Subsections

• Survey geometry
• No. of sources
• Source positions
• Receiver z-plane
• Raytracing numerical parameters
• No. wide-angle ray phases
• No. trace specifications
• No. segments
• Layer sequence
• S/P sequence
• Theta range
• A sample .trc file
• No. of rays (theta)
• Phi range
• No. of rays (phi)
• No. normal-incidence ray phases
• No. trace specifications
• No. segments
• Layer sequence
• S/P sequence
• X range
• No. of rays (x)
• Y range
• No. of rays (y)
• Distance tolerance
• Log
• Ray Log

Geometry and ray-tracing (.trc file)

The .trc file specifies the geometry of the seismic experiment used to obtain data for the inversion. It also contains parameters that control how ray-tracing is performed.

The experimental geometry is described in terms of sources and targets. Jive3D traces rays from each source, through the model to all specified targets. The reciprocity of travel times in ray theory allows rays to be traced in either direction between actual seismic sources and the recording instruments in a survey. Since Jive3D works most efficiently with a small number of sources and a large number of targets, the experimental geometry is usually optimally configured as follows:

Marine wide-angle surveys

Use ocean-bottom instruments as sources and shot points as targets;

Land surveys

Use shots as sources and seismometers as targets;

Multi-channel marine surveys

Use shots as sources and hydrophones as targets^5.2;

Vertical seismic profiles

Use shots as sources and geophones as targets.

Any number of sources may be defined within the model. In addition, any number of wide-angle phases may be defined. A phase describes the path through the model taken by seismic energy between a source and its targets. The phase is specified as a sequence of layers through which the ray must pass between a source and the target surface. For example, a phase that reflects from the 4th interface might be described as [1 2 3 4 4 3 2 1], and one that is refracted in the 2nd layer would be [1 2 1]. Both examples would require that the source position and target plane are within layer 1.

Survey geometry

This parameter (a character array) can take the value `Marine' or `Land'. It specifies the general location of target positions as follows:

Marine

Targets lie on a horizontal plane within the model. The `Receiver z-plane' parameter gives the depth of this plane, and is required. The plane is normally located at the acquisition streamer depth for marine surveys.

Land

Targets lie on the first interface in the model, which represents the surface of the Earth. The `Receiver z-plane' parameter is not required.

No. of sources

The number of sources (an integer) from which rays will be traced through the model to the specified targets.

Source positions

A set of $\left(x,y,z\right)$ co-ordinates (real numbers) specifying the position of each source (in km units). Comments may be added after the co-ordinates to indicate, for example, the name of an instrument.

Receiver z-plane

Depth of the receiver plane (real number, in km) on which targets lie for `Marine' inversions. Not required for `Land' inversions.

Raytracing numerical parameters

These control some technical aspects of the ray-tracing. There are four parameters as follows:

1. Maximum primary perturbation allowed per cell during ray-tracing (in km; suggested values $10^{-3}$ or $10^{-4}$);
2. Maximum secondary perturbation allowed per cell during ray-tracing (in km; suggested values $10^{-3}$ or $10^{-4}$);
3. Accuracy to which each interface intersection is located by binary search (in km; suggested value $10^{-4}$);
4. Number of points along a ray to test for intersection with an interface (suggested value 8).

If large values are allocated to the primary and secondary perturbation parameters (1 & 2), the efficiency of ray-tracing will increase, but errors will be introduced into the ray paths obtained which may lead to instability in the inversion. These errors are difficult to quantify, and are best assessed by experimenting with different values and observing their effects. Parameter 3 determines the size of the random error introduced at each intersection of a ray with an interface. This may be adjusted according to the number of interfaces in the model, but a value of $10^{-4}$ is recommended--increasing it will not significantly speed up the ray-tracing. Parameter 4 need only be adjusted when rays are required to intersect interfaces of very high complexity but appear to be missing the correct intersection points.

No. wide-angle ray phases

The number of different ray phases (see section 3.2 for a definition of `phases') for wide-angle data (an integer). These ray phases are defined for all sources, but not all source-phase combinations need contain targets.

No. trace specifications

The number of specifications (an integer) for wide-angle ray fans that are listed for each source and phase. The use of more than one specification allows very rapid ray-tracing during the early stages of the inversion when the model contains only large-scale structure, and a more detailed but time-consuming mapping of the model at later stages when the model contains complicated structural features.

The next three parameters define a single phase. They must be repeated, with the corresponding fan specifications that follow, for each subsequent phase.

No. segments

The number of segments in this phase (an integer)--see section 3.2 for an explanation of ray phases.

Layer sequence

Indices of the layers through which the ray path must pass from each source to all targets in this phase (integers; one per segment).

S/P sequence

This specifies whether energy propagates as P-waves or S-waves in each segment of the ray path (integers; one per segment). A value of 1 corresponds to P-waves and 2 to S-waves.

The next four parameters describe the fan ray-tracing for a single source within the current phase. They must be repeated for each subsequent source, and then the whole set repeated again for each subsequent fan specification.

Directions of ray propagation from the source are specified in terms of the angular pair $\left(\theta, \phi\right)$, in which $\theta$ measures the angle between the direction of propagation and the $z$-axis and $\phi$ measures the azimuth. A value of $\theta=0$ indicates the downwards vertical direction, a value of $\phi=0$ indicates the direction parallel to the $x$-axis in which $x$ increases, and a value of $\phi=90^{\circ}$ indicates the direction parallel to the $y$-axis in which $y$ increases.

The total number of rays in the fan is equal to [No. of rays (theta)] $\times$ [No. of rays (phi)]. The ray-tracing time will increase in proportion to this number.

Theta range

The minimum and maximum values of $\theta$ (real numbers) that specify the bounds of the solid angle fan of rays to be sent out for this source and phase.

A sample .trc file

 

*** Jive3D forward-modelling spec for "caswani11" - see .inv file for details

Survey geometry = Marine

No. of sources = 10

Source positions =
16.8418 24.3224 1.3917 - 11a
18.7451 21.7271 1.4086 - 12a
16.0990 19.7225 1.4090 - 13a
13.4879 17.8448 1.3175 - 14a
11.7109 20.6201 1.3359 - 15a
15.1011 18.9810 1.3120 - 11b
14.3041 18.3801 1.3200 - 12b
13.4789 17.8141 1.3169 - 13b
12.9396 18.6454 1.3232 - 14b
12.3334 19.4640 1.3355 - 15b

Receiver z-plane =
0.006

Raytracing numerical parameters =
1.e-4 1.e-4 1.e-4 8

No. wide-angle ray phases = 2

No. trace specifications = 1

** Wide-angle phase 1

No. segments = 3
Layer sequence =
1 2 1
S/P sequence   =
1 1 1

** Wide-angle phase 1 trace specifications
** Turning rays above the BSR

** - range in theta/phi min,max in degrees

** Wide-angle spec 1

** Source 1 - 11a
Theta range =
30. 100.
No. of rays (theta) = 20
Phi range =
0. 360.
No. of rays (phi) = 30

** Source 2 - 12a
Theta range =
30. 100.
No. of rays (theta) = 20
Phi range =
0. 360.
No. of rays (phi) = 30
[cut to save space]

** End of wide-angle phase 1

** Wide-angle phase 2

No. segments = 4
Layer sequence =
1 2 2 1
S/P sequence   =
1 1 1 1

** Wide-angle phase 2 trace specifications
** BSR reflections

** - range in theta/phi min,max in degrees

** Wide-angle spec 1

** Source 1 - 11a
Theta range =
0. 60.
No. of rays (theta) = 20
Phi range =
0. 360
No. of rays (phi) = 30
[cut to save space]

** End of wide-angle phase 2

No. normal-incidence ray phases = 1

No. trace specifications = 1

** Normal-incidence phase 1

No. segments = 2
Layer sequence =
2 1
S/P sequence   =
1 1

** Normal-incidence phase 1 trace specifications
** BSR reflections

** Normal-incidence spec 1

X range =
9.1 18.9
No. of rays (x) = 100
Y range =
15.1 25.9
No. of rays (y) = 100

** End of normal-incidence phase 1

Distance tolerance = 0.002

Log = 0

Ray Log = 0

No. of rays (theta)

The number of evenly-spaced values of $\theta$ at which rays will be sent out in the fan (an integer).

Phi range

The minimum and maximum values of $\phi$ (real numbers) that specify the bounds of the solid angle fan of rays to be sent out for this source and phase.

No. of rays (phi)

The number of evenly-spaced values of $\phi$ at which rays will be sent out in the fan (an integer).

No. normal-incidence ray phases

The number of different ray phases (see section 3.2 for a definition of `phases') for normal-incidence data (an integer). For each phase, rays are traced from the reflecting interface up to the receiver plane or model surface.

No. trace specifications

The number of specifications (an integer) for normal-incidence ray grids that are listed for each phase. The use of more than one specification allows very rapid ray-tracing during the early stages of the inversion when the model contains only large-scale structure, and a more detailed but time-consuming mapping of the model at later stages when the model contains complicated structural features.

The next three parameters define a single phase. They must be repeated, with the corresponding fan specifications that follow, for each subsequent phase.

No. segments

The number of segments in this phase (an integer)--see section 3.2 for an explanation of ray phases. Note that normal-incidence rays are traced from the reflecting interface to the model surface, i.e. only half the propagation path is actually traced.

Layer sequence

Indices of the layers through which the ray path must pass from the reflecting interface to reach all targets in this phase (integers; one per segment).

S/P sequence

This specifies whether energy propagates as P-waves or S-waves in each segment of the ray path (integers; one per segment). A value of 1 corresponds to P-waves and 2 to S-waves.

The next four parameters describe a single grid specification within the current phase. They must be repeated for each subsequent grid specification.

A grid of rays propagates from the reflecting interface towards the model surface or receiver plane. All rays in the grid propagate away from the interface at normal-incidence. Their initial $\left(x,y\right)$ co-ordinates form a grid of evenly-spaced points, as defined by the following parameters. The total number of rays in the grid is equal to [No. of rays ($x$)] $\times$ [No. of rays ($y$)]. The ray-tracing time will increase in proportion to this number.

X range

The minimum and maximum values of $x$ (real numbers, in km) that specify the boundaries of the grid of rays to be sent out from this interface.

No. of rays (x)

The number of evenly-spaced values of $x$ at which rays will be sent out in the grid (an integer).

Y range

The minimum and maximum values of $y$ (real numbers, in km) that specify the boundaries of the grid of rays to be sent out from this interface.

No. of rays (y)

The number of evenly-spaced values of $y$ at which rays will be sent out in the grid (an integer).

Distance tolerance

The accuracy to which each target position is located during two-point ray-tracing (a real number in km). This should be kept below the accuracy with which target positions are known in order to avoid introducing additional uncertainties into the travel-time data. Increasing this value will not significantly speed up the ray-tracing.

Log

This defines the level of commentary written to the .out.log file during ray-tracing (an integer), defined as follows (see pages 48-56 of the thesis[1] for explanations of the terms used here):

Log	Level of commentary
0	No commentary
1	List each ray & the segments of each ray
2	As 1 plus ray tracking between segments
3	As 2 plus ray tracking between cells
4	As 3 plus info on ray bisections
5	As 4 plus all values of $\tau$ and $\Delta \tau$

The log flag is used for troubleshooting and should normally be set to zero to maximise the efficiency of ray-tracing.

Ray Log

This determines whether ray paths are written to a new file for each iteration of the inversion ( Ray Log$= 1$) or written over the results from the previous iteration ( Ray Log$= 0$). It should normally be set to zero (in order to avoid excessive disc usage) unless ray diagrams for successive iterations are specifically required.