Hacker News Reader

Comment by jessriedel | Hacker News Reader

Putting aside the climate aspects and just concentrating on companies that are trying to identify methane leaks: If I have a device that can accurately see methane leaks, I am very surprised it's really more economical to put them on satellites than to operate them locally. Yes, I get that you can have one device in orbit that can do the job for multiple companies, but satellites are just fantastically expensive. Unlike photography, where the alternative is to use an also-expensive airplane, why can't companies just survey from the ground?

View on HN

> satellites are just fantasically expensive

Are they? A CubeSat can be launched for less than $100k, and their payload can contain off-the-shelf consumer hardware. That's what some upper-middle class families pay for an extra car. I think you're overestimating the cost of satellites.

At the same time, you're underestimating the challenges of operating a ground sensor network. How would identify which sites to survey from the ground without a broader search? Would the oil industry get to regulate itself? What about temporal variability or detecting flares? How would you detect new leaks and add new sensors? What about large-scale natural leaks that spread over many kms in inaccessible terrain?

Even assuming you could pinpoint the location of every active methane source ahead of time, how would you get a sensor there? How would you secure access for installation and maintenance? How would you ensure connectivity? How would you monitor in areas controlled by hostile governments?

Each of the hundreds of thousands of sites has its own logistical challenges which explode the cost. And because you need to keep the sensor network up to date, it's a constant operational battle, not a one time expense. Compare that to a satellite: the one-shot launch, then lower cost, lower operations, and superior temporal detection abilities and global coverage - not hard to understand why satellites are preferred.

I mostly agree but the $100k is only launch cost, it doesn't include the engineering and manufacturing of the satellite or the ongoing operational costs.

Well funded universities are putting up cubesats. I cannot imagine the operational costs are outside the scope of a serious organization.

$100k would only hire a single employee, it wouldn't even cover getting them into the field let alone setting up and maintaining infrastructure over the course of a year.

It's possible to reduce this number by taking advantages of labor conditions and goodwill, but that turns out to be a ton of legwork if you want a reliable team in the area year after year.

For example, in an anthro field school you might source local labor by identifying people who want to learn about their ancestors, and you'll expect a lot of community outreach and Q&A sessions around what you're studying.

You could attempt a similar approach for studying methane but I'd personally expect the Q&A sessions to include a heavy undertone of "convince me we shouldn't just kill the people involved", and all of my arguments against would be from the perspective of a system they don't really give a shit about. I'd certainly never put myself in that position.

(disclaimer I work on climate/methane mapping)

Companies do survey from the ground, and for large facilities there is often permanent monitoring.

Detecting methane from the ground still requires an expensive instrument, and it covers a very tiny area.

Having said that, a lot of methane leaks are from abandoned well heads, and in many places there just are no economic incentives to care about it.

Anyway, satellites are not cheap from a capex standpoint, but they are surprisingly cheap in terms of $/m^2. A camera moving over the earth at 7km/s mapping a 200km swath can cover a million square km for O($1000). Getting a team with expensive equipment into the field for a day on one site could easily be more than that.

I used to work on this at a methane monitoring shop. Others have good points.

Monitoring from the ground using methane sensors that detect the gas directly requires a lot of human labor. You need to send people out to often remote locations to set up bulky tripods full of (still expensive) sensors. You need some expensive data backhaul because these places often don’t have cellular access. We were billed by the byte to have a 24/7 contract with uptime guarantees in some of these locations. One big issue is that the companies paying for self-monitoring want suspiciously low false positives (for legal compliance reasons). You have to work with the land owners if you expect them to let you put sensors on their sites. This limits who can access that data compared to a NGO running a satellite, and requires you to filter out “noise” that’s probably not noise to appease the customers. They’re also not very accurate at detecting source in that close distance. Gases tend to quickly rise above where a human can place a sensor array and the wind and environmental factors play a huge impact. Very few companies are willing to stand behind their physics modeling with any kind of SLA. The human cost also shouldn’t be understated: when my coworkers would go and set up these sensors they’d be dizzy for days/weeks due to the emitted gasses they breathed in, our employer basically couldn’t find affordable medical insurance for us.

If you want to use spectroscopy/cameras, you can do it from a (small) distance. But it’s fantastically expensive equipment that only can be used for 1 site. And the “looking up” angle is worse than “looking down”. It’s an open question in the industry if the data can be trusted from these setups. You still have a lot of the complexity of maintaining them and exfiltrating the data.

If you want to just fly a plane… that’s expensive, you need to do it regularly, you still have a lot of “not actively monitored” time, etc.

In calibrating satellites getting folks like you to go out with sensors on the ground to send back well vetted data is an expensive and limiting factor. Much cheaper (and often preferred) to do as much as possible from space and do smart ground processing/data analysis to build confidence.

Ground-based sampling has different capabilities, e.g. these days you can mount a small mass spectrometer (see the ones going to Mars, the Moon) or related devices on a drone and get data about a much wider range of particulates and compounds. e.g.

"VOC Sampler on a Drone Assisting in Tracing the Potential Sources by a Dispersion Model - Case Study of Industrial Emissions" (2023)

https://aaqr.org/articles/aaqr-23-07-tn-0169

> "Two plants which manufacture petrochemical products at an industrial complex in Kaohsiung City, southern Taiwan, were applied as the targets for VOCs sampling and further qualitative and quantitative analysis in the laboratory. Aromatic hydrocarbons, including toluene of 433 ppb, ethylbenzene and xylenes of 100–200 ppb and phenol of 111 ppb were identified."

US EPA also has a similar program.

I had to dig into this very topic for a previous product that I was managing.

Satellite photography/sensing vs Airplane based vs UAV based vs local vs point detection.

The higher the altitude of your sensor, the larger the swath of land you can sense, but with height comes tradeoffs... cost, weather, satellite availability, latency/how often the satellite is going to be over where you want to sense just to name a few.

Closer to or on the ground means you can get pinpoint accuracy but that comes with increasing cost of labor. Think cost of an airplane or UAV pilot vs an individual walking around with a methane sensor sniffing around flanges, valve stems, etc.

The different approaches are somewhat complimentary. Satellite can you tell you've got a problem area. You can then deploy people on the ground to "zoom in", but if you're operating a ton of sites, you wouldn't start on the ground.

It's probably simply efficiency. You can check an entire pipeline faster from space than from the air, and from the air vs. the ground. Those efficiencies save a fair bit of money, and probably also lead to earlier detection.

And while they are expensive, we're talking about low single-digit millions of dollars for a satellite which has a lifespan of 5+ years. Relatively inexpensive for the companies who would want to use them.

I can't imagine it's cheaper to do terrestrial search for methane leaks unless your goal is to only find the methane sources you are incentivized to find, and the cost of the satellite gets amortized over thousands of sources and millions to billions of stakeholders.

"putting aside the climate aspects" (???) It's 2024, this is the equivalent of putting a bit "#if 0/#endif" around your comment. I'm taking the time to post this because I hope if you reflect on it, over time, you might realize that the devil actually doesn't need an advocate.

> why can't companies just survey from the ground?

From a layman's perspective:

(1) the incentive is not aligned to accurately report leaks [1][2][3][4][5]

(2) Satellites provide global monitoring. This reaches to places/countries/regions that are not regulated, and secondly provides monitoring for natural sources of methane emissions (eg: permafrost). For the latter, even if there is nothing to be done about natural emissions, the more accurate data will provide for better modelling and help humans better understand the Earth's ecosystem.

[1] June 2022: "Oil and gas companies underreported methane leaks, new study shows" https://www.washingtonpost.com/climate-environment/2022/06/0...

[2] Oct 2023: "Oil and gas companies are missing significant methane emissions. Here’s how to fix that." https://rmi.org/oil-and-gas-companies-are-missing-significan...

[3] "Gross under-reporting of fugitive methane emissions has big implications for industry" https://ieefa.org/resources/gross-under-reporting-fugitive-m...

[4] July 2023: "Australian Fossil Fuels Giants Severely Underestimate Methane Leaks" - "Fugitive — or hidden — methane emissions in the country are likely underestimated by 80% for coal and 90% for oil and gas, a new report from IEEFA say" https://www.bloomberg.com/news/articles/2023-07-04/australia...

[5] March 2022 "Stanford-led study: Methane leaks are far worse than estimates, at least in New Mexico, but there’s hope" https://news.stanford.edu/press-releases/2022/03/24/methane-...

I'm specifically referring to the cases in the article where companies pay for this.

Companies and countries tend not to report catastrophic methane leaks (or really any contamination event) unless they're caught in the act. It's absolutely staggering the rate at which methane leaks are identified now that satellite data is being analysed and published.

It'd be much more efficient all round for firms & countries to just test locally & report their methane leaks, but they can't be trusted to do so.

Ground-based sensors historically undercount methane emissions in 3rd party trials. They have a limited FOV which makes it hard to accurately measure a plume of gas, particularly when the wind blows. I suppose you could create a Dyson sphere of fixed sensors around an O&G asset, but that quickly becomes more expensive than a space based sensor.

Because these leaks can occur spontaneously, you really want real-time, continuous monitoring of a wide area with high resolution. Aerial sensors are impractical for this purpose. Only a large constellation of satellites can offer this.

The short answer is to measure methane you rely on methane absorption features. To do that you need to be looking at a source through the methane plume, from a satellite this source is the ground. From a ground instrument you can use the sun as a source, but it limits the times of day/direction/location you can measure from.

> satellites are just fantastically expensive

This has become less and less the case over the last decade or so. There are more or less mass-produced satellite buses now, and a SpaceX rideshare puts 150kg into orbit for less than a million. If you go small, you can probably procure, launch and operate a satellite for less than $1000 per day of usage.

Reply on news.ycombinator.com