. As we continue to see the popularity of SAR technology grow, we may see this get easier.
Tackling Signal Noise in Urban Environments
Telecommunications in urban environments creates a lot of noise for SAR sensors as they use the same frequencies. Reflective surfaces also create a lot of noise. This problem can be overcome by using different polarizations. A SAR signal can be sent in three ways: vertical, horizontal, or as an alternating combination of the two in some cases.
When the signal bounces off different materials or noises, the polarised signal goes from vertical to horizontal and vice versa. Detecting these polarization changes, and measuring their magnitude makes it possible to identify the source of noise that caused the change of polarizations and correct for it.
Building a Business Around SAR Tech
The problems that Asterra faced while building a business around SAR technology are quite the same for many companies that are developing new solutions in tech. Despite putting in a lot of effort to convince clients that the technology actually works, it is even more difficult to convince a client to create a new budget to buy that solution.
For Asterra, these were passive utility companies that were not actively looking for leakages in their infrastructure, but primarily relied on citizens to report suspected leakages. One of the more common ways a customer might notice a leak is a sudden spike or even gradual increase in their water bill, without having changed their habits. As utility companies did not have an existing budget for a field team that looked for leakages, it was difficult for them to justify a new expense.
Why You Should “Speak the Same Language” As Your Clients
The best kind of clients are the ones that are already solving a problem. It is easier to convince a client to buy your solution if you can acknowledge their existing ideas and solutions, and then offer your ‘new’ solution as a complement to theirs. It will not sound realistic to them to ask them to throw everything out – what is often very expensive equipment – and replace it with your solution. A general good tip in convincing anyone of something, is to make them feel like they came up with it themselves.
For Asterra, it was easier to convince utilities that already had a field team that were actively looking for leakages in their infrastructure, and already had a budget for it. The companies could just reappropriate the budget towards buying the new solution, and even save costs as Asterra’s solution is often cheaper than the original way.
There were also advanced utilities that not only have a field team, but also very expensive IoT equipment for finding leakages. Due to the high costs involved, it is not economical to install the equipment throughout an entire city. For these cases, Asterra’s solution could be adopted as a complement to these devices in order to identify the most problematic areas in the city, and shift the more costly equipment to those locations where it is needed most.
Embracing Competition in Tech Businesses
For most businesses, the thought of competitors may be dreadful, but in tech businesses, competition might just be the force that drives the business forward. Being the only player in the game may be viewed by clients and investors to mean that the market is not viable. It leaves them with many questions of why there are no other solutions in that area already if there is so much money to be made. Competition fuels more discussion around a technology, which is an opportunity for the newest solutions to gain exposure.
Other Applications of SAR Technology
SAR technology is also useful in monitoring other important infrastructure like highways and railways for potential issues. Since SAR is looking underground, the issues it identifies may not be apparent yet from ground level. Accumulated water causes most of the issues in infrastructure. Pinpointing locations with very high soil moisture can help railway companies, or a country’s Department of Transportation to identify infrastructure that may fail soon. SAR’s ability to penetrate the ground at night and in any kind of weather can be harnessed in many other applications, such as mineral explorations, defence, and identifying contaminated soils. Who will be the ones to make it happen?
Recommended Podcast Episodes
In Conversation
From Searching for Water on Mars
Daniel: Lauren, welcome to the podcast. You’re the CTO and founder of Asterra, and the way I understand it, you look for leaky pipes from space. Would you introduce yourself and confirm whether that’s what Asterra does?
Lauren: Thank you for having me. I’m Lauren Guy, CTO and founder of Asterra. I’m a physicist by trade — my bachelor’s is in geology and I continued in my master’s to be a geophysicist. At Asterra we are looking from space for leakages under the ground, without even setting a foot on the ground — totally remote sensing at its purest.
Daniel: How did you get from a geophysics background to a project looking for water on Mars?
Lauren: As a kid I always loved maps — I used to go to sleep with an atlas every night. When I thought about what to study I couldn’t decide between geography and geology, and geology was more interesting. But I realised I was more interested in the processes in the atmosphere and on Earth and other planets than in rocks and dinosaurs, so I focused on the geophysical side. During my master’s, one of the projects I was involved in used radar sensors orbiting Mars to look for indicators of water. If there’s water on Mars it’s under the ground, so you need a sensor that can penetrate the ground — like an x-ray machine — and luckily, we found indications of actual water buried in Mars.
To Finding Leaky Pipes on Earth
Daniel: How does that technology translate from finding water on Mars to identifying leaky pipes on Earth?
Lauren: After my studies my first job was a geology survey here in Israel, on a big highway project next to Jerusalem. Speaking with the engineers at the local utility, I found out that 25 to 40% of the drinking water transported for hundreds of miles from reservoirs or desalination plants is leaking into the ground — sometimes just a few meters before your house. Once I heard that, I had a light-bulb moment: I can look for water under the ground. That started the motion of how to take this technology and apply it on Earth. New issues came up, of course — on Mars we looked for any kind of water, but here we want to find very specific drinking water, which is treated.
Daniel: Do you know how much energy is lost moving that water around?
Lauren: Pushing water takes a lot of energy. From the statistics we saw, in developed countries — Europe, the US, Australia — around 17% of the country’s energy is pushing water from the source to the end client. Now think: out of that 17%, 30% is leaking. You can imagine the huge amount of energy being lost. It’s not just the magnitude of water — it’s also the energy, and the carbon footprint.
Creating a Signature for Drinking Water
Daniel: How do you create a SAR signature for drinking water specifically?
Lauren: What synthetic aperture radar measures is the dielectric constant of the materials the signal touches — how well a phenomenon conducts electricity. Drinking water is treated, so it has a very distinct salinity level, and salinity affects conductivity. If you look only for the signal of treated water in the ground, you can remove all other water sources and emphasise just the treated water. We have the assumption that if we see treated water under the ground, it has to come from a pipe — there’s no other source. That’s the founding idea of the solution.
Daniel: How do you avoid false positives — people watering lawns, swimming pools?
Lauren: In a big city there are thousands of water sources above and below ground. Beyond focusing only on treated water, we calibrated our algorithms to only show moisture in the ground that’s been accumulating for more than 48 hours — most people don’t water their lawn for 48 hours straight. So we only focus on the leaks that are actually there and staying there. It’s a monitoring solution — we take an image of the city at a defined interval that we agree with the utility, and we alert if we find changes that might indicate a leakage.
Penetration Depth and Georeferencing
Daniel: How far can SAR penetrate the ground?
Lauren: People expect us to say hundreds of meters, but that’s not the case — SAR or microwave can only penetrate a few meters. It depends on the soil type and what’s on top — asphalt, pavement, or nothing. Within cities we can penetrate up to about 2 meters, which is fine because the pipes are around 1 meter deep. In rural areas the penetration can reach 5 meters, and in very sandy soil even 10.
Daniel: Georeferencing SAR is difficult because of the oblique looking angle — how do you solve that?
Lauren: SAR satellites look from the side, and those side images don’t make sense unless you georeference them to the right location. The georeferencing error isn’t constant — it can be 20 meters here and 100 meters or even kilometres a block away, related to the topography and radar errors like foreshortening. Exact location is very important for us — we don’t want to send the client to the wrong place. In the beginning we had a team manually connecting each pixel of the radar image to the optical image — one image of 5 million pixels took a week, which doesn’t scale. So we developed a way to auto-georeference L-band SAR images in a few seconds, taking a deviation of hundreds of meters down to 5 or 10 meters, which is more than okay for us.
Daniel: And the noise from telecommunications in urban environments?
Lauren: When a young researcher tells me they want to work with SAR, I tell them to run away — it’s very chaotic and problematic. The frequencies we work at are the same as cell phones and telecommunication devices, so an image of a city is a white cloud of noise masking everything beneath. We overcome it mostly using different polarizations — a signal can be sent vertically or horizontally, and when it touches different materials the polarization changes. By detecting and measuring those changes, we can identify the source of the noise and remove it.
Ground-Truthing and Closing the Loop
Daniel: How do you ground-truth the leaks you find?
Lauren: A lot of remote sensing solutions don’t close the loop — think of NDVI; no one really goes to the location to check the value. But because we have to find very specific issues, that wasn’t an option for us. From the beginning we simulated thousands of leakages under the ground in multiple places around the world, because we wanted one global product. Now, when the utility goes to the field to verify a location, they report back on our app whether it’s a leak — and that re-feeds our algorithm, making it better in a machine learning process. We don’t tell them to dig everywhere. A leak usually makes an acoustic sound on the pipe, so we give them a precise location — a buffer of a few tens of meters — they verify with their own acoustic devices, and only if it sounds like a leak do they dig. In the beginning we were only 20% correct, which sounds awful, but it was still commercial — they could find two leaks a day, twice better than before. Today we’re at 86 to 90% success.
Convincing Old-Fashioned Utilities
Daniel: What did people say when you first told them you could find leaks from space?
Lauren: No one wanted to buy it. I thought everyone would line up and I’d be a millionaire within days — that wasn’t the case. Water utilities are very old-fashioned; they’ve used the same methods for a very long time. They saw us as a group of charlatans. So to prove it, we sent our salespeople to the first meeting with the utility’s leaks already examined — because we can take an image of any location on Earth, we’d scan the utility before the meeting and arrive with a map of their entire system and its leakages. They were shocked we could take an image without their permission, but once they saw it they couldn’t ignore it. We were a bit naughty and aggressive in the beginning, but it was the only way to create assurance in the technology. Then the discussion moves from “does it work?” to “what is the ROI?” — and that’s a good thing, because you want to move from the technology discussion to the business discussion.
Daniel: Are utilities eager to understand where they have leakages?
Lauren: No — many utilities are very passive; they’re not actively looking, they just wait for a citizen to call. If you tell them to spend tens or hundreds of thousands of dollars on an image, it’s very hard for them to justify an expense they didn’t have before, even though they can see the ROI. The best clients are the ones already solving the issue — those walking the streets randomly with a team, where it’s easy to show the ROI versus our solution, measured in “price of leak.” They already have a budget, so they can just shift it to us. The third kind are advanced utilities with expensive IoT equipment — we complement them; because even big cities can’t cover the whole city with that equipment, they use us at a macro level to find the most problematic areas, then lift and shift their equipment there.
Competition, Other Applications, and Lessons Learned
Daniel: Does it worry you that someone could copy what you’re doing?
Lauren: I really love competition. If you go into a blue ocean where you’re the only solution, it’s very hard to justify the market — clients wonder why no one else is selling to it. So when we see more competitors — not exactly in SAR or our bands, but remote sensing solutions for soil moisture and leak detection — we love it. It shows our technology is superior because the client can compare, and it creates discussion. Eight years ago no one spoke about looking for leakages via remote sensing; now it’s becoming a legit, common solution, which makes our life easier. We’ve also expanded — from drinking water to soil moisture in general, monitoring infrastructure like highways and railways for accumulated water that hasn’t drained, plus contaminated soils, defence, and mineral exploration.
Daniel: If you could go back eight years, what would you tell yourself?
Lauren: Honestly, I don’t think water leakages was the best business to start with — it’s a very old-fashioned market and it was very hard to prove the business case. A new startup wants to go fast, in a vertical that’s more private and less governmental, and one that isn’t such a blue ocean — some competition helps. The company was founded in 2013, and I was alone until 2015 or 2016 — a one-person show, because the algorithm was only around 25% accurate and it was hard to prove the commercial case. The big push came in 2016 when the algorithm improved and people who actually know how to sell joined, including a CEO with a lot of experience. You can find me on LinkedIn — I’m one of the only people with the name Lauren Guy — and at our website, asterra.io.
