The dataset consists of a time series (251 steps) of 3D scalar fields on a spherical 180x201x360 grid covering 500 Myr of geological time. Each time step is 2 Myrs, and the fields are:

• temperature [degrees K],
• three Cartesian velocity components [m/s],
• thermal conductivity anomaly [Watt/m/K],
• thermal expansivity anomaly [1/K],
• temperature anomaly [degrees K], and
• spin transition-induced density anomaly [kg/m^3].

The simulation was performed in double precision, however, to reduce downloading time, we provide the data in single precision. Each file was saved in a NetCDF Climate and Forecast (CF) convention format, with each 3D scalar field being a function of latitude [degrees north], longitude [degrees east], and radius [km]. The model’s inner and outer radii are 3485 km and 6371 km, respectively.

The spin transition-induced density anomaly is formally defined at the top of the second column of page 11 in Shahnas et al. (2011) – paper [1] below. Without going into complex physics, you can think of it as the local anomaly (compared to the laterally-averaged value) of the density with the spin transition taken into account.

All four anomalies above (thermal conductivity, thermal expansivity, temperature, and spin transition-induced density) were computed relative to the laterally-averaged values of the respective variables and consequently will be positive in some regions and negative in others.

References

1. M. H. Shahnas, W. R. Peltier, Z. Wu, R. Wentzcovitch (2011): The high pressure electronic spin transition in iron: potential impacts upon mantle mixing. J. Geophys. Res. 116, B08205
2. M. H. Shahnas, R. N. Pysklywec, and D. A. Yuen (2016): Spawning superplumes from the midmantle: The impact of spin transitions in the mantle. Geochemistry, Geophysics, Geosystems 17, 4051-4063
3. M. H. Shahnas, D. A. Yuen, R.N. Pysklywec (2017): Mid-mantle heterogeneities and iron spin transition in the lower mantle: Implications for mid-mantle slab stagnation. Earth and Planetary Science Letters 458, 293–304
4. Researcher’s page at the University of Toronto

Accessing the Dataset

The 251 steps are grouped into 25 gzipped tar files mantle{01..25}.tgz – see the table below. To start working on the project, you can download only mantle01.tgz, but for a production-quality animation you will need all 25 files.

File	Timestep range	Size	MD5 checksum
mantle01.tgz	001..010	3.2GB	6aa435a58ac9487f48291c363eccde6e
mantle02.tgz	011..020	3.2GB	dbc46f9cffb665a5bccfc95a4a864079
mantle03.tgz	021..030	3.2GB	cffe250d3388bbd317a44eda88ccf0df
mantle04.tgz	031..040	3.2GB	84979f5daf4470eb81b9cbdb0c981fa6
mantle05.tgz	041..050	3.2GB	4dce2df4da2a7ef4b5b174b5a49af4d0
mantle06.tgz	051..060	3.2GB	97bd7e2ae99b49ac1732ced3b20ba77d
mantle07.tgz	061..070	3.2GB	5a4fab7180456d99fbb19175a05ea3b1
mantle08.tgz	071..080	3.2GB	4caaea5992f66e391535e079575e29f6
mantle09.tgz	081..090	3.2GB	e282bbd474529a858ddddb627204f88f
mantle10.tgz	091..100	3.2GB	1899ecec0756540957f2d18094c39948
mantle11.tgz	101..110	3.2GB	daaac28e6989f68bd4bff4117a2645d5
mantle12.tgz	111..120	3.2GB	e335c9fed2448ca7bfc24a01a78759e6
mantle13.tgz	121..130	3.2GB	01a85d3402c89d964b9158ca6bc6665d
mantle14.tgz	131..140	3.2GB	c1195cd3ba48b17874507f5f03ed0dd5
mantle15.tgz	141..150	3.2GB	8c25f1996c5eaf5ac0a50114170cd436
mantle16.tgz	151..160	3.2GB	5659e592a8dbbc7d6d3f6ebc10a1e3c0
mantle17.tgz	161..170	3.2GB	e59fccb6cabb87e232f30a6c7ce2cbbb
mantle18.tgz	171..180	3.2GB	38a0007cf2676e64eac91ec9c9f3034c
mantle19.tgz	181..190	3.2GB	8ee1cf466264303d5822aede27477151
mantle20.tgz	191..200	3.2GB	9c6163d659ebac7d9d8f72b6592a3994
mantle21.tgz	201..210	3.2GB	fabba593dd5274d98aa9c48d5fea701a
mantle22.tgz	211..220	3.2GB	8e4ce089409629c7f5122c1c6f19d6db
mantle23.tgz	221..230	3.2GB	a78d3fd385baf8c1ec78eda7b69a2394
mantle24.tgz	231..240	3.2GB	d551765802f1aa9b399763beadf8f44e
mantle25.tgz	241..251	3.5GB	eff0d9d0b62522559a53a4eb11aea579

To download all files in bash command line:

urls=( edS6be3sk8oQ58N infBBW2Rc9TJwf7 76Esj3yDP9EiaGc AZmt47d48prCZZF
       9fZ4A7ENGR6sQrc B8HC3H4oqwcsWB3 t3zLJWWeirR5zmG YmkYgxM7xxrNAwj
       rMma6W9MBtQH9LX MzcZBCaxaojTZJx dfP6NXHmekQQrHR 2GnLRgPi8W2Dt5p
       MqtoESg2d9DsF2P ysGoJK6B3pLYaDB Ae32XwCpt7bHo9D AysWSPnxFS6e5B2
       4NcnJkPYWpkXrmb mBRfrnfEEEaKJ9m J63KxeCppK8ssGc NeqnHBNPWx4PRwd
       JdzZQCKiHaRfL9L DXnWtA5fymHBsxA HzgtF42Pf9AnxGm yy8FASeC8Dm54Sy
       TC8QekmjokmBkWA )
for i in $(seq 0 24); do
    wget https://nextcloud.computecanada.ca/index.php/s/"${urls[$i]}"/download -O mantle"$(printf "%02d\n" $((i+1)))".tgz
done

Loading the Data in ParaView

The dataset can be read directly in ParaView (tested in 5.5 and 5.8), both as single files and as a time series.

The 3D velocity vector can be assembled via the Calculator filter

velocity = (iHat*vx + jHat*vy + kHat*vz) * 1e9

where we recommend to change the scaling to [nm/s] to avoid dealing with very small numbers.

Loading the Data in Python

In Python you can read each time step into an xarray.Dataset containing multiple variables:

import xarray as xr
data = xr.open_dataset('spherical001.nc')
print(data)                         # show all variables inside this dataset
print(data.temperature.values)      # this is a 180x201x360 numpy array
print(data.r)                       # radial discretization

Alternatively, you can use the traditional netCDF4 Python interface:

import netCDF4 as nc
all = nc.Dataset('spherical001.nc', 'r')
print(all)                                   # show all variables inside this dataset
print(all.variables['temperature'][:,:,:])   # this is a 180x201x360 numpy array
print(all.variables['r'][:])                 # radial discretization

Acknowledgments

Data courtesy of the Pysklywec Lab (Russell Pysklywec and Hosein Shahnas) at the University of Toronto. The simulation was conducted using Compute Canada’s Niagara cluster.