Eyes on Earth Episode 32 - Lunar Calibration

TOM ADAMSON

Hello everyone. Welcome to another episode of Eyes on Earth. This is a podcast that 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, your host for this episode. Today's guest is Cody Anderson, the EROS Calibration Validation Center of Excellence Project Manager. Cody, some of our discussion will be on how accurate Landsat is. There are a lot of other Earth observing satellites. What makes Landsat unique?

CODY ANDERSON:

I think one of the best things about Landsat is the age of it or the age of the entire archive. It goes back to a time where Landsat was really unique and one of the only ones up in orbit taking imagery at a global scale. It's the only one that has continued since that time. There are still new Landsats being launched today and acquiring data. We have the only continuous earth archive back from 1972 when we launched the first one. 

ADAMSON: 

So, this super accurate record of earth imaging spans nearly 50 years and it's global imagery. What is one of the main strategies that keeps Landsat data so accurate? 

ANDERSON: 

I guess that would another one of the unique things about Landsat. I know a lots of other missions out there, they go and look at their entire archive and reprocess and do things. But, that's really been a core to Landsat data, for probably 2-3 decades now. We look at all the old Landsats and we compare them to the newer ones. We do this kind of large cross calibration approach where we are looking at the best data we have from the newest Landsats. Landsat 8, which is our most recent one, has seven onboard calibrators, multiple lamps, solar diffusers, shutters, onboard black bodies. And then we use multiple vicarious techniques called Pseudo Invariant Calibration Sites, global land surveys for the geometric reference. We have been sending field teams out for many years with handheld spectrometers making measurements. Then we flow the data from the newer satellites all the way back to the older ones, so it's really a continuous dataset from the first Landsat up to Landsat 8 and then the future Landsats that we are planning.

ADAMSON: 

We are always looking at the older Landsat data to make sure it is still accurate compared to the newer Landsats. Is that right?

ANDERSON: 

Yes. Absolutely. The first ones were collected back in 1972. We are ready to release the Landsat collection 2 data. Involved in that will be a reprocessing of that early data. We are always looking at it, Landsat science team, the Landsat user community still downloads it, they still use it. It is not being forgotten. 

ADAMSON:

There are a lot of methods of calibration that you talked about. There is another method of calibration where you look at different sites across the earth that are very stable like very dry locations. The term you used was Pseudo Invariant Calibration Sites. So, Landsat images those sites, right?

ANDERSON: 

Yes. Our favorite one is called Libya-4. It is out in the Libyan desert. Every time a Landsat satellite passed over that site, we would acquire an image. This is even back in the '80s and early '90s where we could not get all the data at EROS. We were limited on the number of acquisitions that we could take due to the onboard memory of the actual satellite. This is where the International Cooperator agreements came into place where we were actually downloading Landsat data to multiple other countries and institutions throughout the world. Most of the time that data would not actually come back to EROS. The data we always made sure to get was Libya-4. I am sure some time in the future people will be looking back at Landsat collections and realize that, "What was happening over this Libyan desert that we have 10 times more images of this Libyan desert than anywhere else in the world?" Well, it's our favorite PICS site. That's the reason. 

ADAMSON: 

Yes. It would be kind of weird to see that nothing happened in this region. It hardly changed at all. 

ANDERSON: 

Yes. Every picture looks exactly the same. That's the reason we wanted it. 

ADAMSON: 

Yes. At some point though, somebody probably said, "Hey, I know a place that doesn't change. Has no vegetation, it doesn't even have any atmosphere. Maybe we should image the moon." How did that even come up? 

ANDERSON: 

The moon is a great one. No atmosphere is very key there. The surface doesn't change. People have been looking at it for centuries. Lunar calibration really got started with what is called the ROLO Project, or the Robotic Lunar Observatory, which is down at USGS Flagstaff, Arizona. That started from a NASA grant back in 1995. The entire mission of that was to record many, many images of the moon, model out the moon's response throughout the whole waxing and waning phases, the different crescent moons and the full moons. The moon doesn't change, the surface.  But the orbit of it, it wobbles a bit. They call it librations, where you are seeing different parts of the moon. There is actually a pretty involved model behind the lunar response.

ADAMSON: 

Ok. But it still does help us get accurate earth data, the moon helps us do that.

ANDERSON: 

Yes.

ADAMSON: 

So, Landsat 8, is Landsat 8 the first Landsat to do moon imaging?

ANDERSON: 

Yes. It is. It kind of started back in 1995. But, Landsat 8, launched in 2013, was the first Landsat that actually acquired the moon. MODIS around 2002 or 2003 beat us to the punch a little bit on that. 

ADAMSON: 

The MODIS sensor that is on the NASA satellites Terra and Aqua, MODIS has imaged the moon too, right?

ANDERSON: 

Yes. 

ADAMSON: 

So, Landsat 8 turns to look at the full moon once a month. Does it also howl and grow hair and big teeth and claws?

ANDERSON: 

I don't know that it is quite a werewolf, but I do know that the entire maneuver that we need to do in order to rotate the satellite to acquire an image does make the flight ops team irritable every now and again. Maybe, they are wolves about once a month getting upset with us.

ADAMSON: 

Ok. Maybe this will get at what makes them a little irritable, what is exactly involved when Landsat 8 does this turn to look at the moon? 

ANDERSON: 

It is not too difficult. It's pretty easy. There is about a 70-page document and a couple of 10+ term equations that the NASA Jet Propulsion Laboratory puts out. Automatically, you run these equations and it cranks out the position of the moon. So, the flight ops team issues commands where the satellite will pitch and roll in order to acquire the moon. It's really not that difficult. Just a couple long equations.

ADAMSON: 

Just a couple long equations, easy enough. I can't help but picture some guy sitting down here with a joystick and carefully moving the satellite. But that's not how that works.

CA: No, no joysticks involved. 

ADAMSON: 

Ok. Mostly it's these equations, these commands that have to be sent up to the satellite?

ANDERSON: 

Yes. 

ADAMSON: 

How do those commands get up to the satellite?

ANDERSON: 

The flight ops team, which is underneath the Landsat Mission Operations Project. Most of these team members actually sit out at NASA Goddard. They are the ones that are routinely monitoring the satellite. They tell it when to image, when to download the image and when to do these extra maneuvers that they normally don't do. Cal/Val team, we like to play along and challenge people from time to time. Usually these extra maneuvers come from us. The flight ops team would issue a command that says at a certain time you will move the satellite. You will pitch the satellite for a certain amount of time. You'll stop that and then you'll roll the satellite for another certain amount of time. Then, the moon should be there. We really don't know until we get the final images down to the ground and can process them. We haven't missed it yet. I guess we have a good track record.

ADAMSON: 

How much fuel does it take for the satellite to do that maneuver?

ANDERSON: 

It actually doesn't take any fuel. The only fuel that we use is when we are actually raising or lowering the orbit of the satellite. If we have to push it down or push it up. For these spinning, pitching, rotating, yawing maneuvers that we do, they actually use something called a reaction wheel. These are wheels that are physically on the satellite and they are always spinning. That is how it maintains its stability and direction it is pointing. Then if you change the direction that these wheels are spinning, that will cause a rotation of the satellite. So, it is actually by these wheels that are always spinning and you turn one and that will make the satellite respond in kind. It doesn't take fuel. It takes a little bit of electrical power. That's another thing that we have to watch out for when we turn the satellite to look at the moon, we are actually moving the solar panel as well. It would normally be pointed at the sun to get electrical energy. That's also pointing away from the sun now, so we can't do this maneuver for too long a time or else we won't get the power from the sun. 

ADAMSON: 

Do we worry about losing images of the land while we do this lunar observation?

ANDERSON: 

No, we don't lose any land imagery. If you can kind of think of how the sun, the earth and the moon are all aligned and how they work out. When we have a full moon, that is actually when the moon is behind the earth in reference to the sun. So as the sun, the earth and then the moon kind of in line, we are actually imaging the moon when we are on the dark side of the earth, on the ascending orbit from Landsat. So, we can take an entire swath of earth imagery when the earth is daylit. Then as the satellite passes onto the dark side where it would be dark, there is no light there, that's when we can actually do the maneuver and look at the fully illuminated moon. Then, we complete the maneuver and move back before we would need to acquire earth imagery again. 

ADAMSON: 

That sounds like good thinking. We only look at the moon when it is not scheduled to image the earth anyway. 

ANDERSON: 

Yes. Physics helps us out a little bit there. 

ADAMSON: 

Ok. We have to know a little bit about orbits and that's a good thing we have people who know that. So, the Landsat sensors are designed to image the ground from 438 miles away. How can it effectively image something that is 240,000 miles away?

ANDERSON: 

That is a very good question. If you try to think of this in terms of your camera, and if you are going to go out there and take a picture of your family or take a picture of the landscape or whatever you want to do, you know you have to use either auto focus or manually focused if you are looking at something close to you or you are looking at something farther away. If you have ever tried to go out there and take pictures, you know that focus can always mess you up. There is a huge difference. It is 200,000 miles of difference. But, 438 miles is also pretty far away. How we actually set up these sensors, it is called infinity focus. Basically, set up the sensor to focus on something far away and whether it is 400 miles away or 240,000 miles away, they both are far enough away that we don't have to mess with that focus. The real difference is just the size of the pixels that we get out of it. Normal Landsat earth imagery, a lot of people know that it is 30 meters per pixel. For the lunar imagery that we get, it is actually 16-kilometer pixels that we get when we process the moon. It's in focus and we don't have to do anything there, but the actual pixel sizes vary greatly. We only get a small portion of the earth when we take a single image. We acquire the entire moon within one shot within one scene and we get the whole surface of the moon.

ADAMSON: 

Ok. That's what makes the difference. You get a lot more area really when you are looking at the moon. Are those Landsat images of the moon publicly available?

ANDERSON: 

They are not typically publicly available. We are not hiding them or anything. As I said, it is in focus but the pixels are quite large. It's just not that clear, crisp, clean lunar images that we've seen. Right? Everyone likes that nice pretty picture of the moon. It's really not that. You can tell it's the moon but it's really not that impressive to look at. We kind of keep those internal in calibration. But, if somebody wants it, there is no reason that we are hiding it from them. 

ADAMSON: 

How are these lunar images used in calibration? Why is this important?

ANDERSON: 

I think a lot of the ideas we've been talking about up until now have been based on that PICS testing, or the pseudo invariant calibration site. You look at the same spot on the earth or in this case on the moon, and it doesn't change over time. You want a flat line. There are a few other uses that we've found for the data. One is to measure the sharpness of the imagery. If you are a photographer, one of the things you know we talked about that focus versus blurriness versus sharpness. That's a system parameter that we can measure. The moon is pretty round, pretty defined, right? It's dark space and then it is moon. It's a really rapid transition. We can look for that sharpness of those moon edges on something that is called the MTF or the modulation transfer function, which is kind of a pretty complicated engineering-speak way of saying, "How sharp is the imagery?" 

ADAMSON: 

We've been talking to Cody Anderson, one of the people at EROS who make Landsat one of the most accurate satellite systems ever to observe the earth. We hope you come back for the next episode of Eyes on Earth. This podcast is a product of the US Geological Survey, Department of the Interior. Thanks for joining us.