This codebook.txt file was generated on 20201014 ------------------- GENERAL INFORMATION ------------------- 1. Title of Dataset: Data for Dynamic upwelling beneath the Salton Trough imaged with teleseismic attenuation tomography 2. Author Information Principal Investigator Contact Information Name: Joseph Byrnes Institution: University of Minnesota Email: byrnes.joseph@gmail.com ORCID:0000-0002-6161-399X Associate or Co-investigator Contact Information Name:Max Bezada Institution:University of Minnesota Email:mbezada@umn.edu ORCID:0000-0002-7337-3276 3. Date of data collection: 2011-01-13 to 2020-10-07 4. Geographic location of data collection (where was data collected?): Salton Trough 5. Information about funding sources that supported the collection of the data: Sponsorship: NSF EarthScope EAR-1827277 -------------------------- SHARING/ACCESS INFORMATION -------------------------- 1. Licenses/restrictions placed on the data: CC0 1.0 Universal 2. Links to publications that cite or use the data: Byrnes, Joseph S, & Bezada, Maximiliano. (2020). Dynamic Upwelling Beneath the Salton Trough Imaged With Teleseismic Attenuation Tomography. Journal of Geophysical Research. Solid Earth, 125(11), N/a. https://doi.org/10.1029/2020JB020347 3. Recommended citation for the data: Byrnes, Joseph S; Bezada, Maximiliano. (2020). Data for Dynamic upwelling beneath the Salton Trough imaged with teleseismic attenuation tomography. Retrieved from the Data Repository for the University of Minnesota, https://doi.org/10.13020/m9jh-s045. --------------------- DATA & FILE OVERVIEW --------------------- Data Sets contained in the folder ./Data/ ST_data3.mat - data set measured with SSIP data, with ray paths ST_syn1.mat - synthetic dataset shown in Figure 5f ST_syn3.mat - synthetic dataset shown in Figure 5d ST_syn6.mat - synthetic dataset shown in Figure 5b ST_syn6.mat - synthetic dataset shown in Figure 5b ST_syn6.mat - synthetic dataset shown in Figure 5b ST_syn6.mat - synthetic dataset shown in Figure 5b ST_syn6.mat - synthetic dataset shown in Figure 5b ST_syn_trap2 - synthetic dataset inverted for Figure 10c ST_syn_trap3 - synthetic dataset inverted for Figure 10d ST_syn_tri150km - synthetic dataset inverted for Figure 10a ST_syn_tri225km - synthetic dataset inverted for Figure 10b Inversion results (see Plotting script to view) ST_data3ADistribution - see Figure 9. ST_data3CDistribution - see Figure 9, but with no discontinuity used ST_syn1ADistribution - see Figure 8c ST_syn1EDistribution - see Figure 8d ST_syn3FDistribution - see Figure 8a ST_syn3GDistribution - see Figure 8b ST_syn6ADistribution - see Figure 7a ST_syn6BDistribution - see Figure 7d ST_syn_trap2Distribution - see Figure 10c ST_syn_trap3Distribution - see Figure 10d ST_syn_tri150kmDistribution - see Figure 10a ST_syn_tri225kmDistribution - see Figure 10b Plotting scripts plot_distribution_parameters.m - make plots of the median solutions for Inversion results. *Check captions to confirm contouring intervals* plot_raypaths.m - makes Figure 4 Inversion scripts run_batch.pbs - example execution of MakeDistributions.m MakeDistributions.m - writes qsub scripts for supercomputer to run inversions, according to variables defined in run_batch.pbs TD_inversion_* - called by MakeDistributions.m Linear Inversions smoothTSinversion3.m - linear tomographic approach shown in Figure 6. Modify data variable to make different inversionms. The current work flow is to run through qsub with the run_batch.pbs script Run batch submits several jobs of 24 workers, each which call the TD_inversion_function. This circumvents matlab's inability to distribute across multiple machines, and shoul be rethought for each problem. Demonstration_script produces the result in Figure 6 of Byrnes and Bezada, and Demonstration_script_localmachine runs in serial without the supercomputer overhead. To load new data into the inverison, the dataStruct must be built. An example is in load_data_SaltonTrough3D The fields are: dataStruct.dtS t* data. The d will be removed when absolute t* is enabled dataStruct.allLats station location in lat/lon dataStruct.allLons dataStruct.allSig dummy field for initilaization. Set to any value with same size of dtS dataStruct.allSta = allSta; dataStruct.dataX mapping of lat/lon to x/y cartesian coodinates. dataStruct.dataY dataStruct.dataE event index dataStruct.xVec vectors that define the imaging domain dataStruct.yVec dataStruct.zVec allSta and coastFields in load_data_SaltonTrough3D are not used then there are two substructures. dataStruct.discontinuity is defined if you want a discontinuity in the model (see Byrnes and Bezada, JGR for an exmaple) dataStruct.ray contains ray tracing information. The load for Salton Trough script contains and example of how to built the ray structure with the previous dataStruct fields, event lat/lon/depths, and matTaup (included). To tweak the inverison once you have the data, see define_TDstructure. Comments in there define parameters. For the Salton Trough dataset, check chain takes approximately ~24 hours, but each chain is independant and can be run in parallel. This is a small dataset (220 observations) and while time will scale with the size of the dataset, performance can be tweaked with Znodespacing (finer spacing means longer times), or the algorithm can be compiled for larger datasets. Finally, if running a synthetic, you *must* add noise to the synthetic data. See salton trough examples for examples of synthetic datasets, and Demonstration_script for how to add noise. Body wave tomography is high non-unique, and noise-free datasets will be perfectly fit by a bizarre set of models that could in theory converage, but would take forever so far as I can tell. Results are gathered from many chains by the MakeDistributions script. [name]Distribution.mat files are plotted by the plot_distribution_parameters.m script. plot_model_histogram plots a PDF of zeta values at a point in the imaging domain plot_raypaths shows the ray paths in dataStruct plot_voronoi can be used to look at individual models, either after a run or if you are in debug mode while a particular model is running. Please contact Joseph Byrnes at byrnes.joseph@gmail.com or jsbyrnes@umn.edu for help building dataStruct, or anything else. 2. Relationship between files: The unlisted files are never called by any user or contain any data, but are called by scripts the user calls. -------------------------- DATA-SPECIFIC INFORMATION -------------------------- This package contains all the scripts for reproducing the figures from Byrnes and Bezada, JGR - https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2020JB020347?af=R The current work flow is to run through qsub with the run_batch.pbs script Run batch submits several jobs of 24 workers, each which call the TD_inversion_function. This circumvents matlab's inability to distribute across multiple machines, and should be rethought for each problem. Models can be produced with the pair of Demonstration scripts. Also included a batch file that called MakeDistributions to make the main result Demonstration_script produces the result in Figure 6 of Byrnes and Bezada. Demonstration_script *can't* be run locally; it is run by the .pbs scripts on a supercomputer. Demonstration_script_localmachine runs in serial without the supercomputer overhead. The main result in Byrnes and Bezada, 2020 is produced by the default run_batch.pbs script. Output in all these cases is [name]Distribution.mat file, which contains the set of models produced by the Markov Chain. This file is called by plot_distribution_parameters.m to make Figures such as Figure 7, 8, and 9 in Byrnes and Bezada, 2020. To load new data into the inverison, the dataStruct must be built. An example is in load_data_SaltonTrough3D The fields are: dataStruct.dtS t* data. The d will be removed when absolute t* is enabled dataStruct.allLats station location in lat/lon dataStruct.allLons dataStruct.allSig dummy field for initilaization. Set to any value with same size of dtS dataStruct.allSta = allSta; dataStruct.dataX mapping of lat/lon to x/y cartesian coodinates. dataStruct.dataY dataStruct.dataE event index dataStruct.xVec vectors that define the imaging domain dataStruct.yVec dataStruct.zVec