Code for:
Population-level migration modeling of North America’s birds through data integration with BirdFlow
Intermediate data and tuned models are available at: https://doi.org/10.5281/zenodo.16985511
Shield:
This work is licensed under a Creative Commons Attribution 4.0 International License.
scripts/01.Run_validation_more_species
Train and validate BirdFlow models using BirdFlowPipelne package.
BirdFlowPipeline package is the workhorse to fit and validate those models.
The rest of the steps in this repository is mainly for summarizing the results and making figures.
scripts/02.Summarize_validation_preliminary
After fitting the 225 grid search and validate each model, summarize the validation results and pull them into a single dataframe.
scripts/03.Summarize_species_specific_tuning_prediction_metrics
Summarize the predictive metrics. Including log likelihood improvement and distance gain, and the performance decay along space and time.
scripts/04.Summarize_species_specific_tuning_bio_metrics
Summarize the biological features of the tuned models. Including the validation of migration speed, straightness, and number of stopovers.
scripts/05.Compare_species_specific_and_LOO
Investigate the performance of parameter transferring using leave-one-out (LOO) method, including All-LOO, Family-LOO, and order LOO.
scripts/06.Summarize_data_composition
Summarize the data composition for tracking, banding, and Motus for each species.
scripts/07.Regression_for_hyperparameters
Regress the selected hyperparameters using morphological and distributional traits.
scripts/08.Demonstration_map
Simulate migration routes using BirdFlow and the case study of phenology. To give a taste of what BirdFlow can do.
scripts/10.Calcualte_important_result_numbers
Calculate some numbers for the Result part of the manuscript.