A netcdf-formatted file containing the original binned data (described in Shore et al [2017]), in their state before they were subjected to EOF analysis.

These have had additional processing applied to the SuperMAG data (publically available at http://supermag.jhuapl.edu/ ) in the form of sampling them to the centroid of the bins, thus they are worth providing here despite the large file size (approximately 12GB).

To conserve file space, we have removed empty bins, thus the temporal and spatial basis for these data are provided for each filled bin element. Please note that the binned data had not had the temporal mean values (described in Shore et al [2017], and available in the Supporting Information) removed when they were stored in this netcdf file.

The file contains 144 (monthly) sets of 8 variables.

These variables are named:

1: filled_bin_data_YYYYMM_r

2: filled_bin_data_YYYYMM_theta

3: filled_bin_data_YYYYMM_phi

Variables 1 to 3 contain the nanoTesla vales of the binned data for each of the three magnetic field components in the Quasi-Dipole frame.

4: filled_bin_contrib_stations_YYYYMM

The three-letter SuperMAG acronym of the station which contributed to each 5-minute mean data point.

5: filled_bin_colats_YYYYMM

6: filled_bin_longs_YYYYMM

Variables 5 and 6 are the co-latitude and longitude coordinates of each filled bin element.

7: filled_bin_times_YYYYMM

The 5-minute-mean epoch of each filled bin element, with columns in the order: year, month, day, hour, minute, second).

8: filled_bin_indices_YYYYMM

A set of fiducial values describing how the sparse elements of the 1D vector of filled bin values relate to the fiducials of the (transposed!) EOF prediction a 2D matrix product of the spatial and temporal eigenvectors with values in every bin.

An example of the usage of these data is given in the MATLAB program Shore-ms01.m, provided in the Supporting Information of Shore et al [2017].

***** PLEASE BE ADVISED TO USE VERSION 2.0 DATA *****

The VERSION 2.0 data set has been corrected for a bug which led to the bins which span the local midnight meridian having fewer samples than they should. The data density in these bins is now in-line with the rest of the polar coverage. Apart from that change, the original and updated data sets are the same.