Building earth and environmental data science capacity at tribal and hispanic serving schools - post 2

In a previous blog, I introduced a program that I developed in 2019 called the Earth and Environmental Data Science Corps. This program targeted students and faculty at tribal and hispanic serving institutions. It’s goal was to build capacity for faculty to teach and for students to learn earth and environmental data science at these institutions (and beyond).

In that blog I talked about:

  • What earth and environmental data science is (and why it’s important),
  • The various program components that drove the earth and environmental data science corps
  • I broadly discussed what elements worked and which ones could be improved
  • The importance of evaluation and adaptive program development and design.

Here, i’ll discuss how we delivered the program amidst the COVID 19 pandemic and infrastructure that we used in the program. I’ll wrap up with lessons learned.

I hope that these posts are useful to others building programs in this space. I’m using what I learned here as I develop the diversity components of pyOpenSci (my current Python open source software project that also supports open science!

Earth and Environmental Data Science Corps program design

Previously, I mentioned the importance of being comfortable with adaptive program development. Adaptive program development is driven by evaluation and data collection. It assumes that your first design is not the final design.

You check in with participants to see how things are going. And you make modifications as you go. I’ve always taught this way. You can build an amazing amount of trust with students if they know that you are listening to them and trying to address their needs (within reason).

This same approach was implemented in the earth and environmental data science corps (EDSC).

My original design - the one I developed with Jenny, was great. However, after working closely with faculty, we modified the entire program design to meet the needs of all involved. for instance, our fall/spring semester program turned into a summer program. Students had more time in the summer and fewer demands outside of the school year. Many of these students had family to support. Thus, a paid summer internship was appealing and more accessible than during the school year.

Timeline showing when each program element would occur starting with the carpentries training, then our training and then an applied internship. The program ultimately looked very different once we actually implemented it but conceptually was this type of design.
Our original program design was a 3-semester program. After working more closely with faculty and considering resource challenges at the schools, we opted to create a summer program to make student and faculty participation easier. Students often seek employment in the summers but are busy during the semester with school, family and often other jobs.

But also we were hit with another surprise -

Hello Pandemic!

Funny meme showing a woman in a work suite sitting at a professional desk with the words - 1st day of zoom meetings. Below she is now in sweat suit, flashing peace signs and sunglasses in a movie theatre seat. below it says 14th day of zoom meetings
The COVID-19 pandemic official hit the United States in March 2020. Our program started at the end of April 2020. Let #ZoomLife and quarantine begin. src

First, the program began in April 2020. The COVID-19 pandemic forced us to move the hybrid teaching model that I had researched and collected data on for several years, into a fully online environment. Literally over night!

Given it was a quick change, I went to the tool I knew best - Zoom to teach content.

SPOILER ALERT: Zoom was not good enough! More on why this wasn’t ideal in a separate blog.

Hybrid teaching CAN work if done well using the right tools and right design.

Second, our students were stressed given the pandemic. Underserved and underrepresented groups were hit EVEN HARDER than many of us. None of us were prepared. My heart went out to everyone there.

But, I was also stressed; as was my team!

Adaptive program development: modifying your programs as you go

We quickly learned that Zoom is not an ideal way to mimic a classic classroom teaching environment.
Students need to work in small groups to be empowered. And faculty need to have 1:1 interactions students to make them supported but also to keep tabs on student.

Some students stopped participating and it took us weeks to realize that was happening. Others were there every day but shy about asking for help in a large zoom call or in breakout rooms (understandably).

We had some successes in the first year but I knew that we needed to modify our approach.

Interactive online teaching platforms

During that year I started experimenting with other tools in my courses that allowed for more personalized online interactions.

I started using spatial chat with my professional students each week - for both teaching and office hours. And they really enjoyed it (see quote below from one student).

The spatial chat option was nice for one-on-one meeting time with the instructor. Whether or not you decide to stick with spatial chat specifically…

Spatial chat mimicked a more natural teaching environment. Students can move themselves around in the online space, work together and not hear other groups. Or they can chose to work alone.

They can also share screens.

Spatial chat allowed me to checkin individually with every student who needed help each week.

I will save the details of this teaching approach for another blog, but we ended up using this approach with some success in the EDSC program.

Thank goodness for technology!

Democratizing access to earth and environmental data science skills through open education

Lucky for us we had already committed to posting all of our lessons online as open education resources. This approach was also a way to handle the pandemic stress as it allowed students (and faculty) to work at their own pace outside of workshop sessions. Many of our students and faculty had families who were now at home full time.

I had I spent a few years designing and building this website to support our professional graduate program. This portal had hundreds of lessons focused on earth and environmental data science skills. I had hoped that if we published all of the EDSC program materials online, that others could benefit from them.

In the context of the EDSC these lessons were also available for faculty to use in their own courses.

The resources for th EDSC built on top of existing resources. Specifically we built out the introductory lessons in our intro to earth data science textbook to support the EDSC.

These lessons covered introductory concepts related to working with tabular (spreadsheets) data in pandas, time series, and the basics of spatial data manipulation in Python.

Note: the website is not currently under active maintenance.

Through open education, the program could scale, reaching hundreds if not thousands of people around the world.

Read more about publishing high quality open education materials here

But something was still missing. Students and faculty needed somewhere to work on the lessons. A place where they could install Python packages and run code.

An online environment where students and faculty could code

One of the biggest challenges when teaching beginners to code is environment setup. This challenge is particularly large if you are using Python as your choice programming language. Python environments are notoriously tricky to setup.. Further, we couldn’t expect every student to have a computer that they could use to write and run code.

I had already built a JupyterHub to support the professional graduate program that I designed and was teaching at the time. And I had a tested Python environment that we used to run the hub backend. So, I thought we could use that same hub to support our earth and environmental data science students and faculty.

What is a JupyterHub?

A JupyterHub is an online cloud based environment that provides Jupyter notebooks and memory and processing resources that you allocate for students to use. The benefit of a JupyterHub for teaching is:

  • it removed the barrier of a student needing to have a particular type of computer
  • it removes the need to install specific tools on a computer.
  • it can be run from any device with internet - including a tablet!

JupyterHub is a great platform but wasn’t the right fit for this program

After the first year of the program, I realized that the JupyterHub was not a good long term solution for this EDSC program that served small schools with less resources.


We were maintaining it. Mostly I was maintaining it with amazing support from my colleague Karen. And it wasn’t easy or cheap to maintain outside of a funded program. And when it broke, it was stressful for me as a teaching. And for the students.

But also, students and faculty wanted to continue to use the hub after our summer program ended. And they couldn’t because I couldn’t afford to maintain it!

This meant that faculty couldn’t easily add this curriculum to their courses without first knowing how to

  1. setup python environments and
  2. support environment issues that students would inevitably have.

All of our earth and environmental data science participants were thus dependent on us. This platform was not a means towards capacity building!

A truly cloud based portable solution - Google Colab.

To resolve the issues above, I started to use Google Colab. Google Colab allows you to launch Jupyter Notebooks - from within Google Drive. You can even store data on drive and access it in these notebooks.

And you can install Python tools there too.

Google Colab was a winning solution for this program. And I ended up also using it in our professional program for the same reasons.

Putting all of the elements together

In the end, our EDSC program looked something like this:

  • We kicked the program off with online workshops using Spatial chat as a preferred tool
  • Students coded together using Google Collab. All data were stored on Google drive. This allowed students and faculty to take the coding platform with them after the program ended!
  • Faculty, rather than teaching each day, mentored the students. In mentoring they further developed their coding skills!
  • In year two, students from year one were paid to come back as peer mentors. This was a powerful way to build community and sense of belonging.
  • Faculty received training too - empowering them to add content to their existing courses.
  • Students engaged in real-world applied projects as a way to apply their skills.
  • All participants were paid for their time spent both in training, mentoring and project work.

All in all I think the program had some great successes. For one, several of the students did go on to graduate school after the program!

And I have been lucky enough to stay in touch with a few of them as well.

I learned a lot from the EDSC participants. Below I share a few of the core takeaways that have stuck with me the most.

Core Lessons Learned From This Program

Working with under-served institutions, students and faculty requires care, time and consideration. We learned a tremendous amount from the faculty and students who participated in this program.

1. Design & Build the Program With the Participants From the Beginning

I made the mistake of starting work on the proposal only weeks before it was due. It was my first NSF proposal and I didn’t account for the time needed to truly build relationships with our partners and include them in the core design. Because of this, I had to redesign the program once we began actual work. It was important to us that the faculty and students felt invested in the program and gained a large return from it.

Lesson learned: Don’t try to design a training program involving these communities without involving the students and faculty from the very beginning. While you might have considerable expertise in teaching, program design, and timing of activities… and you may have experience working in the diversity equity Inclusion and accessibility (DEIA) space, each group has individual needs. How content is delivered may all need to be adjusted depending upon your target audience.

2. Allow for a considerable amount of extra time when working with these communities

Time is critical to building trust with underserved communities. Spend extra time with students individually and in small groups. Get to know the issues they are dealing with that may get in the way of learning and participation. Accommodate these challenges in every way possible. This all takes time to do well.

Trust is critical.

3. Be Flexible to Adapt when things don’t work

I am a strong believer in adaptive program design and development. What this means is develop surveys for participants and collect data often and frequently. Then adapt and change based upon their needs and feedback. Finally participants know when you are making changes and why (because of their helpful feedback).

This not only improves your program but also helps participants feel more a part of the entire process rather than feeling “left behind”.

It also builds trust and respect. It enforces that you are there to help and support their needs. This, in my opinion, holds true in all higher education and training spaces.

4. Mentorship and Check-ins Are Critical

Setting up a support system for students is critical for project success. This support system could be multi-tier. In our case we invited students from previous years of the program to mentor in following years. While students can look up to their instructors, it’s even more powerful to learn from and receive support from their peers. This is particularly powerful if they identify with the background of their peers. And there aren’t many of their peers in that professional space to begin with.

Also ensure that students have structure. Schedule check-ins with faculty and instructors regularly. Make sure it is noticed, if a student doesn’t show up one day. Ensure that each student is seen.

5. Working Online is Hard - Zoom Doesn’t Cut It

When I talk with others about programs that have DEIA goals, the immediate response is often “oh, online approaches won’t work”.

While I do think there is a place for in-person interactions. I also think that immediate response to online learning does not take into consideration, one thing:

teaching online requires a redesign of your curriculum and how you teach / deliver content.

I will spend more time talking about this in another blog. In our case, we had to move our entire program online with a few weeks notice due to the pandemic. So we learned the hard way what worked and what didn’t work.

However, it helped us that in our professional program, I was already teaching in a hybrid in person / online setting. Thus, I had adapted a lot of my content to support online participants.

In year two of the program we moved from Zoom to an interactive online platform called Spatial Chat. I had started using this platform in my courses and found it to be a great way to recreate the classroom environment where you can check in with students individually or small groups. The platform also allowed student to work together dynamically during class and share screens.

This approach worked. Students were happy. And students learned a lot! In short online learning can work well, if you work hard to recreate a healthy classroom environment.

6. Make Sure You Have A (Relatively) Fail-Proof Compute Environment

In the early days of teaching earth and environmental data science in our professional graduate program, I created a Jupyter Hub (Many thanks to Tim Head and Karen Cranston for help with that effort!). What I loved about this platform was the ability to skip over troubleshooting student environment issues on their local computers during class.

However, the problem with the JupyterHub was that it wasn’t portable. Students couldn’t take it home with them. Faculty couldn’t use it in their classes the following year. This was because it was hard for me to maintain (there were costs!).

Google Colab is a free tool can be used from within Google Drive. These platforms are available to students at no cost. This allowed students and faculty to take all of the resources with them.

Since we were using Python, that meant Jupyter Notebooks that were stored in participants Google Drive (now Google Workspace). This was a huge win for everyone - especially faculty who waned to adapt the content into their own courses.

Wrapping This Up

The above is a summary of a program that had many pieces. I shared some of what I learned and some of the pain points that we experienced when running the program.

As I mentioned we had some great successes too. Several of our tribal students went on to Graduate school. Several returned in following years and were incredible mentors and advocates during their time in and after the program.

I will never forget my time designing and launching this program. All of the amazing people I got to know and work with and how much I learned from them all.

Notes & thank yous!

Some of the amazing faculty and students that I personally worked with on this project are listed below.

  • Elisha Yellow Thunder (A true leader and rock star, Oglala Lakota College)
  • Jim Sanovia (Oglala Lakota College)
  • David Parr (Metro State University)
  • Jeremy Guinn (United Tribes Technical College - now at NSF)
  • Emily Biggane (United Tribes Technical College),
  • Nate Quarderer (CU Boulder, who ran and taught a significant portion of the program in years 2 and 3 ),
  • Bill Travis (CU Boulder faculty representative)

Tags: data science , deia , earth and environmental data science , open-education , teaching

Categories: deia