This month I attended the Earth Surface Processes Institute (ESPIn) 2021 summer course, provided by the Community Surface Dynamics Modeling System (CSDMS). Spoiler: It was amazing!
The event was virtual due to the pandemic but would have been in person in Boulder, Colorado, US. The group was about 25 graduate students and 4 incredible instructors. We met for 8 days, over two weeks, 8 hours a day, on Zoom. Despite being intense and tiring mainly because it was virtual, the group was very lively and cheerful, creating a light atmosphere. I could feel very comfortable with them, almost as if I was with close friends, even though I hadn’t ever met anyone before the course. Everyone was so friendly and nice! I was impressed with the course organization.
What did I learn at the ESPIn 2021?
We learnt about the basics of python programming, jupyter notebook, the shell, text editors, conda, version control with git and github, toolkits for modelling earth surface processes, Landlab model, how to create our own model (the Basic Model Interface – BMI – and Best Practices in Software Development) and high-performance computing.
Phew! That’s a lot!
Lots of content for only eight meetings in two weeks. The other students and I were beginners in basically all those topics, so we asked a lot of questions – from the silliest ones. Before this course, I had tried to learn some of these subjects by myself, but it was too confusing content to learn randomly by just googling it, so I had a lot of doubts. This course helped me a lot to clear up all these concepts and uses. Before that, I was all the time confused with the terminal, conda, git, pull, commit, push, fork, branch… Everything was a mess in my mind!
We used our learnings from the course to apply them to a team project on the last few days. It was time to practise and learn by doing! The idea was to learn how to do a collaborative project, and, in fact, I feel what I learned the most was to use git.
My team project was about hillslopes and channels. We used the SPACE and Network Sediment Transporter (NST) components of Landlab model to represent how soil erodibility changes due to fire on hillslopes increase erosion and sediment transport in rivers. SPACE simulates over long periods of time (centuries or millennia), while NST simulates on smaller time scales (some years). We created a way to take the hillslope erosion data generated by SPACE to be used as an input into the NST, coupling the models. The idea is to route the same amount of sediment supply from hillslopes into the same river network in both models and compare the results. The project has yet to be adjusted and will be published on the CSDMS website.
Table 1: Comparison of SPACE and Network Sediment Transporter (NST) Landlab component models.
|
SPACE | NST |
Data required |
· Topography / grid
· Soil and bedrock erodibility |
· River network / grid · Sediment parcel characteristics (location, volume…) |
Model resolution (timestep) |
Year |
Seconds or day |
Time scale | Centuries or millennia |
Days or years |
Figure 1: Illustration of how fires increase the supply of sediment in rivers.
Figure 2: Conceptual model of ESPIn 2021 team project ‘Hillslopes and Channels’: the object was to compare the sediment yield at catchment outlet resulted by Network Sediment Transporter (NST) and SPACE models, using the same sediment input into rivers and river network grid in both models.
How can I apply it to my PhD project?
First, all the git learning, and collaborative working is sure to be very useful to i-CONN, as our group also wants to work together. In addition, it was extremely useful to finally ask and clarify everything that I wanted about the terminal, conda, jupyter notebook and basics of python.
The team project that I developed in the course is closely related to my PhD Project. My PhD project aims to analyse sediment cover patterns in riverbeds due to external influences, such as variations in river discharge and sediment supply. I intend to use Landlab NST model to analyse the sediment cover patterns. I also think about using SPACE or HyLands to simulate the sediment supply from hillslopes into rivers. The ESPIn 2021 team project was helpful to learn how to apply these models and solve errors that I had when I ran the model locally on my computer. It was also helpful to learn that I can couple models relatively easily, using the hillslope component of another Landlab component to add sediments and route them in a river network using the NST. I also found out about other possibilities, such as the Sediment Pulser developed to NST, which I still have to explore and learn more about.
Figure 3: a) Topographic grid map and river network generated on SPACE; b) River network exported from SPACE and used on NST with links and nodes.
I think the main challenge that remains on my project is to couple models with different time scales and how to compare their results. Working with scales isn’t easy. Also, I still have a long way to go to adjust the models, test them and apply them to the goals of my PhD project. I will continue to apply and improve my skills acquired in ESPin 2021.
Conclusion
In summary, it was extremely valuable to have taken part in this summer course. I think it was important not only to develop my PhD project but also to develop my career further. The opportunity to meet several of the developers of Landlab and other Earth surface process models was especially valuable. I expect all the knowledge I gained about software development in this course will be useful for my career development. I felt this course was a kick-off for my training as a developer of Earth surface process models, providing all the foundation I needed. I am so grateful!
Mel Guirro