Workflow for detecting plumes in CitcomS global mantle-flow models. This code was written as part of Hassan et al. (2015) -- please consider citing this work if you find the code useful.
scons: A python build utility that can be installed on linux systems using the system package manager, e.g. apt on Ubuntu.
libf2c: A fortran compatibility library that can also be installed on most linux systems using the system package manager
scons: Can be installed with homebrew from the terminal using brew install scons
libf2c: See instructions here
cd into the folder named fast and run scons from the command line. The scons build system will compile an executable named plumeTrackFast in build\release\program.
The workflow consists of two parts: (i) CitcomS raw cap files for a given number of timesteps are processed through the plumeTrackFast program (ii) the output from the first step are post-processed through a python script to generate a list of plumes detected at each given timestep.
./plumeTrackFast -h produces the following help message:
************************************************************************
* plumeTrackFast (v 0.1) *
************************************************************************
plumeTrackFast features:
USAGE: ./plumeTrackFast [REQUIRED OPTIONS]
OPTIONS:
-h, -help, --help, --usage Display usage instructions.
--cutoff-percentile ARG Cut-off percentile below which field values are
ignored for cluster analysis.
--data-dir ARG Path to available cap-files.
--data-file ARG Model name.
--output-file-basename ARG Base-name for output files
--radius ARG Radius (m)
--start-depth ARG Depth (m) at which the model-space is scanned for
plumes
--start-time ARG Time (Myr) from which the model-space is scanned
for plumes
--stop-time ARG Time (Myr) from which the model-space is no longer
scanned for plumes
--time-file ARG A two-column text file listing model-times (Myr)
and model-time-steps corresponding to available
cap-files.
--tracer-flavour ARG Flavour (1 based) of deep tracers associated with
LLSVPs. Set to -1 to ignore tracers.
--validation-depth ARG Depth (m) at which the model-space is scanned to
validate that plume-conduits found at depth
'--start-depth' are indeed plumes
--velocity-scale ARG Velocity-scale (m/Myr)
EXAMPLES:
./plumeTrackFast --data-file <gpm19> --data-dir <pathToCapFiles> --time-file <modelTimes.txt> --velocity-scale 4.953209857165279 --radius 6371e3 --start-time <0> --stop-time <230> --output-file-basename <gpm19.plumes> --start-depth 350e3 --validation-depth 1500e3 --cutoff-percentile 5 --tracer-flavour 5
plumeTrackFast v(0.1), Copyright (C) 2014 Rakib Hassan (rakib.hassan@sydney.edu.au)
This program is free and without warranty.
Steps to generate cluster analyses outputs are as follows:
- In your work area untar and unzip
data.tar.gz, which contains cap files for a model at times 0, 65 and 70 Myr. Typically, cap files are output every 5 Myr -- to keep the data volume manageable, example data for only three timesteps are provided here. Note that cap files for the 0th timestep are mandatory. - create a folder in your work area named e.g.
output - Run the following command to generate cluster analyses output for the above timesteps.
<path-to-plume-tracker>/fast/build/release/program/plumeTrackFast --data-file gpm58 --data-dir data/gpm58/cap --time-file data/gpm58/cap/times.txt --velocity-scale 4.953209857165279 --radius 6371e3 --start-time 0 --stop-time 230 --output-file-basename output/gpm58.plumes --start-depth 350e3 --validation-depth 1500e3 --cutoff-percentile 5 --tracer-flavour 5
Note that the --velocity-scale parameter is used to convert the non-dimensional model velocities to units of m/Myr and is obtained as described in the CitcomS manual. The above should produce three files, corresponding to the three timesteps, in the output folder. The column-descriptions are as follows:
#column-name description
r : normalized radial distance from centre
theta : latitude (degrees)
phi : longitude (degrees)
time : model time (Myr)
magGradVr : magnitude of the gradient of radial velocity vector
v_r : radial velocity (m/Myr)
T : nondimensional temperature
Eta : viscosity
data : magGradVr after applying percentile cutoff
cid : cluster-id, assigned by clustering algorithm
magGradVrVD : as magGradVr, but at validation-depth
v_rVD : as v_r, but at validation-depth
TVD : as T, but at validation-depth
EtaVD : as Eta, but at validation-depth
dataVD : as data, but at validation-depth
cidVD : as cid, but at validation-depth
tracComp : composition (as a fraction) of tracer id
given by --tracer-flavour, at a model depth
given by --start-depth
./plotPlumeAscentRate.py -h produces the following help message:
Usage:
Plots results from cluster-analysis from plumeTrackFast
Usage 1: plotPlumeAscentRate.py -b <base file-name> -d <reconstruction-directory> -n <neighbour file-name> -r <radial-velocity minimum> -s <start-age> -o <output file-name>
Options:
-h, --help show this help message and exit
-b BASENAME, --base-name=BASENAME
File name for clustered data
-d RECONSPATH, --directory=RECONSPATH
Path to reconstruction files
-s STARTAGE, --start-age=STARTAGE
Start age
-n NBFILENAME, --neighbours=NBFILENAME
File containing node-neighbours from spherical
triangulation
-r RADIALVELOCITYMIN, --radial-velocity-minimum (cm/yr)=RADIALVELOCITYMIN
Minimum radial velocity that a plume-conduit must meet
to be considered as such
-o OUTFILENAME, --output-file=OUTFILENAME
Output-file name
Steps for postprocessing output from (i) are as follows:
- Create another sibling output folder, e.g.
plume350 - Run the following:
<path-to-plume-tracker>/postprocess/plotPlumeAscentRate.py -b ../output/gpm58.plumes.clustered -d ../data/reconstructedShapeFiles/ -n ../data/neighbours.txt -r 10 -s 230 -o plume350/plumes.gpm58.txt
Note that the reconstructedShapeFiles folder contains reconstructed shapefiles for ages corresponding to the available model times in this example. The neighbours.txt contains natural neighbour indices for each node of the CitcomS mesh at the surface and is obtained through a spherical triangulation. Note that this file corresponds to a CitcomS mesh of resolution (129x129x65 x 12); a different resolution mesh would require regenerating this file.
The above should produce the following in the plume350 folder:
plumes.gpm58.txt, a text file (column descriptions below) containing entries, grouped by model time, for plumes detected
#column-name description
time : model time (Myr)
theta : latitude (degrees)
phi : longitude (degrees)
T : nondimensional temperature
Vr : radial velocity (m/Myr)
Flux : buoyancy flux (Mg/s)
Area : conduit area (km2)
Eruption : if a plume wasn't detected within 500 km of
the current location in the preceding 5 Myr,
a detected plume is considered a new eruption
and this flag is set to 1; 0 otherwise
meanBgTemp : a secondary mean nondimensional temperature,
computed based on nodes that are warmer than
the mean temperature within a ~400 km region
around a plume conduit
avgConduitTemp : mean temperature within a plume conduit.
See appendix A in Hassan et al. (2015) for a
detailed account of how plume conduits are
delineated
concentration : concentration of anomalous material, as a
fraction, near the surface
-
Three plots, corresponding to the model timesteps, showing plumes detected in each, e.g. at 65 Myr below:
-
plume350/supp/, a folder containing supplementary plots for each detected plume, shown in a local coordinate system
