The Value of Retrospection- A look back at DeFROST
2019 was the year when we started our adventure to build DeFROST from scratch. After around a year working on the project here at WeGaw, we have decided to do a little retrospective on the project and share it with you.
We hope that this retrospective will help us reflect on what worked well, but most importantly, what we could improve and how to do it. For you, our beloved reader, we hope that you can gain some insights how it works for a start-up to run an ESA demo project, or at least have some fun reading it.
First things first
For those of you unfamiliar with the project: DeFROST delivers near real-time snow data based on satellite data, ground data, and meteorological models. This project is a European Space Agency IAP business application program demo project, meaning is that it has been done within a European Space Agency framework and with support from ESA. For the project, we teamed up with WSL Institute for snow and avalanche research SLF to create an awesome validation system for our snow data. With our partners and the team ready, we had a year to work on this project.
So now that we know what we are talking about, time to retrospect!
The Good
Let’s start with some of the best things that have happened during the project.
- The Power of Agile
One of the first challenges that WeGaw found for this project is to set up a new team and find a framework for our collaboration. Luckily, we found a suitable methodology quickly. We decided on using Scrumban as our main agile management methodology, and it worked well for us since it is build to support development in a dynamic and fast-paced environment.
- The Help of Experts
The ESA demonstration project framework has also proven incredibly valuable during the project. ESA expects detailed documentation throughout the project and provides useful templates in order to create that documentation. These documentation templates provide many different tools that help manage the project, for example risk management tools that helped us to clearly identify potential risks, their likelihoods and their impact. This knowledge was a great support for internal strategy and decision-making.
Thanks to the support and criticism offered by experts at ESA, we were able to spot problems and shortcomings in the project early and navigate back to the desirable track.
- The Ecosystem
During the project, we have found a great ecosystem in Earth Observation, especially around snow data. The support and knowledge we found online has helped tremendously in making progress with our work, due to the big number of libraries, documentation and information that can be found. In addition, open source and open access communities are always ready to help when problems arise. We are proud to be contributing to this awesome cause by releasing parts of our snow detection system based on the Let-It-Snow algorithm by CESBIO back online.
- The Value of Validation
Thanks to the validation system I mentioned before, we have been able to benchmark our product against well-established algorithms from reputable institutions like SLF. This has helped us improve and fine-tune our algorithms and reach high accuracy for our validation. You can see some of the results in November and December in the table on the left. If you want to learn more about how we do our validation, our Product Manager Diego wrote an article earlier this year. You can find it here.
The Bad
As we all know all that glitters is not gold, and neither was our journey without obstacles. Here are some challenges we faced during the project:
- The Documentation Overload
In the good part of the project we have stated that we have found that ESA documentation has been a valuable tool to get the project done. It was indeed, but there is a caveat. The documentation framework provided by ESA is detailed and specific, and therefore extensive. For a small team like ours (and in a project with such a time constraint) the workload can be overwhelming.
- Clouds Gaps
Since we use optical satellite imagery to detect snow, clouds can pose a problem. The obstructed view on the ground means that instead of collecting observed data, we need to use our technique based on other satellite products and weather models to fill these gaps. Thanks to our validation system we know that we still reach high accuracy, but we want to go further. Therefore, we are currently working on new technology to mitigate uncertainties caused by cloud gaps and reach even higher accuracy based on observed data.
- Underestimating the time
A personal problem throughout the project (that many of you can probably identify with) has been that I underestimated the time it takes to do simple tasks. These underestimations tend to pile up and create delays in the project. In addition, estimating the time it takes to research and gather new knowledge is very complicated, and can have an significant impact on the deadlines.
In Summary
Retrospection is a useful exercise to identify strengths and weaknesses, to learn from past mistakes, and to move forward with new knowledge and confidence. For WeGaw, it is clear that we experienced a lot of challenges and identified ways we can and will improve. All things considered, I think that the DeFROST project has been a success. We have created a reliable and accurate product. We met the strict requirements used by ESA and delivered the product on time. And most importantly, we have a great team that is up for the challenges ahead in 2020.
One thing is certain: We will come to find many opportunities to improve in our next retrospection. Project work is a constant circle of improvement not only for us, but for companies of all shapes and sizes. A clear structure and organization is key, and looking back and reviewing what has been done in the past is a great tool for a clearer path into the future.
