New Space Blog

Perfecting EO Data Quality

Written by NewSpace Team | Dec 14, 2022 5:31:49 PM

Isn’t Earth Observation (EO) data great? We get to use satellites to visualize our planet in entirely new ways. In most cases, we are trying to detect change. The evolution and health of a large agricultural crop, for instance, or geographical changes to a coastal shoreline.

But detecting change is hard. You need to know exactly what you’re looking at and exactly what’s different from the last time you looked at it. And our planet isn’t exactly a still-life model; it’s ever-changing.

The quality of EO images can vary from day to day, and from system to system. With these inconsistencies, it can be hard to reach a solid conclusion.

“If you have images which are improperly correlated in terms of their quality, what you think is a change is actually a change in the quality from a day-to-day basis,” says Don Osborne CEO of EarthDaily Analytics.

Thankfully, EarthDaily Analytics has a solution for the data quality conundrum: “We are imaging everywhere on Earth every day.”

 

Getting the Right Perspective

“The cornerstone of our system is about quality,” Don says. “It’s the sensors that we use. It’s the quality of the satellite we use. It’s the way we fly the satellite. All of that gives us … quality of the pixel stream.”

“We have 22 spectral bands. So that means we’re taking imagery in 22 different ways, which gives us 22 sources of information.” These data streams, which include such information as methane and thermal measurements, essentially enable EarthDaily Analytics to look at the same image from 22 different angles.

“Part of this is utilized to correct for other areas,” Don says, as continuous cross-calibration is built into the system to ensure precision data accuracy.

 

Precision Consistency

End users of EO data need to be able to trust that the data they’re using to make decisions is accurate.

That’s why EarthDaily Analytics puts an emphasis on temporal consistency—“looking … from the same point in space at the same time every day.” So, users can be sure that the image they’re looking at from a year ago can be effectively compared to the more recent image.

“We actually take a picture looking straight down on the Earth at 10:30 in the morning every day,” Don says. “So, shadows are consistent. The light and the angles of the sun. All of that is consistent on a day-to-day basis.”

 

Global Coverage, Every Day

“We put up a number of satellites,” Don says, “and the way they rotate allows us to look everywhere, every day. So, it’s basically 240-kilometer strips of the Earth, which we then aggregate and assemble to produce images as we go.”

The reason for this everywhere on Earth every day strategy? “We don’t think anybody’s cracked the code on how to do broad-area change detection,” Don says.

“When you’re talking about change detection, you don’t always know when you get to a future point in time where you want to compare back to,” Don says. “If you come out a year from now and you say, ‘I want to compare back to a year ago,’ we actually have the image, where other … service providers wouldn’t necessarily have that available.”

 

Prodigious Processing Power

Of course, processing this “huge amount of information” is a big job. And making it useful to end-users is even harder.

“We built a virtual environment to do the corrections,” Don says. “We correct for cloud coverage. For aerosols in the atmosphere, which … distort some of the imagery. And we’re able to do this using a fully automated processing system that we’ve spent about seven years developing.”

In the end, the system creates “what we call a file of analytics-ready data, which is ready for machine-learning or AI to then have a look to see what the changes are from day to day.”

 

Powerful Use Cases

EarthDaily Analytics plans to launch a constellation comprising 10 total satellites by the middle of June 2024. And as the company focuses mainly on commercial and agricultural business opportunities, Don sees a robust market moving forward.

“I think anybody who’s going to be doing any significant business in agriculture will be using a data feed like ours or similar to ours” in the future, Don says.

The EarthDaily Analytics business model can be useful in defense and intelligence, as well. “One of the things we can do because we’re looking at everything every day is, we can be used for what’s commonly called tipping and queuing,” Don says.

For instance, EarthDaily Analytics can monitor a camp in the wilderness that has been deemed a potential source of criminal or terrorist activity. “We can actually signal that we’ve seen more cars come in. Now, we can’t read the license plates, but we know there’s more comings and goings. Therefore, now you send in the high-res asset to … see exactly what’s going on.”

In the end, Don says, EarthDaily Analytics's work is poised for big things. “You’ve got a number of big, global problems. And we believe that a global change detection system based on Earth observation can contribute significantly to solving these problems.”