224 lines
7.7 KiB
Plaintext
224 lines
7.7 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"<div style='background-image: url(\"../images/header.svg\") ; padding: 0px ; background-size: cover ; border-radius: 5px ; height: 250px'>\n",
|
|
" <div style=\"float: right ; margin: 50px ; padding: 20px ; background: rgba(255 , 255 , 255 , 0.7) ; width: 50% ; height: 150px\">\n",
|
|
" <div style=\"position: relative ; top: 50% ; transform: translatey(-50%)\">\n",
|
|
" <div style=\"font-size: xx-large ; font-weight: 900 ; color: rgba(0 , 0 , 0 , 0.8) ; line-height: 100%\">Noise</div>\n",
|
|
" <div style=\"font-size: large ; padding-top: 20px ; color: rgba(0 , 0 , 0 , 0.5)\">Lab: Probabilistic Power Spectral Densities</div>\n",
|
|
" </div>\n",
|
|
" </div>\n",
|
|
"</div>\n",
|
|
"\n",
|
|
"\n",
|
|
"Seismo-Live: http://seismo-live.org\n",
|
|
"\n",
|
|
"##### Authors:\n",
|
|
"* Tobias Megies ([@megies](https://github.com/megies))\n",
|
|
"\n",
|
|
"---"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"%matplotlib inline\n",
|
|
"import matplotlib.pyplot as plt\n",
|
|
"plt.style.use(\"bmh\")\n",
|
|
"plt.rcParams['figure.figsize'] = 10, 6"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
" * read waveform data from file `data/GR.FUR..BHN.D.2015.361` (station `FUR`, [LMU geophysical observatory in Fürstenfeldbruck](https://www.geophysik.uni-muenchen.de/observatory/seismology))\n",
|
|
" * read corresponding station metadata from file `data/station_FUR.stationxml`\n",
|
|
" * print info on both waveforms and station metadata"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"from obspy import read, read_inventory\n",
|
|
"\n",
|
|
"st = read(\"data/GR.FUR..BHN.D.2015.361\")\n",
|
|
"inv = read_inventory(\"data/station_FUR.stationxml\")\n",
|
|
"\n",
|
|
"print(st)\n",
|
|
"print(inv)\n",
|
|
"inv.plot(projection=\"ortho\");"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
" * compute probabilistic power spectral densities using `PPSD` class from obspy.signal, see http://docs.obspy.org/tutorial/code_snippets/probabilistic_power_spectral_density.html (but use the inventory you read from StationXML as metadata)\n",
|
|
" * plot the processed `PPSD` (`plot()` method attached to `PPSD` object)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"from obspy.signal import PPSD\n",
|
|
"\n",
|
|
"tr = st[0]\n",
|
|
"ppsd = PPSD(stats=tr.stats, metadata=inv)\n",
|
|
"\n",
|
|
"ppsd.add(tr)\n",
|
|
"ppsd.plot()"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"Since longer term stacks would need too much waveform data and take way too long to compute, we prepared one year continuous data preprocessed for a single channel of station `FUR` to play with..\n",
|
|
"\n",
|
|
" * load long term pre-computed PPSD from file `PPSD_FUR_HHN.npz` using `PPSD`'s `load_npz()` staticmethod (i.e. it is called directly from the class, not an instance object of the class)\n",
|
|
" * plot the PPSD (default is full time-range, depending on how much data and spread is in the data, adjust `max_percentage` option of `plot()` option) (might take a couple of minutes..!)\n",
|
|
" * do a cumulative plot (which is good to judge non-exceedance percentage dB thresholds)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"from obspy.signal import PPSD\n",
|
|
"\n",
|
|
"ppsd = PPSD.load_npz(\"data/PPSD_FUR_HHN.npz\", allow_pickle=True)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"ppsd.plot(max_percentage=10)\n",
|
|
"ppsd.plot(cumulative=True)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
" * do different stacks of the data using the [`calculate_histogram()` (see docs!)](http://docs.obspy.org/packages/autogen/obspy.signal.spectral_estimation.PPSD.calculate_histogram.html) method of `PPSD` and visualize them\n",
|
|
" * compare differences in different frequency bands qualitatively (anthropogenic vs. \"natural\" noise)..\n",
|
|
" * nighttime stack, daytime stack\n",
|
|
" * advanced exercise: Use the `callback` option and use some crazy custom callback function in `calculate_histogram()`, e.g. stack together all data from birthdays in your family.. or all German holidays + Sundays in the time span.. or from dates of some bands' concerts on a tour.. etc."
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"ppsd.calculate_histogram(time_of_weekday=[(-1, 0, 2), (-1, 22, 24)])\n",
|
|
"ppsd.plot(max_percentage=10)\n",
|
|
"ppsd.calculate_histogram(time_of_weekday=[(-1, 8, 16)])\n",
|
|
"ppsd.plot(max_percentage=10)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
" * do different stacks of the data using the [`calculate_histogram()` (see docs!)](http://docs.obspy.org/packages/autogen/obspy.signal.spectral_estimation.PPSD.calculate_histogram.html) method of `PPSD` and visualize them\n",
|
|
" * compare differences in different frequency bands qualitatively (anthropogenic vs. \"natural\" noise)..\n",
|
|
" * weekdays stack, weekend stack"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"ppsd.calculate_histogram(time_of_weekday=[(1, 0, 24), (2, 0, 24), (3, 0, 24), (4, 0, 24), (5, 0, 24)])\n",
|
|
"ppsd.plot(max_percentage=10)\n",
|
|
"ppsd.calculate_histogram(time_of_weekday=[(6, 0, 24), (7, 0, 24)])\n",
|
|
"ppsd.plot(max_percentage=10)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
" * do different stacks of the data using the [`calculate_histogram()` (see docs!)](http://docs.obspy.org/packages/autogen/obspy.signal.spectral_estimation.PPSD.calculate_histogram.html) method of `PPSD` and visualize them\n",
|
|
" * compare differences in different frequency bands qualitatively (anthropogenic vs. \"natural\" noise)..\n",
|
|
" * seasonal stacks (e.g. northern hemisphere autumn vs. spring/summer, ...)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"ppsd.calculate_histogram(month=[10, 11, 12, 1])\n",
|
|
"ppsd.plot(max_percentage=10)\n",
|
|
"ppsd.calculate_histogram(month=[4, 5, 6, 7])\n",
|
|
"ppsd.plot(max_percentage=10)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
" * do different stacks of the data using the [`calculate_histogram()` (see docs!)](http://docs.obspy.org/packages/autogen/obspy.signal.spectral_estimation.PPSD.calculate_histogram.html) method of `PPSD` and visualize them\n",
|
|
" * compare differences in different frequency bands qualitatively (anthropogenic vs. \"natural\" noise)..\n",
|
|
" * stacks by specific month\n",
|
|
" * maybe even combine several of above restrictions.. (e.g. only nighttime on weekends)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": []
|
|
}
|
|
],
|
|
"metadata": {
|
|
"jupytext": {
|
|
"encoding": "# -*- coding: utf-8 -*-"
|
|
},
|
|
"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": "3.12.3"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 2
|
|
}
|