Modelling the flow of nocturnal bird migration with year-round European weather radar network.

Raphaël Nussbaumer1,2 , Lionel Benoit2 , Grégoire Mariethoz2 , Felix Liechti1 , Silke Bauer1 and Baptiste Schmid1

Biorxiv, Researchgate, Demo, .

Dataset selection and pre-processing

  1. script_download_cleaning.m
    • L3-L10: Download all data available on enram repository for the entire year of 2018.
    • L50-L75: Delete the data of all radar with aparrent error in them
    • L96-L117: Reshape data to be on a regular grid: 5min interval, same altitudinal bins.
    • L119-L157: Compute sunrise and sunset at each radar location
    • L256-L277: Manually clean the data (use a MATLAB app CleaningV.mlapp)
    • L281-L345: More automatic post-cleaning processing: removal of single data point isolated by 30min. Noise removal (replace value exeeding twice the average of 8 surounding values by the mean of the surounding values), interpolate to 5000m.
    • L449-L498: Cleaning of flight speed componenet and sd_vvp based on the clearning performed on density.
    • L502-L771: Insect removal procedure
  2. script_elevation_correction.m
    • L91-L181: MPS simulation for the lower elevation
    • L212-L362: Compute volume available for flight
    • L365-L400: Export data for Zenodo

The final interpolated spatio-temporal map can also be downloaded from zenodo DOI.


See each livescript for more information.

  1. Inference of Bird density Density_inference.mlx
  2. Interpolation of Bird density Density_estimationMap.mlx
  3. Inference of flight speed Flight_inference.mlx
  4. Interpolation of flight speed Flight_estimationMap.mlx

Flow Model


This file contains both the code to compute the fluxes and also produce the figures of the paper.