This documentation is not for the latest stable Salvus version.
Salvus version: 0.12.16
This document describes all the various parameters and settings for meshes of the Cartesian2D type available when using the simple_mesh
interface of SalvusMesh:
from salvus.mesh import simple_mesh m = simple_mesh.Cartesian2D(...) # Accessing the elements_per_wavelength parameter in the basic group m.basic.elements_per_wavelength = 2.0
[basic]
The most important parameters for seismic wave simulations.
Required: True
Detailed description of all parameters:
model
Pretty name: 1D Model
Description: Model Name for predefined 1D models or filename for external 1D model file.
Parameter type: string
Required: True
min_period_in_seconds
Pretty name: Minumum Period in Seconds
Description: The desired minmimum period resolved by the mesh.
Parameter type: number
Required: True
elements_per_wavelength
Pretty name: Elements per Wavelength
Description: Number of elements per wavelength at the mimimum period. Used to calculate element sizes in the 1D model as a function of depth.
Parameter type: number
Required: True
[advanced]
Advanced parameters for more control on seismic wave sisimulation meshes.
Required: True
Detailed description of all parameters:
courant_number
Pretty name: Courant Number
Description: Courant Number used to estimate dt.
Parameter type: number
Required: True
Default: 0.6
tensor_order
Pretty name: Tensor Order
Description: Spatial order used for shape mapping and model representation.
Parameter type: integer
Required: True
Default: 1
model_parameters
Pretty name: Model Parameters
Description: Model parameters to put into the mesh, defaults to the parametrization used by the 1D background model. This can be used for example to create an anisotropic parametrization of an isotropic model.
Parameter type: array of strings
Required: False
Default: []
Possible values: "g"
, "dg"
, "VP"
, "VS"
, "VSV"
, "VSH"
, "VPV"
, "VPH"
, "RHO"
, "QMU"
, "QKAPPA"
, "C11"
, "C12"
, "C13"
, "C14"
, "C15"
, "C16"
, "C21"
, "C22"
, "C23"
, "C24"
, "C25"
, "C26"
, "C31"
, "C32"
, "C33"
, "C34"
, "C35"
, "C36"
, "C41"
, "C42"
, "C43"
, "C44"
, "C45"
, "C46"
, "C51"
, "C52"
, "C53"
, "C54"
, "C55"
, "C56"
, "C61"
, "C62"
, "C63"
, "C64"
, "C65"
, "C66"
, "A"
, "C"
, "L"
, "N"
, "F"
, "XI"
, "PHI"
, "LAMBDA"
, "MU"
, "KAPPA"
, "ETA"
velocity_model_representation
Pretty name: Velocity Model Representation
Description: How to represent the velocity model on the mesh. element_nodes: values are stored on each node of each element, allows discontinuities and variations inside each element, but produces large files. elements: one value per element, allows discontinuities but the model will be piecewise constant. nodes: one value per node, enforces a continuous model, using mean at values on discontinuities. The Salvus solver currently only supports 'element_nodes'.
Parameter type: string
Required: True
Default: "element_nodes"
Possible values: "element_nodes"
, "elements"
, "nodes"
model_spline_order
Pretty name: Model Spline Order
Description: Spline order used for model interpolation in case of layered models. Ignored for polynomial models like PREM.
Parameter type: integer
Required: True
Default: 3
compression
Pretty name: Compress Mesh
Description: If true, the final mesh will be compressed with gzip compression level 2. Compression is most effective for constant values and the compression level does matter much in that case. Slows down the mesh IO.
Parameter type: boolean
Required: True
Default: true
[attenuation]
Parameters for the memory variable approximation of attenuation. A detailed description can be found in van Driel et al. 2014b.
Required: True
Detailed description of all parameters:
number_of_linear_solids
Pretty name: Number of Linear Solids
Description: The number of standard linear solids used. 5 is conservative, in many applications 3 may be sufficient.
Parameter type: integer
Required: True
Default: 5
frequencies
Pretty name: Frequencies
Description: Minimum and maximum frequencies in the attenuation band. If either is not set, it will be determined automatically. Will be overwritten by 'auto_band'.
Parameter type: array of numbers
Required: False
Default: [0.001, 1.0]
power_law_reference_frequency
Pretty name: Power Law Reference Frequency
Description: Reference frequency for power law Q = Q_0 (f / f_0)^alpha, where Q_0 is taken from background model [Hz]. Also see Fichtner & van Driel 2014.*
Parameter type: number
Required: True
Default: 1.0
power_law_alpha
Pretty name: Power Law Alpha
Description: Exponent alpha for power law Q = Q_0 (f / f_0)^alpha. Also see Fichtner & van Driel 2014.*
Parameter type: number
Required: True
Default: 0.0
auto_band
Pretty name: Auto frequency bands
Description: Automatically determine attenuation band [f_min, f_max] using an empirical relation for the bandwidth and the resolved frequency of the mesh (provided in basic). Will overwrite any given frequencies.
Parameter type: boolean
Required: True
Default: true
[cartesian]
Parameters defining the domain for 2D cartesian meshes.
Required: True
Detailed description of all parameters:
x
Pretty name: X Range
Description: X dimension of the domain in km. The second item defaults to the scale of the 1D model.
Parameter type: array of null or number
Required: True
Default: [0.0, None]
y
Pretty name: Y Range
Description: Y dimension of the domain in km. The second item defaults to the second item of the x dimension if not specified.
Parameter type: array of null or number
Required: True
Default: [0.0, None]
[absorbing_boundaries]
Parameters for the addition of absorbing boundary elements.
Required: True
Detailed description of all parameters:
number_of_wavelengths
Pretty name: Number of Wavelengths.
Description: Width of the absorbing boundary in number of wavelengths at the reference velocity and frequency. For best performance, we recommend that this parameter is set to 3.5 or more. If set to zero, only first order absorbing boundaries are used.
Parameter type: number
Required: True
Default: 0.0
reference_velocity
Pretty name: Reference Velocity
Description: Reference velocity which is used in the 'number_of_wavelengths' calculation. If the velocity is varying along the boundary, an average value may suffice. If undesirable reflections are still encountered it is best to trend towards the lowest velocity in the mesh, leading to a thicker absorbing layer.
Parameter type: number
Required: True
Default: 0.0
reference_frequency
Pretty name: Reference Frequency
Description: Reference frequency used in the 'number_of_wavelengths' calculation. This may require some tuning for optimal performance, but a good starting point is setting it to the center frequency of your source. Use lower values to be more conervative.
Parameter type: ['number', 'string']
Required: True
Default: inf
side_sets
Pretty name: Side Sets
Description: List of absorbing side sets to extended for sponge layers. At the moment only supports ['all'] and ['all_except_free_surface'] for cartesian and only ['all_except_free_surface'] for spherical meshes, or ['none'].
Parameter type: array of strings
Required: True
Default: ['none']
[refinement]
Parameters for refinement of the meshes based on the 1D model.
Required: True
Detailed description of all parameters:
refinement_style
Pretty name: Refinement Style
Description: The mesh refinement style. 'doubling' produces the highest quality elements and lowest element number, 'tripling' may be best for models with strong velocity constrasts. 'local' and 'local_stable' allow non monotonous size functions, the refinement factor is 3 for both.
Parameter type: string
Required: True
Default: "doubling"
Possible values: "doubling"
, "tripling"
, "local"
, "local_stable"
hmax_refinement
Pretty name: h_max Refinement
Description: Criterion (vertical oversamping factor BEFORE the refinement) for moving refinement layers downwards to avoid small timesteps caused by placing a refinement into a thin layer. Smaller values = more aggressive, needs to be > 1. For coarse meshes, larger values may be required. Ignored for 'local' refinement styles, if available.
Parameter type: number
Required: True
Default: 1.5
refinement_bottom_up
Pretty name: Refinement Bottom Up
Description: There are two approaches of locating refinements in the mesh: top down approach means minimizing number of elements at the surface at the cost of more elements at the bottom (default). If the option refinement_bottom_up is set, the bottom up approach is used instead, that is minimizing number of elements at the bottom at the cost of more elements at the surface. Which one is more efficient depends on the velocity model and refinement style. Ignored for 'local' refinement styles, if available.
Parameter type: boolean
Required: True
Default: false
horizontal_max_slowness_layer_index
Pretty name: Horizontal maximum slowness layer index (EXPERIMENTAL)
Description: For low velocity layers at the surface, horizontal refinements can be avoided if the source is below the LVL and hence cannot excite waves above a certain slowness. This is a 1D argument and might not me valid in 3D, hence check carefully. Ignored for 'local' refinement styles, if available.
Parameter type: integer
Required: True
Default: 0