A package to calculate coherent sets from geospatial trajectory data.
pip install GeoCSThe package follows object orientation and is centered around classes handling trajectories (Traj), point-wise distances (Dist), point-cloud boundaries (Bound), and diffusion maps (DiffMap). Each class can be calculated, saved, loaded and plotted.
from GeoCS import Traj, Dist, Bound, DiffMap
from datetime import datetime
T = Traj(path_to_your_trajectories, datetime(Y, M, D, H))
T.load()
r = 1e5 # cut-off radius
k = 15 # scaling parameter
D = Dist(path_to_distances, r=r, k=k, traj_data=T)
D.save()
B = Bound(path_to_boundaries, k=k, convex=True, traj_data=T)
B.save()
eps = 1e5 # diffusion bandwidth
DM = DiffMap(path_to_diffusion_maps, eps=eps, bound_data=B, dist_data=D)
DM.save()
DM.plot()
Full documentation is available on readthedocs: https://geocs.readthedocs.io/.
Project repository is at github: https://github.com/hschoeller/GeoCS
Licensed under the MIT License.
Development has been financed by the DFG funded CRC 1114.
Largely based on theory laid out in Banisch & Koltai, 2017. Application and extension in the context of atmospheric flow will be detailed in future publication (Schoeller et. al, 2024).
Banisch, Ralf and P ́eter Koltai (Mar. 2017). “Understanding the Geometry of Transport: Diffusion Maps for Lagrangian Trajectory Data Unravel Coherent Sets”. In: Chaos 27.3, p. 035804. issn: 1054-1500, 1089-7682. doi: 10.1063/1.4971788.