Skip to content


Repository files navigation

Meteo-Particle model for wind and temperature field construction using Mode-S data

This repository is a Python library for wind field estimation based on the Meteo-Particle particle model. The wind and temperature are computed from ADS-B and Mode-S data using the pyModeS library.

This tool was developed together with a research paper by Junzi Sun and his colleagues at TU Delft CNA/ATM research group. If you find it useful for your research, please consider citing it as:

  title={Weather field reconstruction using aircraft surveillance data and a novel meteo-particle model},
  author={Sun, Junzi and V{\^u}, Huy and Ellerbroek, Joost and Hoekstra, Jacco M},
  journal={PloS one},

Dependent libraries

  1. You need to install pyModeS library for ADS-B and Mode-S decoding.
  2. You also need to install the following common scientific libraries: numpy, pandas, and matplotlib.
  3. You may install optional geomag library, to support the correction of magnetic declination in BDS60 heading.

For a fresh install, run following commands:

$ pip install git+
$ pip install numpy pandas matplotlib geomag

Code examples

Examples of using the model with recorded data and real-time streaming are given in and file.

To quickly test the model out of the box, try:

$ python

or if you have access to a ModeSBeast raw stream on TCP port:

$ python --server xx.xx.xx.xx --port xxxxx

Configurable model parameters (with defaults) are:

AREA_XY = (-300, 300)           # Area - xy, km
AREA_Z = (0, 12)                # Altitude - km

GRID_BOND_XY = 20               # neighborhood xy, +/- km
GRID_BOND_Z = 0.5               # neighborhood z, +/- km
TEMP_Z_BUFFER = 0.2             # neighborhood z (temp),  +/- km

N_AC_PTCS = 300                 # particles per aircraft
N_MIN_PTC_TO_COMPUTE = 10       # number of particles to compute

CONF_BOUND = (0.0, 1.0)         # confident normalization

AGING_SIGMA = 180.0             # Particle aging parameter, seconds
PTC_DIST_STRENGTH_SIGMA = 30.0  # Weighting parameter - distance, km
PTC_WALK_XY_SIGMA = 5.0         # Particle random walk - xy, km
PTC_WALK_Z_SIGMA = 0.1          # Particle random walk - z, km
PTC_VW_VARY_SIGMA = 0.0002      # Particle initialization wind variation, km/s
PTC_TEMP_VARY_SIGMA = 0.1       # Particle initialization temp variation, K

ACCEPT_PROB_FACTOR = 3          # Measurement acceptance probability factor
PTC_WALK_K = 10                 # Particle random walk factor


One minute simulation: simulation

Wind field from the sample dataset (snapshot at 01/01/2018 09:02 UTC) real-wind-field

Temperature field from the sample dataset (snapshot at 01/01/2018 09:02 UTC) real-wind-field


Weather (wind and temperature) field reconstruction based on meteo-particle model using ADS-B and Mode-S data








No packages published