Skip to content

Commit

Permalink
Added some python notebook tutorials.
Browse files Browse the repository at this point in the history
  • Loading branch information
@jkroenke
jkroenke committed Jan 10, 2019
1 parent 348bb68 commit 09d1bd4
Show file tree
Hide file tree
Showing 3 changed files with 471 additions and 4 deletions.
239 changes: 239 additions & 0 deletions examples/tutorials/climate_network.ipynb
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,239 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# Climate Network Tutorial\n",
"\n",
"Tutorial on analyzing climate networks using Python.\n",
"\n",
"Uses the Python packages ``core`` and ``climate`` providing all kinds of tools\n",
"related to climate networks. Written as part of a diploma / phd thesis in\n",
"Physics by Jonathan F. Donges (donges@pik-potsdam.de) at University of Potsdam\n",
"/ Humboldt University Berlin and Potsdam Institute of Climate Impact Research\n",
"(PIK),\n",
"\n",
"Copyright 2008-2018."
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Download the data set and copy it to the path below or change the path to the location of the data set"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"DATA_FILENAME = \"../../../Daten/Reanalysis/NCEP-NCAR/air.mon.mean.nc\""
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Create ClimateData object containing the data and print information"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from pyunicorn import climate\n",
"\n",
"# Settings\n",
"\n",
"# Type of data file (\"NetCDF\" indicates a NetCDF file with data on a regular\n",
"# lat-lon grid, \"iNetCDF\" allows for arbitrary grids - > see documentation).\n",
"# For example, the \"NetCDF\" FILE_TYPE is compatible with data from the IPCC\n",
"# AR4 model ensemble or the reanalysis data provided by NCEP/NCAR.\n",
"FILE_TYPE = \"NetCDF\"\n",
"\n",
"# Indicate data source (optional)\n",
"DATA_SOURCE = \"ncep_ncar_reanalysis\"\n",
"\n",
"# Name of observable in NetCDF file (\"air\" indicates surface air temperature\n",
"# in NCEP/NCAR reanalysis data)\n",
"OBSERVABLE_NAME = \"air\"\n",
"\n",
"# Select a subset in time and space from the data (e.g., a particular region\n",
"# or a particular time window, or both)\n",
"WINDOW = {\"time_min\": 0., \"time_max\": 0., \"lat_min\": 0, \"lon_min\": 0,\n",
" \"lat_max\": 30, \"lon_max\": 0} # selects the whole data set\n",
"\n",
"# Indicate the length of the annual cycle in the data (e.g., 12 for monthly\n",
"# data). This is used for calculating climatological anomaly values\n",
"# correctly.\n",
"TIME_CYCLE = 12\n",
"\n",
"data = climate.ClimateData.Load(file_name=DATA_FILENAME,\n",
" observable_name=OBSERVABLE_NAME,\n",
" data_source=DATA_SOURCE,\n",
" file_type=FILE_TYPE,\n",
" window=WINDOW,\n",
" time_cycle=TIME_CYCLE)\n",
"\n",
"print(data)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Generate climate network using various procedures"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {
"scrolled": true
},
"outputs": [],
"source": [
"# Indicates whether to use only data from winter months (DJF) for calculating\n",
"# correlations\n",
"WINTER_ONLY = False\n",
"\n",
"# fixed threshold\n",
"THRESHOLD = 0.5\n",
"\n",
"# fixed link density\n",
"LINK_DENSITY = 0.005\n",
"\n",
"# Create a climate network based on Pearson correlation without lag and with\n",
"# fixed threshold\n",
"net = climate.TsonisClimateNetwork(data, threshold=THRESHOLD, winter_only=WINTER_ONLY)\n",
"\n",
"# Create a climate network based on Pearson correlation without lag and with\n",
"# fixed link density\n",
"# net = climate.TsonisClimateNetwork(data, link_density=LINK_DENSITY, winter_only=WINTER_ONLY)\n",
"\n",
"# Create a climate network based on Spearman's rank order correlation without\n",
"# lag and with fixed threshold\n",
"# net = climate.SpearmanClimateNetwork(data, threshold=THRESHOLD, winter_only=WINTER_ONLY)\n",
"\n",
"# Create a climate network based on mutual information without lag and with\n",
"# fixed threshold\n",
"# net = climate.MutualInfoClimateNetwork(data, threshold=THRESHOLD, winter_only=WINTER_ONLY)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Calculate some measures"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"print(\"Link density:\", net.link_density)\n",
"\n",
"# Get degree\n",
"degree = net.degree()\n",
"# Get closeness\n",
"closeness = net.closeness()\n",
"# Get betweenness\n",
"betweenness = net.betweenness()\n",
"# Get local clustering coefficient\n",
"clustering = net.local_clustering()\n",
"# Get average link distance\n",
"ald = net.average_link_distance()\n",
"# Get maximum link distance\n",
"mld = net.max_link_distance()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Save results to text files"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"import numpy as np\n",
"\n",
"# Save the grid (mainly vertex coordinates) to text files\n",
"data.grid.save_txt(filename=\"grid.txt\")\n",
"\n",
"# Save the degree sequence. Other measures may be saved similarly.\n",
"np.savetxt(\"degree.txt\", degree)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"Plot the results into the file \"climate_network_measures\". This only works if you have the pyngl package!"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"map_plots = climate.MapPlots(data.grid, DATA_SOURCE)\n",
"\n",
"# Add network measures to the plotting queue\n",
"map_plots.add_dataset(\"Degree\", degree)\n",
"map_plots.add_dataset(\"Closeness\", closeness)\n",
"map_plots.add_dataset(\"Betweenness (log10)\", np.log10(betweenness + 1))\n",
"map_plots.add_dataset(\"Clustering\", clustering)\n",
"map_plots.add_dataset(\"Average link distance\", ald)\n",
"map_plots.add_dataset(\"Maximum link distance\", mld)\n",
"\n",
"# Change the map projection\n",
"map_plots.resources.mpProjection = \"Robinson\"\n",
"map_plots.resources.mpCenterLonF = 0\n",
"\n",
"# Change the levels of contouring\n",
"map_plots.resources.cnLevelSelectionMode = \"EqualSpacedLevels\"\n",
"map_plots.resources.cnMaxLevelCount = 20\n",
"\n",
"# map_plots.resources.cnRasterSmoothingOn = True\n",
"# map_plots.resources.cnFillMode = \"AreaFill\"\n",
"\n",
"map_plots.generate_map_plots(file_name=\"climate_network_measures\",\n",
" title_on=False, labels_on=True)"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.6.6"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
8 changes: 4 additions & 4 deletions examples/tutorials/climate_network_lagged.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,7 @@

import numpy as np

from pyunicorn import CouplingAnalysis
from pyunicorn.funcnet.coupling_analysis import CouplingAnalysis
from pyunicorn import climate

#
Expand Down Expand Up @@ -94,10 +94,10 @@
#

# Get array of anomaly time series
anomalies = data.get_anomaly()
anomalies = data.anomaly()

# Get data grid
grid = data.grid()
grid = data.grid

# Create CouplingAnalysis object
ca = CouplingAnalysis(anomalies)
Expand All @@ -108,7 +108,7 @@
# similarity_matrix = \
# np.abs(ca.get_cross_correlation(tau_max=LAG, lag_mode='all')[0, :, :])
similarity_matrix = \
ca.get_mutual_information(bins=16, tau_max=LAG, lag_mode='all')[0, :, :]
ca.mutual_information(bins=16, tau_max=LAG, lag_mode='all')[0, :, :]

# This matrix is not symmetric in general. To create an undirected climate
# network, symmetrization is necessary! This point is critical in interpreting
Expand Down

0 comments on commit 09d1bd4

Please sign in to comment.