Studying the green wave: ten years of research software collaboration

For ten years, scientists in the Netherlands have used research software to study the seasons. The software has been developed by the Netherlands eScience Center, the world’s first national centre of excellence for research software, to meet the specialised needs of the research community, with computing infrastructure provided by SURF, the national research and education network for the Netherlands.

“As we are approaching fall, plants are losing their leaves. In spring, we will see them bloom. The science that studies these cyclical biological events is called phenology. The timing of these occurrences varies from place to place and from year to year, depending on the environmental conditions”, explains Raúl Zurita Milla, a professor of Spatio-temporal analytics at the University of Twente and pioneering user of eScience Center research software.

Green wave models

“We study changes in what we call the ‘green wave’: the transition from winter, when everything is dormant, to the arrival of spring, when everything comes back to life. We make maps that show this green wave moving from south to north. In the Netherlands, spring comes later than in southern Spain, where I am from, and it arrives even later in Norway.”

The green wave models are based on phenological observations from volunteers, Zurita Milla says. “You take your children to school, and you notice: hey, this tree didn’t have leaves last week and today it has. You can report this to a phenological network. These networks help to collect millions of observations.”

Zurita Milla and his team combine this information with environmental data such as temperature and day length to map the arrival of spring.

“Seeing how the green wave changes over time, gives us insight into the impact of climate change. It also has a practical agricultural application: farmers may need to plant at different times or use different varieties. And, as you can imagine, it would be a disaster if there was a temporal mismatch between blooming and pollinating insects, or between the arrival of spring and the timing of the last frost.

“There are health aspects too. Ticks also have phenology, and humans go into nature at certain times of the year. So phenological studies can help to forecast the chance of more bites. At the moment, we are also working on modelling pollen. So that patients with hay fever can take their medication on time or avoid high pollen areas.”

Because he wanted to do large-scale phenological studies over Europe and the continental US, Zurita Milla approached the eScience Center a few years ago.

“I had a lot of data. Our models were also relatively slow and we wanted to speed them up by running them on distributed systems.”

Figure 1: an example of aggregated green wave data over North America

Learning from each other

The technology that the phenology researcher works with has changed enormously over time.

“When I was a student we had a desktop application, we were told what buttons to press, and all data was analysed locally. During my PhD, I moved into programming to automate data analysis. Now I need to switch to big data programming frameworks because the data simply don’t fit on my computer. Researchers should not download data anymore but move their code to the cloud, where all the data are available.”

“Thanks to the collaboration with the eScience Center, I was a bit of a pioneer with this way of working at my faculty. Just before the COVID-19 pandemic, we started our own big data centre, because we realised that we had to invest in these technologies and tools.”

Ready-to-use tools for earth observation

For several years, Zurita Milla, the eScience Center and SURF worked in a so-called alliance project, that enabled them to learn from each other.

“They wanted to know more about geospatial data, and we wanted to acquire knowledge about big data solutions. We discussed the algorithms in and out with the research software engineers at the eScience Center, and SURF provided the computing infrastructure.”

With the insights from this project and similar ones, the eScience Center and SURF have now developed ready-to-use tools, infrastructure and storage for Earth Observation data.

“This has lowered the entry barrier to big data for the geospatial community.”

Founded to make itself redundant

“A centre of excellence for research software as a national organisation is unique in the world. Others envy that,” says Joris van Eijnatten, director of the Netherlands eScience Centre since 2020. The centre employs scientists with digital expertise, who help researchers develop research software.

It started with about ten people, now there are a hundred. The organisation therefore recently moved to larger offices at the Amsterdam Science Park. With SURF, the Center mainly collaborates around the use of SURF’s computing and data infrastructure. “In addition, we give joint training courses and exchange ideas around innovation and open science,” says Van Eijnatten.

“We aim to let this expertise land within institutions and among researchers.”

“Besides the collaborative projects we do with research institutions, we want to expand our training programme in the coming years. Such as deep learning, parallel programming and online collaboration. So that this expertise also lands within institutions and among researchers.”

The eScience Center was intended as a temporary entity when it was set up.

“We do not have a deadline, but it is conceivable that our mission will be accomplished at some point.”

Investigating new technologies

In the early days, the type of researcher knocking on the Center’s door was mainly from the STEM field. They were already convinced of the usefulness of digital research methods and could code themselves. “Nowadays, the emphasis is shifting towards the social sciences and humanities. That’s a completely different target group. It means we have to work at different levels and make our language more accessible. We do offer a certain advanced level; we don’t teach programming courses. But universities do that now.”

“Research institutions are now investing more in digital know-how, for example by appointing data stewards. The next step is for them to start investing in research software engineers. It is a process; we are far from there. Young researchers have more digital skills, but computational training should really be part of every university study. At the same time, technology is constantly changing. Universities themselves cannot keep up with all that. We research new things, such as artificial intelligence, machine learning, quantum technology and digital twins. So for now, there is plenty for us to do.”

Find out more about research software collaboration in the Netherlands

Published: 02/2023

For more information please contact our contributor(s):