The software package “ClassifyStorms” version 1.0.1 performs a classification of geomagnetic storms according to their interplanetary driving mechanisms based exclusively on magnetometer measurements from ground. In this version two such driver classes are considered for storms dating back to 1930. Class 0 contains storms driven by Corotating or Stream Interaction Regions (C/SIRs) and class 1 contains storms driven by Interplanetary Coronal Mass Ejections (ICMEs). The properties and geomagnetic responses of these two solar wind structures are reviewed, e.g., by Kilpua et al. (2017, http://doi.org/10.1007/s11214-017-0411-3). The classification task is executed by a supervised binary logistic regression model in the framework of python's scikit-learn library. The model is validated mathematically and physically by checking the driver occurrence statistics in dependence on the solar cycle phase and storm intensity. A detailed description of the classification model is given in Pick et al. (2019) to which this software is supplementary material.Under “Files” you can download ClassifyStorms-V1.0.1.zip, which contains the jupyter notebook “ClassifyStorms.ipynb” (https://jupyter.org/) and the python modules “Imports.py”, “Modules.py” and “Plots.py”. Check for an up-to-date release of the software on GitLab via https://gitext.gfz-potsdam.de/lpick/ClassifyStorms (under Project, Releases). The “Readme.md” file provides all information needed to run or modify “ClassifyStorms” from the GitLab source.The software depends on the input data set “Input.nc”, an xarray Dataset (http://xarray.pydata.org/en/stable) saved in NetCDF format (https://www.unidata.ucar.edu/software/netcdf), which you can also download under “Files”. It contains1. the HMC index: a three-hour running mean with weights [0.25,0.5,0.25] of the original Hourly Magnetospheric Currents index (HMC index, http://doi.org/10.5880/GFZ.2.3.2018.006).2. the geomagnetic observatory data: vector geomagnetic disturbances from 34 mid-latitude observatories during 1900-2015 in the Cartesian Centered Dipole coordinate system. The original observatory data was downloaded from the WDC for Geomagnetism, Edinburgh (http://www.wdc.bgs.ac.uk/) and processed as described in section 2.1 of Pick et al. (2019).3. the “reference” geomagnetic storms: universal time hours of 868 geomagnetic storm peaks together with their interplanetary drivers (class labels 0 or 1, see above) as described in section 2.2 of Pick et al., 2019. These events are taken from published lists (Jian et al., 2006a, 2006b, 2011; Shen et al., 2017; Turner et al., 2009), which are gathered in the separate ASCII file “ReferenceEvents.txt” (under “Files”) for a quick overview.4. additional quantities for plotting: time series of Kp (since 1932) and Dst (since 1957) geomagnetic indices from the WDC for Geomagnetism, Kyoto (http://wdc.kugi.kyoto-u.ac.jp/wdc/Sec3.html) as well as the yearly mean total sunspot number from WDC-SILSO, Royal Observatory of Belgium, Brussels (http://sidc.be/silso/datafiles).The output of ClassifyStorms is "StormsClassified.csv" (under “Files”). This table lists the Date (Year-Month-Day) and Time (Hour:Minutes:Seconds) of 7546 classified geomagnetic storms together with the predicted interplanetary driver class label (0 or 1) and the corresponding probability (between 0 and 1).Version history:20 Sep 2019: Version 1.0.1: Correction of plotting mistake in Figure m / Figure S4 (see gitlab repository for details)
This dataset comprises the PCEEJ equatorial electrojet model current intensity values (mA/m). The PCEEJ is an empirical model based on the principal component analysis of satellite and ground equatorial electrojet data, described in detail in Soares et al. (2022), to which this data publication is supplement to. The model data is provided as text files (.csv extension) and Matlab-formatted files (.mat extension). For text files, there is one file per year (file name labeled with the corresponding year). For the Matlab format, there is only one Matlab file that contains all years as separate variables (variable name labeled with the corresponding year). Each yearly file/variable corresponds to a matrix: the rows represent local time/longitude bins and the columns represent days of year.
The local time/longitude bins (rows) always sum up to 432 (12 local time intervals and 36 longitude intervals). The day of year (columns) always starts in January 1st and ends in December 31st, leading to a total of 365 or 366. The PCEEJ model values of 13 years from 2003 to 2010 and from 2014 to 2018 are provided. The PCEEJ basis functions (principal components) are provided in the text and Matlab files labeled as ‘PC_Functions’. The ‘PC_Functions’ data is given as a 432x10 matrix, in which 432 stands for the aforementioned local time/longitude bins and 10 represents the 10 principal components used to obtain the PCEEJ model (in ascending order). Two additional auxiliary indices, namely ‘lt_index’ and ‘lon_index’ are also contained as text and Matlab files. These indices represent the corresponding local time and longitude values of each row of the PCEEJ yearly files and ‘PC_Functions’ files.
The HMC (Hourly Magnetospheric Currents) index measures the activity of large-scale magnetospheric currents on Earth's surface from 1900 to 2015. It resolves the absolute intensity of low-frequency variations, especially at periods relevant to the solar cycle, more robustly than existing geomagnetic indices. HMC is based on hourly means of vector magnetic field measurements from 34 mid latitude geomagnetic observatories obtained from WDC Edinburgh (http://www.wdc.bgs.ac.uk/catalog/master.html). This data has been manually revised to correct for spikes, jumps and drifts. A detailed description of the derivation method is given in Pick et al., 2018 to which these data are supplementary material.This directory contains the HMC index (hmc1900phor.hor) and the modified observatory data that it is based on (data.zip). The index and the observatory data files are formatted in compliance with the IAGA-2002 ASCII exchange format (https://www.ngdc.noaa.gov/IAGA/vdat/IAGA2002/iaga2002format.html).Individual file names are composed of:[IAGA code of observatory] + [first active year during 1900-2015] + [p(provisional)] + [hor(hourly)] + [_mod(modified)].horAlso included is information on how the data modifications (list in modifications.pdf) were applied (readme.txt).