Small Steps, Giant Leaps: Episode 176: Calibrating Satellite Image Data

Host Andres Almeida: Image data from Earth-observing satellites like Landsat and the PACE mission helps scientists monitor everything from wildfires and hurricanes to crop health and changing coastlines. 

But before scientists can use those images for research, the data must be carefully calibrated to ensure the measurements are accurate and consistent over time. A lot of this happens inside NASA’s Goddard Calibration Laboratories at Goddard Space Flight Center in Maryland.

How does image calibration work? Julia Barsi, calibration scientist, is going to paint the picture for us.

This is Small Steps, Giant Leaps. 

[Intro music] 

Welcome to Small Steps, Giant Leaps, the podcast from NASA’s Academy of Program/Project & Engineering Leadership, or APPEL. I’m your host Andres Almeida.  

Today, Julia Barsi is here to tell us about the behind-the-scenes work that makes satellite observations reliable. 

Hi Julia, welcome.  

Barsi: Hi! Good to see you.

Host: Likewise! So, image calibration is really interesting. What does your work entail? 

Barsi: So, I run a collection of laboratories that do radiometric calibration and characterization and reflectance characterization of materials for various types of instruments. 

So, calibration and characterization are the processes that allow us to understand the properties and performance of an instrument, so that we can interpret those results for scientific application. 

There’s many types of calibration and characterization that go into the process of building an instrument, but in my labs, we do radiometric and spectral calibration and characterization. 

The radiometric calibration establishes the relationship between the sensor response and a physical parameter – in this case, radiance, or the amount of energy that is hitting that instrument – and we work in the UV, visible, near-infrared, and shortwave. And so, for every wavelength of light, we want to know exactly how that instrument responds. 

Being able to represent that signal as a physical parameter means that scientists can quantify the processes of what they’re seeing in an image. So, then that applies for the single point image or over time, where you’re trying to stack up multiple images to see what happens over time. And that then allows us to better understand the processes going on on Earth that we have captured in these images. 

So, to do that, I have three labs at Goddard that perform various types of this radiance reflectance and spectral calibration for spaceflight hardware but also ground-based radiometers and spectrometers and airborne instruments that we put in an airplane and fly occasionally. 

To do that, we have a collection of sources and detectors in the labs that are NIST traceable, that means their response is what is known and traceable to a common standard. In the United States, that’s to the National Institute of Standards and Technology baseline, and so any instrument that has NIST traceable standards is comparable. 

We can say that we can compare one instrument measurement to another instrument measurement through that traceability. And so, the collection of sources and detectors that we have in our lab allow us to give that traceability to to these flight instruments, we also do reflectance and and transmission measurements. 

That’s normally on materials, they’re commonly on solar diffusers, which go on spaceflight instruments as a calibration target, but also materials like baffles that you would use inside a telescope, you want to know the reflectance of that material to know whether it’s going to scatter light throughout your telescope or whether it’s going to stop the light from bouncing around. 

So, an instrument team can come to my lab for a request for calibration or characterization, and we work with them to build up the appropriate test plan for their spectral range, brightness range, environment, and then we work with them to get the test implemented. 

Host: How do you ensure the satellite data remains consistent and accurate? 

Barsi: For on-orbit instruments, you really need to start with a rigorous pre-launch calibration and characterization, so you understand everything about your instrument before it goes into space. 

Once it’s on orbit, there’s things you just can’t back out from the information you get on your satellite platform, so we work with the instrument developers to make sure that they’re doing all that characterization and our labs can provide those sources and services to do those characterizations. 

This includes meeting the instrument at their TVAC chamber, where they are doing an environmentally controlled calibration, and we bring the source to them, so that they can test their response in different temperature regimes, so that if something on orbit changed and the temperature of their detector was had changed by a little bit, we would have already understood on the ground how that temperature change could affect their responsivity, so the pre-launch part is key. 

NOAA and NASA instruments are typically built with a number of calibration devices to ensure their stability and accuracy once they’re on orbit, but we also use ground-based measurements to verify and validate. 

And then my lab is also involved with the calibration and characterization of those instruments, so the ground-based instruments that are measuring the surface that then we try to map that to pixel in an image or an airborne instrument that’s going to underfly a satellite, so that on the same day at the same time we have satellite instrument and airborne instrument images, those now are tied both. Both of them are tied to a NIST traceable standard. 

And you can compare them with confidence, and so that that traceability between stand our lab standards and the instruments allows you confidence in being able to compare between sensors, not just in space but in the field and in the air. 

And then that that continues over time, so we can have a field measurement, you know, every day for 10 years, that three different instruments are comparing to, and that helps us track the consistency and accuracy across time. 

Host: How do scientists calibrate sensors in the Goddard Calibration Labs? 

Barsi: So, I have the three distinct facilities, and so I’ll separate them into two parts, basically the Radiometric Calibration Lab and the GLAMR Calibration Lab are provide bright sources of radiance for an instrument to look at and take images of to map to do to apply the calibration radiance to digital counts and any other additional characterizations for their, for their test plan. 

So, an instrument might come to us and say, we want to calibrate from 300 to 800 nanometers, and we will set up their test plan based on what they need, do they want to do, they need to monitor duct stability, do they need to monitor linearity, do they need to do polarization characterization. There’s a lot of things, a lot of different kinds of tests we can, we can help them with, and then they bring their instrument to us, or we go to them, if necessary. If it’s going to be in a TVAC chamber, we can move our hardware to them. 

Host: And TVAC, meaning “thermal vacuum chamber.” 

Barsi:  Yes. The thermal vacuum chamber gives us “test as you fly” spacelike environmental conditions. 

So, for spaceflight hardware, it’s preferable to test your hardware in TVAC, because then it is in the configuration in which it will be when it’s on orbit at the operating temperature, which can change things in an instrument and a telescope, and at the pressure, which can also and then warp things. 

So, ideally for spaceflight hardware, you’re testing in a TVAC chamber, which means we’re not testing here in my labs. We’re moving my hardware to the TVAC chamber and testing in front of the instrument while it’s under vacuum and temperature control, and we work with the instrument team a lot to develop the right test. 

And then once our light source is in front of their instrument, there’s a test plan that we follow to either provide whatever, sources of energy we are providing, either monochromatic or broadband light, we step through those and any other characterizations that we’re going to do. 

As I mentioned, we can do linearity, polarization, saturation, long-term stability, things like that. And so that test plan is all well established. When it’s in TVAC, you’re generally under a lot of pressure to get it done in a certain amount of time. TVAC is very expensive, so you got to get it right the first time, and so those tests are always very well thought out in advance. 

When they’re in our lab, it’s more casual. We can do go-backs, we can see if something’s not working, and we can make changes without the pressure of a TVAC campaign. And then in the reflectance facility, a team will come to us with the material and say, “I want to understand the reflectance of this material at this number of wavelengths at this number of angles,” and we will make those measurements for them. 

Host: What other big challenges are there in calibrating an Earth-observing instrument? And can you think of something that maybe taught your team, taught you an important lesson? 

Barsi: So, the big challenge is for us is when we are in TVAC getting that test right the first time. It means a lot of collaboration ahead of time to verify that our instrument, that our hardware is ready and that the instrument is ready, that everybody’s expectations are going to be met, but some of the challenges we’ve met is because the instruments that NASA and NOAA builds are just the best instruments out there. They are better than any of our ground system monitoring devices. 

So, so we’ve gotten into a TVAC test, and the instrument team comes to us and says there’s something weird about our data, and then that we have to work back and forth to figure out if it’s the image system or the calibration system. 

And then when it’s a calibration system, we have to be able to fix it in a, in a short amount of time, because, again, we’re on TVAC timeline, and it’s something that we couldn’t see our, when that happens, it’s something that we couldn’t see ourselves, because our ground system monitoring devices aren’t just not as good as the NOAA or NASA instrument, so this is this has happened a couple times recently. 

In one test, one space flight hardware test, we started seeing a new noise feature with an instrument that we had been testing with for, for a couple of years. 

And this noise feature wasn’t apparent in the previous test with this instrument, and so we had to work with the instrument team to figure out whether the noise feature was in the calibration hardware or in the instrument. 

And then once we did, the engineers worked together to rule out the different conditions of what it could be and tracked it back to an electrical circuit where one of our components was plugged in. 

And they had plugged something new into that circuit, which introduced noise into our electrical rack. And all we had to do was unplug our thing and put it in a quiet circuit, and that source of noise went away. 

Host: Fascinating. That was something you hadn’t seen before. And then… 

Barsi: Right, we had, we had been in this lab facility for a long time from operating in there for a couple of years, and we, we, we never actually figured out what it exactly was plugged, plugged into this circuit, but there was a lot of stuff plugged into that circuit, and just by moving it, we, we mitigated the problem. 

So, we never figured out exactly what was causing the noise. In another case, we were using a new laser system for a different particular wavelength range, and we’re in the middle of TVAC, so we had actually captured a lot of data already that was very good, and then we get to this wavelength range with this new laser, and it was really noisy, like off the charts, unusable data. 

And the image instrument team came and said, “What happened?” And it was clear in this case it was a result of this new laser system, and we had an idea that there’s this artifact that we, we couldn’t detect with this with our devices in this laser system. 

So, we’re, we take a break, we go to dinner, we think about what we can do to reduce this, this effect, it’s called “speckle,” and what we had on hand. 

And because this is a TVAC test, it was a three-shift test, so somebody in the lab actually had their overnight kit with them, and I was like, “Well, my electric toothbrush could vibrate this thing,” so we pulled out this guy’s electric toothbrush, set it next to the fiber, and vibrated the fiber enough to overcome and then made adequate images for the instrument test. And then we had a rotating cycle of three electric toothbrushes that we used for the rest of of this wavelength range 

Host: I just love that it’s that kind of problem solving happens when you’re away from the office. Maybe you’re, sometimes it’s little napkin ideas, like, “Oh, what if we tried this?”

[Laughter]

Barsi: Yeah, I have, I have a napkin, actually, with some problem solving on it, and some place in my files.

Host: You really are a NASA person!

Barsi: Yep, yep.

Host: What makes your work rewarding? 

Barsi: So, I got into this as a result of seeing a single image of Lake Ontario acquired by Landsat, while I was in high school. It was the coolest thing I’d ever seen. I said, “That’s that’s where I want to work. I want to do imaging from space.” 

And I, and I still feel that awe and excitement when I see images of Earth from space. The Earth is so beautiful, and we, we make such terrific images of it with all of the different instruments we have in space, and so the fact that I’m helping to generate those images and keeping those missions going and providing science data is, is still makes me, keeps me excited.

On a daily basis, the team that I have is so great to work with, you know, we constantly run, come across challenges that they’re like, “Of course, we can solve this. We’re NASA engineers.” And so, my really small group works together to solve problems, and then when we’re with an instrument team, where everybody is so dedicated to get the problems solved. That continues to be like rewarding every time, every time we do an instrument test, it’s like, “Oh, is this, is this going to go as well as the last one, because the last instrument team was really great,” and then it is! They are everybody who is so committed to these instruments that they, you know, that they are willing to support a TVAC campaign, really wants it to succeed, and then we get fantastic images out of it.

I’m the PACE…we calibrated PACE when it was here at Goddard before launch, and you know, I regularly get the PACE image digest, which is just amazing, 

Host: And you can be proud of that. 

Barsi: Yeah.  

Host: You also worked on Landsat. 

Barsi: I did work on Landsat. My first job was as a, as a Landsat radiometric analyst, and that was like 20 years of just making Landsat better and better, because I started with Landsat 7, and I, when I left Landsat, it was at Landsat 9, and it’s just a terrific archive that that mission has built up. 

Host: Julia, I have one more question for you, and that is: What was your giant leap? 

Barsi: My giant leap was pretty recent. As I said, I’d been working with Landsat for a long time. I started my master’s degree with Landsat calibration. 

And I did field work, and that was super exciting, being in the field every day, not always being tied to it to your desk. And you know, solving the engineering problems, just like I’m doing now. 

So that was my master’s degree, and then I came to, came to Goddard to work on the Landsat calibration team, which is primarily a desk job, and it’s great, but it was all at my desk, and during, during the pre-launch calibration of Landsat OLI-2 on Landsat 9, I got involved with the pre-launch calibration, the GLAMR system that I now run. It was my job as the Landsat person to interface with the GLAMR team, and it was so fun. 

So, eventually, a few years later, the opportunity came up to transition to running the pre-launch calibration facilities here at Goddard, and I, and I took that. And it’s really a stretch for me. 

I was in a comfortable analyst chair, where nothing much changed every day, and now I’m in this position where, you know, hardware could break at any moment, and we have to be able to figure out how to move on with or with a substitute. 

But it’s back in that dynamic environment like where I was in college, where it was not out in the field, but in the lab every day [where] something’s different. So, I really rely on my team, but that’s the excitement of this kind of job. Every day is really different, because an instrument can come to us on any given day with a completely new set of parameters that they need characterized. 

Last week, we got a request to do something with a giant field of view that we’ve never had to measure, and we’re like, “Okay, how do we do this?” And we’re going to figure it out, because we’re going to do the test next week,” you know? 

So, and then sometimes they say, “Okay, well, we did all these measurements, but we found something weird. Let’s figure out what that weird thing is. How do we test for that weird thing?” And we figure that out. 

So, it’s exciting to be able to solve these problems in this very dynamic environment and help every instrument get the best characterization and calibration possible. 

Host: Well, thank you, Julia. Thanks for your time. Thanks for explaining your work on image calibration, and how important it is. 

Barsi: You’re welcome. I do think it’s very important, so I’m glad you’re interested.

[Outro music]

Host: That’s it for this episode of Small Steps, Giant Leaps. For a transcript and to hear other episodes, visit nasa.gov/podcasts. While you’re there, check out our other NASA podcasts like Houston, We Have a Podcast, Curious Universe, and Universo curioso de la NASA. As always, thanks for listening. 

Outro: This is an official NASA podcast.