Eyes on Earth Episode 118 – Preparing for Landsat Next, Part 2
Detailed Description
In this episode of Eyes on Earth, we talk with several people involved with the next Landsat mission, targeted for launch around 2030. While that may seem like a ways into the future, it takes a lot of work to prepare for such a big mission. In Part 2 of this two-part series on Landsat Next, we’ll hear about what needs to be done before launch to prepare the ground system, data processing and data calibration. Be sure to also listen to Part 1, where we talk about how different Landsat Next will be from previous Landsat missions, and what scientists are really looking forward to with this mission.
Details
Sources/Usage
Public Domain.
Transcript
TOM ADAMSON:
Hello everyone and welcome to another episode of Eyes on Earth, a podcast produced at the USGS EROS Center. Our podcast focuses on our ever-changing planet and on the people here at EROS and across the globe who use remote sensing to monitor and study the health of Earth. My name is Tom Adamson.
This is the second of two episodes about Landsat Next, the next mission in the Landsat series that is targeted for launch around 2030. In Part 1, we learned about what the Landsat Next mission will be like, and about the benefits to society and to science it will bring. Today in Part 2, we have some more technical information about the ground system, processing the data, and validating the data.
We’ll start with Brian Sauer, Landsat Next Project Manager at EROS, which is where the Landsat data is all processed, archived, and distributed. So Brian, what will be EROS’s role in the development of Landsat Next?
BRIAN SAUER:
First of all, the Landsat project is a joint agency between NASA and the USGS. Really think of us as one project team. We work very, very closely together. And similar to Landsat 9, during development, NASA is responsible for the space segment. That includes the instrument, the spacecraft, and the launch vehicle. USGS is responsible for the development of the ground system. During development, while NASA leads the mission, the joint agency partners in areas such as science, calibration and validation, and mission engineering also work very closely together. NASA leads the mission readiness campaign and launch, early orbit, and commissioning, and then upon success, they hand over the mission to the USGS.
After handover, the responsibilities sort of flip. USGS takes primary responsibility for the mission and they operate it. And then NASA turns into more of a support role. We need to think of this as one joint agency activity where we work very closely together throughout the development life cycle and during the operation.
ADAMSON:
What is EROS’s role exactly once Landsat Next is in orbit?
SAUER:
I’d like to just note that the Landsat Next and Landsat operations project offices are located right here at EROS. We work closely together throughout the development and operations cycles. Once in orbit, EROS will provide the flight operations team, the ground system, and engineering support during launch, early orbit, and commissioning. We call that LEOC because we have an acronym for about everything. While NASA is leading and responsible for the on orbit validation and acceptance periods—that’s right after we launch, we go through this acceptance period and we test out the system. NASA’s responsible for making sure that the system meets the requirements and is verified and validated.
After we get through that period, the USGS then takes ownership of it. And our flight operations team being involved in the deployment phase and that testing phase and especially when we’re on orbit, we’re going through that commissioning phase, really provides a seamless transition for when we get into operations.
ADAMSON:
That ground system staff is here at EROS, but there’s also staff at NASA Goddard. Is that how that works?
SAUER:
Yeah, we have staff at EROS that really take care of all the ground systems, testing, integration, development of the ground system and so forth. But we also have a flight operations team and a flight operations contract in our MOC. For Landsat 8-9, that is located at Goddard Space Flight Center. Notionally, it’s probably located at Goddard Space Flight Center for Landsat Next, and we expect to have a launch support capability there for sure. We haven’t really figured out where Landsat Next is going to be operated yet at this stage when we’re formulating the mission, so we’re going through those processes to determine that and we’re performing studies and other activities, but we haven’t really determined that. But traditionally for Landsat 8 and 9, our operations for monitoring the system, performing commanding to the satellite, health and safety of the satellite, that’s all performed—the Mission Operations Center capabilities are performed at Goddard Space Flight Center for our existing missions. Again, we’re going through the formulation process, so I don’t want to make any promises of where that’s going to be at today, so long-term for Landsat Next.
ADAMSON:
How will the current ground system have to change to support this new mission?
SAUER:
That’s a real good question. Obviously, it poses a new challenge, you know, for the ground system. So really the ground system, we have to make sure that it’s efficient and cost effective for long-term operations. We’re talking about a system that’s going to be used from, you know, the 2030s into the 2040s, so you know we’re going ahead here in time and we know that with three satellites and managing those satellites, ensuring health and safety of those satellites, we know the vast amount of data we’re bringing to the ground. We need to ensure that we engineer it appropriately. So the ground system team has really been leading formulation activities to ensure that this is taking place, that the system, the ground system is engineered to be as efficient as possible and handle the triplet observatories. In summary, the most the most significant changes of the ground system include handling the three satellites efficiently with much greater automation, modifying the system to be able to acquire and downlink the increased volume of the science data, and getting that data to the ground and then modifying the data processing and archive system to handle that data. We’ve been doing quite a bit of work and planning and formulating this ground system. The EROS team has been soliciting industry on several fronts to help us formulate the ground system architecture, the OPS con and the requirements. The team solicited industry requests for information—another acronym, RFIs—to understand the capabilities of the ground system and the Mission Operations Center. The responses to the RFIs were outstanding and are helping the team here at eros formulate the strategy. We’re right now in the heart of those studies and getting information back from those vendors. It’s been very fruitful.
On the ground network side, the antennas needed for Landsat Next to bring the data down to the ground use a different frequency than existing antennas. This will require either acquiring services, maybe modifying existing systems, or potential build outs. So the request for information we used helped us solicit industry to understand what capabilities and what services are out there so we can architect the ground network system appropriately.
I should note that EROS has also recently issued an industry request for information—again, RFI—for upgrading or replacing an existing antenna here at EROS. That antenna would support the existing missions, Landsat 8 and 9, and potentially would be used for Landsat Next. The results of that are still being analyzed, and we’ve just recently gathered those results. So again, we had some very good information from industry, and we’ll be using that in our studies moving forward.
ADAMSON:
Whatever that is that will be needed is still being studied.
SAUER:
Yeah, we are still in formulation. We are formulating our requirements for what this ground system is going to need and we’re in the heart of that work.
ADAMSON:
It always feels like 2030 or the early 2030s is a ways off, but you have a lot of work to do still.
SAUER:
Oh, absolutely, especially with our acquisition cycle in the federal service, we need to make sure that we involve industry, so we understand our requirements more properly and the capabilities that are out there in the marketplace. We use those then in driving out our requirements. We use those in when we issue the solicitations for acquiring the services we have and getting industry involvement in the mission. The industry involvement we have had has been influential in our work right now. We wouldn’t be near as far along as we are if we didn’t involve industry very early, and I think the lessons are the earlier you can, the better it is. So it’s been helping us considerably, even though it’s, you know, you’re talking about early 2030s, but now is the time.
ADAMSON:
Yeah, there’s no procrastinating in a mission like this. Don’t study the night before.
SAUER:
Yeah, you can’t do this like you did your biology test in high school.
ADAMSON:
How important is Landsat Next to EROS?
SAUER:
Wow, that’s an incredible question. And you know Landsat Next to me is an incredible mission. And it’s so important. It’s so vital. You know, Landsat Next is going to serve our science community really through the 2030s and into the 2040s. And I really think of it as a game changer. The measurements provided to the community not only provide continuity with our current 52-year plus archive, but also drive new and emerging applications and science research. Landsat Next isn’t just a one-time science mission. It’s a global Land survey mission, and it’s going to monitor our planet to further quantify and understand changes. For the continuity of Landsat, it’s so important, but also for the new measurements are going to be so important for these emerging applications that are needed in the next decade and beyond, so it’s very important, it’s very important to EROS. It’s very important to the American people and of course to science.
ADAMSON:
That’s going to be a lot of data coming from the three satellites of Landsat Next. Let’s bring in Chris Engebretson, the acting ground system manager for Landsat Next at EROS to see if we’re ready for all this data. Landsats 8 and 9 currently send us a total of about 1,500 scenes per day. Chris, can you tell us how many scenes Landsat Next will acquire?
CHRIS ENGEBRETSON:
So Landsat Next is expected to send about 2,500 scenes per day, and that’s across all three satellites in the constellation. So it’s a significant increase in the amount of data that we’re getting each day.
ADAMSON:
With that higher resolution, more spectral bands, it’s going to be a lot more data overall. Is EROS ready to handle that much more data when it comes?
ENGEBRETSON:
The big thing that we’re doing to get ready for Landsat Next is we’re right in the middle of an activity to take all the historical Landsat processing systems, which are running today at EROS in Sioux Falls, and migrating those systems to the cloud. And that gives us an opportunity to really revisit some of the architecture and design of those systems, which at this point some of them are fairly old. They served us very well, but it really is an opportunity for us to go back and look at those systems, but the other thing that moving to the cloud is really going to give us is a lot of flexibility in terms of our scalability and elasticity. So I mean basically what that means, if we need to scale up our processing resources to handle you know whether it’s the increased load from Landsat Next or whether we want to take the entire archive and process a new Landsat collection, being in the cloud really gives us the flexibility to scale up and scale down to handle those loads as appropriate. So this is an activity that we’ve been working on for a couple of years now. We expect to have it finished in 2026, but that that cloud-based system is really going to give us a lot of power and flexibility.
ADAMSON:
Yeah, we should clarify that this Landsat collections is sort of a reprocessing of the archive that kind of continues, so that we make sure that that data is really usable and really accurate. So how does Landsat Next data fit into Landsat collections?
ENGEBRETSON:
So Landsat Next data is planned to be a part of Landsat collection 3. Our current collection, as folks are probably aware, is Landsat collection 2. I mean, that’s been our active Landsat collection since early 2021. So when we roll out Landsat Next data, our plan is to be on collection 3 and then all Landsat Next data will fall under that Collection 3 umbrella. The timeline for Collection 3 right now is in the 2028-2029 time frame, so our intent is to have Collection 3 rolled out before the launch of Landsat Next and give people an opportunity to use that data to get migrated over to Collection 3 for, you know, a couple of years before we actually add Landsat Next to the Collection 3 specification.
ADAMSON:
Can you tell me more about processing the data in the cloud? I guess specifically I’m wondering where does that processing in the cloud take place exactly?
ENGEBRETSON:
EROS has a processing environment in the cloud that— Actually, the Department of Interior has what’s called the CHS. A cloud hosting solutions contract that provides us with, you know, basically a capability to process in the cloud. So today that capability is being provided by AWS, on Amazon Web Services. And so that’s the environment that we use to actually do the processing today and it’s the environment that we’re targeting for our cloud migration that is ongoing.
ADAMSON:
Does data have to be compressed and is that something that you have to be concerned about, balancing data compression with preservation of essential information?
ENGEBRETSON:
For Landsat Next, just for reference, we’re looking at generating about 15 terabits, you know, upwards of 15 terabits of data per observatory per day. That’s a significant increase over what we do today with Landsat 8 and Landsat 9. So in order to get that data down to the ground, yes, it does have to be compressed and just for reference, compression is nothing new. We do it today for Landsat 8 and Landsat 9. For Landsat Next, given the volume of data that we’re working with, we have to get a little bit more aggressive with the compression in order to get all that data down to the ground. We’re looking at new sophisticated algorithms to increase our compression ratio for Landsat Next. That compression approach is certainly integral to our ability to get data from the spacecraft down to the ground so that we can process it.
ADAMSON:
Just for a quick comparison, do you know how much data we’re getting per day from, for example, just Landsat 9 by itself?
ENGEBRETSON:
So Landsat 9, if you compare our expected science volume output from Landsat Next and compare that to Landsat 9, we’re expecting Landsat Next to generate about 15 times the volume of the data that’s currently being generated by Landsat 9. So, I mean, a lot of that is brought on by the increased spatial resolution, the new bands that we’re adding, which is all great, but the impact of that is that it certainly increases the size of the, the data volume coming down from the spacecraft and then the size of the resulting science products. So yes, it is a significant increase in the amount of data that we’re collecting and the size of the products that we’re producing.
ADAMSON:
And launch is still a little ways away and we’re already planning for processing all of that data. Why do we have to be planning ahead this far?
ENGEBRETSON:
Because that is how long it takes. You know, I’ve worked on Landsat for a long time, and I think I’m not the only one, but when you read news articles and you, you know, you see things about Landsat where it’s referred to as a picture taking mission or a photo snapping mission. Those things always rub me the wrong way because it sort of gives the impression that we’ve got the equivalent of a cell phone camera in orbit and we’re just taking pictures and sending them down to the ground and then making them available for people to download. The actual process that’s involved with taking raw data from a precision science instrument on orbit and generating the, you know, essentially the pretty pictures that people are used to seeing with our level one and higher products is a very complicated process. I mean, the complexity of the algorithms and the amount of number crunching that’s required to actually take that raw data and turn it into a well calibrated, useful science product is quite an undertaking, so it’s never too early to actually start that process. For Landsat Next later this year, we will actually have an instrument vendor on board, and that’s when the work really begins between you know that vendor and those of us on the ground to really be involved with the design of that instrument and the development of all the algorithms that are needed in order to actually take that raw data and turn it into usable science data. But the design of those algorithms, the development, the validation, the integration and testing, that is a process that literally takes years. So yes, we’re still a ways away from launch, but we’re getting to that point where we’re really going to start ramping up on a lot of these instrument and you know these data processing type activities because they are quite complex and it is a lot of work, but we’re excited to be at this point in the mission and we can’t wait to start really digging into this stuff.
ADAMSON:
You mentioned doing some test processing. How is that possible when we don’t have actual Landsat Next data yet?
ENGEBRETSON:
So you know, again, I’ve been working on Landsat for about 30 years and you know, going all the way back to Landsat 7 and even going back that far, this has always been a perennial challenge for us to deal with because it is, you know, your classic chicken and the egg problem. When we start getting science data down from Landsat Next, we want to be able to process that data right away, but we don’t get real data from Landsat Next until we’re actually on orbit. So yeah, that’s always the question. How do we develop and test software to process that data and generate those products if we haven’t seen any of that actual data yet?
So we’ve actually got a three-pronged approach that we’ve developed that we’re going to be using to essentially generate simulated or proxy data for Landsat Next. The first thing that we’re doing is we’re looking at existing sources of multispectral data. So obviously Landsat 9 is one of those sources, but we’re also looking at sources like Sentinel-2. Sentinel-2 will be a good one because it’s already got 20- and 10-m pixels. It already has a lot of the additional spectral bands. Not all of them, but some of the additional spectral bands that we’re adding for Landsat Next. So sources like that are, you know, a good source of data that we can start with. We’ve developed essentially this algorithm where we can take that multispectral data and then combine it with a land characterization database and then a spectral library. And we’ve got an algorithm where we can use all those things and then essentially use the spectral responses of the bands that we have in order to simulate the spectral responses of the bands that we don’t. That process isn’t perfect, but the resulting data should certainly be good enough for us to simulate a lot of our processing algorithms and do some of our software testing and things like that.
So there’s a lot of activity going on there. The second approach that we’re taking is looking at hyperspectral data. We’ve got a fairly large volume of hyperspectral data in the archive at EROS from missions like EO-1 and the Hyperion sensor. We’re also looking at other sources of hyperspectral data such as the EnMAP mission from our German colleagues at DLR, which is a hyperspectral mission that’s at 653 kilometers, which is where Landsat Next is going to be, so having that hyperspectral data gives us the ability to really simulate some of those new spectral bands that were not there before.
So that’s going to be a great source of data for us as well. And then the final thing that we’re doing is we’re working with our colleagues at NASA and at the Rochester Institute of Technology in New York where they have got actually a physics-based model where we can feed that model the exact characteristics of our instrument. And that model does a lot of physics-based simulation of not just the ground but also various atmospheric features. So we can use that model to generate some very high fidelity simulated images of what we would expect Landsat Next data to look like. So yeah, that’s a lot of work, but that’s an activity that that generally is very well worth it and it will serve us well as we get into the development and testing of Landsat Next.
ADAMSON:
All of this information really helps us appreciate what you said that the Landsat imagery isn’t just pretty pictures. I heard one of the scientists refer to them as not pretty pictures, but scientific measurements.
ENGEBRETSON:
Yeah. So I used the term pretty pictures as kind of a pejorative, because that’s the impression, but that’s certainly not the reality because yeah, every pixel in a Landsat product is a scientific measurement and every pixel has been very carefully calibrated and, you know, that’s certainly a point that we endeavor to make.
ADAMSON:
Once the Landsat Next data is brought into the ground station and processed, we need to calibrate the data. One of the truly valuable things about Landsat is how consistent and accurate it is, which helps for mapping landscape changes. Cody Anderson is the EROS Calibration and Validation Center of Excellence project manager. Landsat is often referred to as the gold standard for Earth satellite imaging. Cody, how do we keep it that way with Landsat Next?
CODY ANDERSON:
Oh, I really enjoy that question. The gold standard is something that I take very personally and the EROS Cal/Val Center of Excellence, or ECCOE, one of our catch phrases or mottos or slogans or mission statements, really, is to maintain that gold standard. So looking forward to Landsat Next, everything that’s new on it still has to be at that absolute highest level that everyone else will continue to look at. We’re going to have to tighten up our geometry a bit because we’re going from 30 meters to 10 meters. And with the new spectral bands, we’re gonna have to come up with a couple of new algorithms in order to calibrate those. We’re adding some atmospheric bands so they won’t even see the surface, so some of our traditional techniques will have to adopt those a little bit, come up with new ones to really make sure that Landsat Next is at the same class at the same reliability, the same consistency as previous missions.
ADAMSON:
It sounds like this is a lot of new challenges. Some of the bands on Landsat Next are the same spectral bands as previous Landsats, but even with that, the resolution is different and you have to take care of that too.
ANDERSON:
Yep.
ADAMSON:
So that’s what you mean when you say the geometry is also something that you’re working on.
ANDERSON:
Currently we say that we’re accurate within about half a pixel for Landsat 8-9, which is half of a 30-meter pixel. We’re talking 15 meters. 15 meters is no longer good enough when your resolution is 10 meters. So we’re going to have to improve all those things, then flow all those changes that we’re making for Landsat Next back through the archive so that the whole archive stays internally consistent.
ADAMSON
Is there any way that you can test the new data before we actually have it?
ANDERSON:
Yeah, that’s activity of several different groups on the science branch at EROS as well as the calibration group at EROS and as well at NASA. We can look at different missions that are already in existence. Sometimes they’re called hyperspectral missions or imaging spectroscopy missions. So these are ones that really sample the entire electromagnetic spectrum. So all of those new spectral channels that we’re getting for Landsat Next, these other missions are kind of acquiring data over those same spectral channels, so we’re simulating test data for Landsat Next to use these other sources to simulate something that will look like Landsat Next. But it gives us an idea for what to expect. We can get some of that data out to users as well to get feedback on them. So we are creating test datasets or sample datasets that look like what Landsat Next will potentially be and we can use those to really prototype or set up some new algorithms.
ADAMSON:
And this work is already happening? We already have this simulated data?
ANDERSON:
Yep, Yep, that is already happening and has been going on for a year a little bit more.
ADAMSON:
With a new generation of sensor, how do you make sure the Landsat Next data will be consistent with data from previous Landsats?
ANDERSON:
We are adding a few new spectral bands so there is no history with those, so we’re kind of going in uncharted territory with those ones. But for the existing ones that we have, we are keeping all the bands we’ve had in the past, all of the reflective, right, you’re blue, you’re green, you’re red, you’re near IR channels. Your SWIR channels, as well as the thermal channels. So those channels that we’ve had before have to be consistent from what we’ve had in the past. That’s kind of been the crux of Landsat and is one of the key parts of that gold standard is, when it’s Landsat data, you can trust that it will fit with the data in the past. But there are some challenges there going from 30 meters to 10 meters. We need to make sure that those new pixel resolutions will stack, or lay on top of, the previous data. We’re coming up with new ways of maybe referencing the pixels even to make sure that those different pixel sizes will all lay on top of each other, so you can use older Landsat data with the new data. And for the radiometry, so those spectral channels that are the same that they’ve existed in the past, we will come up with extensive cross-calibration techniques. We’re going to launch three Landsat Next all at the same time, so we have to make sure that those three are all intercalibrated, that all three of them agree with each other. And then it’s assumed that Landsat 9 at least, I’m hoping that we get Landsat 8 and Landsat 9 still operational when we launch Landsat Next, and there will be campaigns of comparing Landsat Next with Landsat 9 and hopefully Landsat 8, and making sure that all those existing channels will be lined up and consistent across the entire archive. So this is what calibration is all about. This is what we’ve been doing with the Landsat archive for 50 years to make sure that when it says Landsat people know what they get.
ADAMSON:
Will Landsat Next do that turn and look at the full moon thing?
ANDERSON:
Yeah, yeah, we’re still planning that Landsat Next will continue lunar calibration. So it’ll still kind of perform at 180° maneuver to do all the moon calibration with all three satellites. And with those new spectral channels that we have, the Moon is one of those targets that we think will be really valuable. There’s no atmosphere on the moon, so all of those new channels that we’re adding that look at the atmosphere, they won’t see the surface of the earth. They’ll still see the surface of the moon, so we’re expecting the moon to be very valuable for us.
ADAMSON:
Of course we have that other podcast that we did a couple of years ago on that. Listeners should reference that one as well to get all the detail on how that works.
ANDERSON:
One thing that you need to keep doing when you’re talking about a gold standard is staying consistent with the past so people know what they have. You also need to be looking toward the future. So we need to keep innovating also as we’re looking forward. And so that’s all the new challenges that the Cal/Val team has are really kind of brought about because we’re enabling all this new science with those new channels. So all of these new challenges that are being put on the Cal/Val team, I think are really being brought about by enabling new science with Landsat Next.
ADAMSON:
It’s a lot of work that you’re gonna have to do in the next few years.
ANDERSON:
Luckily, we got a good team between the USGS and the NASA Cal/Val teams. I think we got the people we need in order to do those things.
ADAMSON:
I’d like to thank Brian, Chris, and Cody for joining us on this episode of Eyes on Earth. Sounds like we’re going to be really ready for Landsat Next. All the teams are already doing the work to prepare for the incredible new Landsat data that is coming. And thank you listeners. Be sure to go back and listen to Part 1 about Landsat Next and its benefits to science and society. And check out our social media accounts to watch for our newest episodes. You can also subscribe to us on Apple and YouTube podcasts.
VARIOUS VOICES:
This podcast, this podcast, this podcast, this podcast, this podcast is a product of the U.S. Geological Survey, Department of Interior.