KITKangu
Members
First name (team leader)
Asena Karolin
Last name
Özdemir
Organisation name
Karlsruhe Institute of Technology
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
First name
Charlotte
Last name
Debus
Organisation name
Karlsruhe Institute of Technology
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
First name
Lars
Last name
Heyen
Organisation name
Karlsruhe Institute of Technology
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
First name
Deifilia
Last name
Kieckhefen
Organisation name
Karlsruhe Institute of Technology
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
First name
Julian
Last name
Quinting
Organisation name
Karlsruhe Institute of Technology
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
First name
Sonal
Last name
Rami
Organisation name
Karlsruhe Institute of Technology
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
First name
Siyu
Last name
Li
Organisation name
Karlsruhe Institute of Technology
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
First name
Sebastian
Last name
Lerch
Organisation name
Karlsruhe Institute of Technology
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
First name
Tobias
Last name
Biegert
Organisation name
Karlsruhe Institute of Technology
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
First name
Julian
Last name
Stefinovic
Organisation name
Karlsruhe Institute of Technology
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
First name
Julian
Last name
Quinting
Organisation name
University of Cologne
Organisation type
Research Organisation (Academic, Independent, etc.)
Organisation location
Germany
Models
Please note that the model information below reflects the latest details provided by the team and does not show previous versions. Only the model summary questionnaires retain a submission history.
Model name
KanguS2SEasyUQ
Model type
Data-driven
Number of individuals supporting model development:
11-20
Maximum number of Central Processing Units (CPUs) supporting model development or forecast production:
8-48
Maximum number of Graphics Processing Units (GPUs) supporting model development or forecast production:
16-64
How would you best classify the IT system used for model development or forecast production:
High-Performance Computing (HPC) Cluster
Model summary questionnaire for model KanguS2SEasyUQ
The list below shows all model summary questionnaires submitted for this model. Participants may submit an updated questionnaire whenever their methodology changes.
Questionnaires are displayed from the most recent to the earliest, with the latest version opened by default.
Which of the following descriptions best represent the overarching design of your forecasting model?
- Machine learning-based weather prediction.
What techniques did you use to initialise your model? (For example: data sources and processing of initial conditions)
We used ERA5T data for initialization and normalized via z-score.
If any, what data does your model rely on for real-time forecasting purposes?
No additional data apart from initialization data.
What types of datasets were used for model training? (For example: observational datasets, reanalysis data, NWP outputs or satellite data)
Reanalysis data
Please provide an overview of your final ML/AI model architecture (For example: key design features, specific algorithms or frameworks used, and any pre- or post-processing steps)
We used PyTorch as a framework to build our model.
As an architecture, we implemented a 2D SWIN Transformer.
We used the time as a variable.
We converted the model's point predictions into probabilistic ones by assuming that the point prediction is the mean of a Gaussian distribution. The variance of that distribution is inferred from the climatology quintiles, again assuming that they are the result of a Gaussian distribution.
Have you published or presented any work related to this forecasting model? If yes, could you share references or links?
Architecture based on: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1714/
Before submitting your forecasts to the AI Weather Quest, did you validate your model against observational or independent datasets? If so, how?
No validation
Did you face any challenges during model development, and how did you address them?
It was odd that the AI-WQ-package included training data but not initialization. Thus, we had to reproduce the preprocessing steps in order to generate initialization data that matches the training data which was provided.
Are there any limitations to your current model that you aim to address in future iterations?
Currently the time dimension is not treated separately to the variable dimension, which we want to change in future iterations. Also, we are using a model that produces point predictions and we would like to convert our model to an ensemble.
Are there any other AI/ML model components or innovations that you wish to highlight?
-
Who contributed to the development of this model? Please list all individuals who contributed to this model, along with their specific roles (e.g., data preparation, model architecture, model validation, etc) to acknowledge individual contributions.
Data Preparation: Siyu Li, Julian Stefinovic
Model Architecture: Deifilia Kieckhefen, Lars Heyen
Training and Inference Script: Sonal Rami
Submission Pipeline: Asena K. Özdemir
Supervision: Charlotte Debus, Julian Quinting
Which of the following descriptions best represent the overarching design of your forecasting model?
- Machine learning-based weather prediction.
What techniques did you use to initialise your model? (For example: data sources and processing of initial conditions)
We used ERA5T data for initialization and normalized via z-score.
If any, what data does your model rely on for real-time forecasting purposes?
No additional data apart from initialization data.
What types of datasets were used for model training? (For example: observational datasets, reanalysis data, NWP outputs or satellite data)
Reanalysis data
Please provide an overview of your final ML/AI model architecture (For example: key design features, specific algorithms or frameworks used, and any pre- or post-processing steps)
We used PyTorch as a framework to build our model.
As an architecture, we implemented a 2D SWIN Transformer.
We used the time as a variable.
We converted the model's point predictions into probabilistic ones by assuming that the point prediction is the mean of a Gaussian distribution. The variance of that distribution is inferred from the climatology quintiles, again assuming that they are the result of a Gaussian distribution.
Have you published or presented any work related to this forecasting model? If yes, could you share references or links?
Architecture based on: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1714/
Before submitting your forecasts to the AI Weather Quest, did you validate your model against observational or independent datasets? If so, how?
No validation
Did you face any challenges during model development, and how did you address them?
It was odd that the AI-WQ-package included training data but not initialization. Thus, we had to reproduce the preprocessing steps in order to generate initialization data that matches the training data which was provided.
Are there any limitations to your current model that you aim to address in future iterations?
Currently the time dimension is not treated separately to the variable dimension, which we want to change in future iterations. Also, we are using a model that produces point predictions and we would like to convert our model to an ensemble.
Are there any other AI/ML model components or innovations that you wish to highlight?
-
Who contributed to the development of this model? Please list all individuals who contributed to this model, along with their specific roles (e.g., data preparation, model architecture, model validation, etc) to acknowledge individual contributions.
Data Preparation: Siyu Li, Julian Stefinovic
Model Architecture: Deifilia Kieckhefen, Lars Heyen
Training and Inference Script: Sonal Rami
Submission Pipeline: Asena K. Özdemir
Supervision: Charlotte Debus, Julian Quinting
Model name
KanguParametricPrediction
Model type
Number of individuals supporting model development:
11-20
Maximum number of Central Processing Units (CPUs) supporting model development or forecast production:
8-48
Maximum number of Graphics Processing Units (GPUs) supporting model development or forecast production:
16-64
How would you best classify the IT system used for model development or forecast production:
High-Performance Computing (HPC) Cluster
Model name
KanguPlusPlus
Model type
Number of individuals supporting model development:
11-20
Maximum number of Central Processing Units (CPUs) supporting model development or forecast production:
8-48
Maximum number of Graphics Processing Units (GPUs) supporting model development or forecast production:
16-64
How would you best classify the IT system used for model development or forecast production:
High-Performance Computing (HPC) Cluster
Submitted forecast data in previous period(s)
Please note: Submitted forecast data is only publicly available once the evaluation of a full competitive period has been completed. See the competition's full detailed schedule with submitted data publication dates for each period here.
Access forecasts data