Scientific Machine Learning at DMI
Hi!
Welcome to the website for people interested in scientific machine learning (SciML) at the Danish Meteorological Institute (DMI) - SciML is the discipline of combining machine learning with scientific computing see e.g. sciml.ai for an overview. This website serves to organise our work and allow others to join us.
If you’re based at DMI and would like to join you can simply make a pull-request to add yourself to this website: people.md
We meet the first Monday of every month (think “ML Mondays”) at 11am local time at DMI - everyone is welcome!
People
-
Leif Denby
Self-supervised/unsupervised learning, convolutional networks
meso-scale cloud organisation, denoising of LIDAR water vapour observations
-
Kasper Hintz
XGBoost, Random Forest
Quality control of smartphone pressure observations
-
Marion Devilliers
beginner, interested in CNN for downscaling
climate models
-
Peter Thejll
supervised learning of CNN, CNN for image regression (and classification)
Radiometry: determine terrestrial albedo from Moon images
-
Maxime Beauchamp
Supervised/unsupervised learning, Neural data assimilation schemes
Geosciences, CAO (Climate, Atmosphere, Ocean), Data assimilation, Spatial Statistics
-
Søren Borg Thorsen
Beginner
Post-processing of NWP data and downstream applications
-
Peter Ukkonen
RNNs, MLPs, CNNs
ESM parameterizations, e.g. radiative transfer
-
Phillip Aarestrup
LSTMs, CNNs
Forecasting and simulation of river networks
-
Tommaso Benacchio
Supervised learning, Physics-informed neural networks
dynamical cores
-
Ruth Mottram
Still learning, convolutional neural networks
Regional climate model emulation, Surface mass balance modelling in Greenland and Antarctica
-
Fabrizio Baordo
interested in supervised/unsupervised learning
ai data driven NWP forecasts, ML for remote sensing products and satellite data assimilation within NWP systems
-
Villy Mik-Meyer
interested in supervised learning
Data driven models for extreme WL predictions
-
Kristian H. Møller
Beginner
Atmospheric Dispersion
-
Tore Wulf
supervised/self-supervised learning, vision models, uncertainty quantification
earth observation (e.g. sea ice mapping in SAR imagery), data-driven forecast models
-
Carlos Peralta
Beginner/intermediate. Hands-on experience with most python ML libraries., Interested in supervised/unsupervised machine learning
Model verification and post-processing, Interested in data-driven forecast models, downscaling, time-series forecasting
-
Kristian Pagh Nielsen
Beginner (in AI)
weather and climate radiative transfer (forward) modelling, remote sensing (inverse) modelling
-
Thomas Bøvith
Machine learning
Nowcasting of precipitation, Data quality
-
Rashpal Gill
Machine learning
Hydrometeor classification, Data quality
-
Flemming Vejen
Machine learning, Data quality
Precipitation estimation
-
André Düsterhus
Interested in supervised/unsupervised machine learning
Effect of North Atlantic system on European Climate, Chances and limitation of ML methods
-
Nicolaj Hansen
Machine learning, Data quality, Emulators
Build emulators that can downscale precipitation over ice sheets
-
Michael Andersen
GNNs, XGBoost, Hyperparameter Optimization, ResNet/CNN, Word2Vec text embedding, OCR, Neo4j, GraphDB
Fraud detection, Image classification, Similarity/context searches
-
Henriette Rilling
Supervised/unsupervised learning, Physics-informed neural networks
Precipitation time series forecasting, spatial + temporal gap-filling, ML for radar products