WindBorne
Members
First name (team leader)
Todd
Last name
Hutchinson
Organisation name
WindBorne Systems
Organisation type
Small & Medium Enterprise or Startup
Organisation location
United States of America
First name
Joan
Last name
Creus Costa
Organisation name
WindBorne Systems
Organisation type
Small & Medium Enterprise or Startup
Organisation location
United States of America
First name
John
Last name
Dean
Organisation name
WindBorne Systems
Organisation type
Small & Medium Enterprise or Startup
Organisation location
United States of America
First name
Anuj
Last name
Shetty
Organisation name
WindBorne Systems
Organisation type
Small & Medium Enterprise or Startup
Organisation location
United States of America
First name
Haoxing
Last name
Du
Organisation name
WindBorne Systems
Organisation type
Small & Medium Enterprise or Startup
Organisation location
United States of America
First name
Chris
Last name
Riedel
Organisation name
WindBorne Systems
Organisation type
Small & Medium Enterprise or Startup
Organisation location
United States of America
First name
Lyna
Last name
Kim
Organisation name
WindBorne Systems
Organisation type
Small & Medium Enterprise or Startup
Organisation location
United States of America
First name
Jack
Last name
Michaels
Organisation name
WindBorne Systems
Organisation type
Small & Medium Enterprise or Startup
Organisation location
United States of America
Models
Model name
WeatherMesh
Number of individuals supporting model development:
6-10
Maximum number of Central Processing Units (CPUs) supporting model development or forecast production:
Unknown
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:
Small cluster
Documentation
Model summary questionnaire for model WeatherMesh
Please note that the list below shows all questionnaires submitted for this model.
They are displayed from the most recent to the earliest, covering each 13-week competition period in which the team competed with this model.
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)
ECMWF ENS analysis for the Thursday 0z cycle. Pressure level and surface atmospheric variables.
If any, what data does your model rely on for real-time forecasting purposes?
only initial conditions from above
What types of datasets were used for model training? (For example: observational datasets, reanalysis data, NWP outputs or satellite data)
ERA5
point observation datasets like METAR
IFS ENS analysis
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)
Transformer-based encode-process-decode architecture similar to the WeatherMesh-3 model
https://arxiv.org/pdf/2503.22235
Additionally adding a noise embedding and training with CRPS loss
This model is only trained on 3-hourly instantaneous or aggregate values for shorter lead times - we have rolled out the model to 32 days out, much longer than we expect it to be well-behaved, and then aggregated.
Have you published or presented any work related to this forecasting model? If yes, could you share references or links?
WeatherMesh-3 model is a previous iteration, published here
https://arxiv.org/pdf/2503.22235
Model code here https://github.com/windborne/weathermesh-3
Before submitting your forecasts to the AI Weather Quest, did you validate your model against observational or independent datasets? If so, how?
Validating on holdout ERA5 data
Validating historical forecasts against the observation data provided by WeatherQuest
Did you face any challenges during model development, and how did you address them?
No answer.
Are there any limitations to your current model that you aim to address in future iterations?
Training for longer lead times, incorporating more S2S-relevant input data
improving raw skill of ensemble mean, and better calibrating spread too
Are there any other AI/ML model components or innovations that you wish to highlight?
No answer.
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.
No answer.
Model name
WeatherMeshBeta
Number of individuals supporting model development:
6-10
Maximum number of Central Processing Units (CPUs) supporting model development or forecast production:
Unknown
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:
Small cluster
Documentation
Model name
WeatherMeshGamma
Number of individuals supporting model development:
6-10
Maximum number of Central Processing Units (CPUs) supporting model development or forecast production:
Unknown
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:
Single node system
Documentation
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