Hackathon 2025 – Introduce Yourself! 🌲👋

[sidekock]

Welcome to the introduction thread for the pySTEPS Hackathon 2025 (8–12 September, Château d’Avin).

Please reply with:

1. Who you are
   • Current role / affiliation
   • A sentence or two on your background
2. What you’ve worked on previously
   • Prior contributions to pySTEPS or other open-source / scientific-computing projects
   • Relevant research or industry experience
3. What you’d like to achieve at the hackathon
   • Features, bugs or documentation you’re keen to tackle (feel free to link to issues on the project board)
   • Skills you hope to gain or share

These details will help us match mentors, form coding squads and plan room assignments. Looking forward to collaborating with all of you!

— The RMI organising team

[vazquezjportillo]

🌀 Nowcasting Cut-off Low Flood Events in a Warming Climate (ardennes-2025)

Who I am
I am a Master's graduate in Meteorology and Atmospheric Physics, involved in early-career research and scientific publishing. I have a strong interest in synoptic and mesoscale dynamics, and I am passionate about applying theory to real-world forecasting challenges.

What I have worked on previously
I am the lead author on a paper currently under review by the QJRM, focused on baroclinic instability and the role of the beta effect. I am also working on a second paper analyzing frontogenesis using the quasi-geostrophic approximation. While I have not contributed to pySTEPS yet, I have a solid Python background and hands-on experience with atmospheric modeling and data analysis in a scientific context.

What I would like to achieve at the hackathon
My goal is to set up a case-based evaluation of nowcasting techniques (radar + ML-based) for recent flood events triggered by cut-off lows and omega blocks, specifically:

•  The October 2024 floods in eastern Spain
  

•  Medicane Daniel and the associated flooding in Libya/Greece
  

I’d like to:

• Prototype a reproducible workflow in pySTEPS for these events
  

• Compare performance of deterministic and probabilistic nowcasting during synoptic blocks
  

• Test ML-based nowcasting under slow-moving systems
  

• Contribute visualizations or Jupyter-based documentation for broader use
  

With global warming increasing water vapor availability in the atmosphere, improving real-time flood forecasting during blocked/cut-off low situations is more important than ever.

I am keen to learn the pySTEPS internals, contribute to the project’s scientific applications, and collaborate with others working at the interface of meteorology, hydrology, and Python-based modeling.

[alanseed]

Who I am

Retired from the Australian Bureau of Meteorology in 2020 and now work on the things that interest me. I have a strong interest in modelling the statistical structure of rainfall in space and time and quantitative radar rainfall estimation.

What I have worked on previously

I am the original developer of the STEPS framework for generating stochastic rain fields. This year I have been working on a parametric approach to generating the stochastic fields and a MongoDB database to manage the pysteps data.

What I would like to achieve at the hackathon

Mostly to assist with the maintenance of pySTEPS and to progress with integrating the parameteric algorithms into the package.

[m-rempel]

Who I am

I am meteorologist who worked for eight years in DWD's SINFONY project on the combination of NWC and NWP. Currently, I am working on generating realistic precipitation structures from marginal distributions and have started working on ML-based NWC approaches.

What I have worked on previously

I have implemented and enhanced Daniele Nerinis combination technique. I have also worked on object-based verification and statistical post-processing of ensemble forecasts.

What I would like to achieve at the hackathon

My aim is to integrate Danieles origin approach into pySTEPS and to introduce my further developments to provide a solid evaluation reference for, e.g., ML-based NWC/combination techniques.

[Loickemajou]

Who I am
I am a Master's graduate in Applied Computer Science from Vrije Universiteit Brussels (VUB). My main interest lies in applying deep learning techniques to solve real-world problems.

What I’ve worked on previously
I completed an internship at the Royal Meteorological Institute (RMI), where I aimed at integrated deep learning nowcasting models as a plugin within pySTEPS.

What I’d like to achieve at the hackathon
I would like to work on prototyping new features in pySTEPS, specifically focusing on neural network-based nowcasting.

Skills I hope to gain or share
I hope to deepen my understanding of the pySTEPS framework and contribute to enhancing its models for improved weather nowcasting through new features.

[Joep1999]

Who I am
I work at Weather Impact, a company focussing mostly on providing weather forecasts and agro-weather advisories for farmers in Africa. However, we also do some more research-focused projects in the Netherlands. My job title is 'Consultant weather and climate change' but actually I am just a programming meteorologist that also does some teaching and report writing.

What I’ve worked on previously
The most relevant work I have done for this hackathon was during my internship at Weather Impact, which was on comparing deep-learning nowcasting methods with pysteps and regular extrapolation-based methods over the Netherlands. I am currently working on a project for the ECMWF, in which the goal is to use an improved version of the blending methodology of Pysteps to blend DGMR with their new 'DestinE ExtremesDT' model.

What I’d like to achieve at the hackathon
The first part of this project is to rebuild the pysteps blending algorithm so it can use DGMR instead of only being able to use pySTEPS. last week I worked on creating a crude first version of this, which I hope to implement properly into pysteps during the week.

Skills I hope to gain or share
I hope to get to know the pySTEPS modelling community better, while also learning more about the several features of pySTEPS as a whole, but especially about the pySTEPS nowcasting and blending algorithms.

[AVI68]

Who I am
Hi everyone, I’m Avijit Majhi, a PhD candidate at the University of Cagliari (Italy), currently a visiting researcher at TU Delft (Civil Engineering and Geosciences). My research focuses on radar meteorology and precipitation nowcasting, with a particular interest in improving urban rainfall predictions.

What I’ve worked on previously
I’ve worked on spectral blending to improve RainFARM, where we merge low-resolution, large-coverage C-band radar with high-resolution, small-coverage X-band radar to build improved composite fields. This technique can also be extended to merge X-band with low-resolution radar or satellite data, producing enhanced 2D rainfall maps that are highly relevant for local nowcasting. I have been an active pySTEPS user for ensemble nowcasting and spectral blending experiments. While I haven’t yet contributed code upstream, I’ve developed workflows, verification routines, and visualization tools based on pySTEPS that I’d be happy to share.

Relevant experience
I am currently working on different deep learning pipelines for urban precipitation nowcasting. As part of the GeoScienceIR project under the EU PNRR programme, we are building a GUI based on pySTEPS modules for our new X-band dual-polarisation radar, including data loaders, preprocessing, and nowcasting integration.

What I’d like to achieve at the hackathon
Learn more about pySTEPS internals and development practices. Contribute to spectral blending and verification modules. Explore ways to link pySTEPS with deep learning pipelines. Support documentation and tutorials, especially for real-time applications.

Skills I hope to gain or share
Looking forward to sharing expertise in radar merging and deep learning frameworks for nowcasting, while learning from the community on pySTEPS design and implementation.

Excited to work with you all in Château d’Avin!

[ladc]

Who I am

Hi everyone, I'm Lesley De Cruz. I'm a professor at Vrije Universiteit Brussel (VUB) and researcher at the Royal Meteorological Institute of Belgium. Since 2021, I've been leading a joint research team on AI for weather and climate modelling. At RMI, I coordinate the seamless prediction programme, project IMA.

I was trained as a physicist and got my PhD in particle physics in 2011. I've been working at RMI since 2012, where I've worked on a range of topics from dynamical systems theory, ocean-atmosphere coupling, data assimilation and regional climate modelling to nowcasting (STEPS).

What I’ve worked on previously

I was a lead developer of the MAOOAM coupled ocean-atmosphere model, which evolved into the more general QGS model in python. I've worked on pysteps blending with NWP and have co-organised the previous pysteps Hackathons at RMI. In a previous life, I also developed LGSL (Lua GNU Scientific library).

What I’d like to achieve at the hackathon

I'd like to finally finish the deterministic member and @alanseed has made the great suggestion of simply maxing out the autocorrelation coefficients to achieve this. I'm also curious to learn about mongoDB. Finally, I'd like to contribute to making pysteps fully interoperable with its ML sister package, MLCAST.
Sounds totally doable in one week, no? 😁