Ecological Forecasting Challenge Registration

JavaScript isn't enabled in your browser, so this file can't be opened. Enable and reload.

Email *

What forecasting challenge are you registering for? *

vera4cast

neon4cast

tern4cast

usgsrc4cast

model_id *

The model_id is the id that you want to appear on the dashboard

No spaces or - are allowed
It can have underscores in the model_id.
It can not include the word "example"

model_id should reflect a method to forecast one or a set of target variables.

Any variables that will be submitted in different files should have different model_ids

Important note for students in a university course: If you prefer not to include your name and email, please provide your instructor's email and name. Do not include any personal identifying information in the model_id or model descriptions.

Long name of the model *

Can include spaces

Contact name *

(or course instructor's name)

Contact email *

(or course instructor's email)

Can we publicly list your name and email as part of the model metadata? *

If no, the name and email are only available to the Challenge organizers.

Yes

Are the name and email for a course instructor? *

Yes

Institution *

If your forecast is being submitted by a team, who are the team members?

First Last, separate by commas. Leave blank if no other team members

Model description

Web link to model code

Ideally, a link to a GitHub repository or, if your GitHub repository contains the code for multiple models, a link to the specific function in the repository that is associated with the specific model.

Which category best matches your modeling approach?
*

Empirical (a statistical model)

Machine learning

Process based

Hybrid approach

Multi-model ensemble

Other:

Describe your modeling approach in your own words. *

Include the names of any established algorithms or functions in a package (e.g., XGBoost, Random Forest, ARIMA, etc.) that you used.

Sources of Forecast Uncertainty

If your forecast was generated from a multi-model ensemble, answer "Not Sure" for the sources of uncertainty below except for the uncertainty from multiple models.

Does the forecast use drivers? *

E.g., weather forecasts. Drivers are also referred to as inputs or covariates.

Yes

Not sure

Does your forecast include uncertainty from drivers (i.e., ensemble weather forecasts)?
*

Yes

What are your model drivers? *

e.g., variable type (temperature, wind, etc.). Use NA if no model drivers

Is your forecast model dynamic? (i.e. is tomorrow’s forecast dependent on today’s forecast)?
*

Yes

Do your forecasts include uncertainty from initial conditions? *

Uncertainty in initial conditions means that you have variation in the starting point for your forecast. If you start with a single value then your model does not have initial condition uncertainty. This only applies to dynamic models, see above.

(Use "NO" if you don't have a dynamic model)

Yes and they were estimated from data

Yes and they were not estimated from data (e.g., assumed initial conditions were the model equilibrium)

Does your model include parameters? *

Yes and at least one is estimated from data

Yes and they are not estimated from data

Does your forecast include uncertainty from the model parameters?
*

Yes

Do you update your initial conditions or parameters between forecast submissions using newly available data (i.e., data assimilation)? *

Initial conditions

Parameters

Both initial conditions and parameters

No updating or data assimilation was used

What method did you use if you updated your initial conditions or parameters using data assimilation? *

Does not apply

Model refitting

Kalman filter (ensemble, etc.)

Particle filter

Other:

Does your forecast include uncertainty from the model (process uncertainty)? *

For example, even after calibrating a model to data, there is still error present because the model is unable to fully represent the data. This uncertainty could be generated by sampling from the residual error in the model fit.

Yes and the uncertainty was estimated from data

Yes and the uncertainty was not estimated from data

Does your forecast include uncertainty from measurement noise? *

This comes from measurement error in the variable(s) you are forecasting. A forecast can have measurement noise added to a model prediction to account for the fact that a model may be predicting what is expected but that there is uncertainty when you actually measure it.

Yes and the noise was estimated from data

Yes and the noise was not estimated from data

Does your forecast include uncertainty from using different models?
*

e.g,. a multi-model ensemble

Yes

Does your forecast include uncertainty from parameter random effects?

This might occur if parameters randomly vary over time or space. This source is less common and typically estimated by a hierarchical model.

Yes and the uncertainty was estimated from data

Yes and the uncertainty was not estimated from data (uncommon)

Optional additional information describing how you determined the uncertainty in your forecast.

For example, if you are submitting a multi-model ensemble, include information on the models used (e.g., model_id if available) and the way that the models were combined (e.g., equally weighted).

Update to previous model_id registrations

This section is only required if you are updating a previously submitted model_id.

What is the previously registered model_id?

(the model_id can be the same if you want the registration to update an existing model_id)

Please describe how you have changed your modeling approach from the prior registration.

Participation Agreement

Do you agree to the following participation agreement? *

All participants agree to:

Have their registration information publically available. This excludes name and email if "No" was selected for the question "Can we publically list your name and email as part of model metadata?"
Allow the Challenge organizers to contact participants using the supplied email about updates to the Challenge, manuscript opportunities, and other correspondence related to the Challenge.
Have their forecasts posted in real-time on public storage, dashboards, and catalogs.
Have submitted forecasts and publically available metadata assigned Creative Commons License BY (CC BY). https://creativecommons.org/licenses/by/4.0/).
Have forecast submissions potentially published in a scientific journal. As part of the Challenge outcomes, manuscripts describing the accuracy of forecasts across teams will be coordinated by Challenge organizers with authorship extended to members of each team, with an opt-in policy. They will be contacted using the supplied email address. If your forecasts are used in manuscripts, code will need to be publicly available (unless not permitted by your institution or model license). These Challenge-wide analyses do not preclude individual teams from conducting analyses of their own submissions.
Cite the grant number associated with a challenge in any publications that use the infrastructure or forecasts submitted to the challenge (vera4cast = NSF DEB-2327030 and DEB-1926388, neon4cast = DEB-1926388). Please email the Challenge organizers with your published manuscript or pre-print.

Yes

A copy of your responses will be emailed to the address you provided.

Submit

Clear form

Never submit passwords through Google Forms.

reCAPTCHA

Privacy Terms

This form was created inside of Virginia Tech. Report Abuse

Forms