People Behind the Data: Kelly Olsen on Finding Meaning in Change

From seismic research to satellite monitoring, Kelly Olsen helps EarthDaily customers understand change and decide where to look next.

EarthDaily’s scientists, analysts and engineers do much of the work of turning satellite data into answers customers can act on. With our new series, People Behind the Data, we’re introducing them, what they do and who they are outside of work. We’ll be featuring more of our team in the months ahead.

We begin with Kelly Olsen, Senior Applied Scientist and Technical Manager, whose background is in geophysics and seismic research. She leads a team working on deformation monitoring and change detection for pipelines and infrastructure. Here she talks about her path into remote sensing, the challenges of turning satellite measurements into results people can trust, and a colleague who occasionally joins her meetings uninvited.

Your career has taken you from geophysics and seismic research into remote sensing. What prompted that move?

Seismic data and geophysics are also forms of Earth observation, although they often solve different problems. My PhD research gave me a foundation in understanding how different areas may move and change over time.

I enjoyed the research, but I wanted to solve more problems that could help make people’s day-to-day lives better. After my PhD, I decided to move into remote sensing. There is so much open-source data available with global coverage, which allows us to tackle a lot of big problems.

My experience also helped me understand sensor limitations and how we can use proxy data to gain insight.

What are you working on right now that you’re particularly excited about?

I lead a team working on several projects, and one of them relates to pipeline and infrastructure monitoring. There are millions of miles of pipelines across North America, which makes manual monitoring a huge task.

We’re developing change detection algorithms to quickly flag subtle changes that may indicate a leak, spill or other incident. The aim is to help owners and operators respond more quickly and minimize environmental impacts.

When a customer brings your team a problem, what does your role look like from the initial question to a result they can use?

Everything starts with understanding what is feasible and how we can help. Customers typically come to us with a description of a problem or the solution they’re looking for. We then need to figure out which datasets might be useful, what methods we want to use and what the deliverable will look like.

I oversee a lot of those higher-level questions, delegate tasks, and also write and review code myself. I lead meetings internally and with customers to decide on technical requirements and help guide technical decisions.

Can you share an example of your work helping a customer focus on a problem?

A customer using our deformation monitoring product told us, essentially: “This is useful, but it’s too much data for me to look through. How can I focus on the hotspots?”

From that feedback, we developed deformation anomalies that can be used as alerts. They point to areas that appear to be moving differently from their usual behavior. For example, if a tailings dam or pit wall starts sliding, even a little, we can flag it as an anomaly.

We may also flag things the customer doesn’t need to worry about, but the alerts narrow down the areas they need to look at. We provide a PDF report for a quick overview and display the anomalies in our monitoring software, so customers can investigate further if they’re concerned.

That has helped customers get more value from the data quickly, rather than having to comb through all of it.

Has the data ever surprised you or challenged what you expected to find?

I work a lot with synthetic aperture radar, or SAR, data. Repeat satellite passes can be used to measure how the ground moves. This typically works very well in urban areas or places with less vegetation, but sometimes we don’t detect motion where we would expect to.

Most often, there’s a reasonable explanation, but sometimes it’s harder to understand why. We still trust the technology, but like everything else, nothing is 100% reliable. That can be difficult to explain to people who expect certainty from the data.

What is one of the biggest challenges in turning satellite data into results people can trust and use?

One of the biggest challenges is the lack of ground truth. Someone might ask us if we can detect something, but there are no examples or ways to verify that what we’re seeing is actually what we think it is.

For example, we might be looking for the construction of a small illegal airstrip or a small stand of trees blown down in a windstorm. If we can’t point to a known example of where that has happened, it’s very difficult to test and verify what we’re seeing.

We can’t know for sure that we would be able to detect something without testing it against a known example.

What might colleagues outside your immediate team be surprised to learn about you?

How demanding my dog, Juniper, can be. She thinks she’s part of the team, and if she hears me on a call, she insists on being let into my office if she’s not already there.

If you’re in enough meetings with me, you’ll know that I sometimes have to pause for a second to open the door and let her come in and do her part.