You are on page 1of 5

GEOL 4714 Field Geophysics

Fall 2013
LAB 4: SEISMIC REFRACTION
Due: 22 October 2013

The goal of this exercise is to determine shallow subsurface structure (number of rock layers,
their thickness and seismic P-wave velocity, and perhaps dip) using the seismic refraction data
collected by the class. The instructions here are basic guidelines.

We will first pick arrival times from the end shots on line 1 (with 3 m spacing) (so files 50 and
52). We will fit this data with a two layer model. We will then pick the 1 m spacing E-W line 2
(files 54 and 55) and make a 2 layer and a 3 layer interpretation of that profile. We will repeat
that analysis with line 3 (the north-south line) (files 57 and 64). Finally we make a geologic
interpretation, including information you might have from your resistivity analysis.

1. Pick arrivals for each shot much as was demonstrated in class. Measure uncertainties on
your picks. Some of the frequently asked questions from previous classes are below to help
you out.
2. Using these picks, solve for the structure between each pair of shots. You can reorder the
geophones from largest to smallest number and back again by clicking on the title in the data
window. Include the uncertainties on your picks. Uncertainties can be shown in Refract by
going to the Preferences Window and choosing to “Use uncertainties on data”. Note that the
RMS misfit for a given model is shown at the top of the Data Window.
3. For line 1 at 3 m spacing, try a 2-layer model. For lines 2 and 3, try both a 2-layer and 3-layer
model. How does adding a layer change the depth to the halfspace?
4. After getting the best solution for each profile, identify a set of layer velocities that you think
might best fit all three profiles. Then use those velocities in all the profiles and solve again,
just changing the thicknesses and dips of layers (this is usually best done in the Model
Window). Discuss whether or not your measurements require any lateral variation in
velocity or lateral variation in depth, providing a justification from your analysis.

For one of your interpretations (i.e., using picks from a forward and reverse shot), you need to
consider how the uncertainty in your picks translates into an uncertainty in the velocity model
you are constructing. To do this, you will determine the range of possible solutions for either the
second or third layer (this is done by choosing steepest possible fits and solving, shallowest
possible fits and solving, steepest forward with shallowest reverse, and shallowest forward with
steepest reverse). Thus there are four possible combinations. Please provide the following for
this particular interpretation: a travel time plot with the picked travel times and their uncertainties
showing the steepest slopes you used, one showing the shallowest slopes used (you can combine
these two, if desired), and a table showing the velocity model parameters for each of the four
extreme fits and your best fit estimate.

Using information in your text, your resistivity profile, and what you know of the geology of the
area, attempt to interpret your profile (i.e., which layers represent which rocks). Make one
interpretive cross section of the subsurface under the entire combined profile (to scale).
Your geologic interpretation should address the possible range in average depth to the Pierre
Shale as well as likely variations in the depth to the Pierre Shale. Consider any relevant
constraints from the resistivity model you obtained in lab 3. If appropriate, consider the
relationships between porosity and seismic velocity described in the text.

When presenting your seismic data, you should list the picks and uncertainties that you made
(you can copy the data from Refract’s data window and paste it into Excel or Word—the fields
are separated by tabs and so can be made to look nice). Please try and align column headers with
data below. For your models, you should list the model characteristics and generally show a
section, indicating the orientation of the section; it is acceptable to combine several related
models in a single table. Material showing the uncertainties on a model parameter can often be
combined into a single table, greatly reducing the bulk of the report. You should include for each
interpretation a plot of the picked travel times vs. distance and the predicted values from your
model.

As always, you should examine the field notes and carefully locate the experiment on one of the
topo maps available online. Of course, there should be a sketch map (to scale) in your report as
well.

Inverting in Refract
First, recognize that you can see the apparent velocity and intercept of any line in the Plot
Window by hovering the mouse over that line. This information can guide you away from
problems.

It is often tempting to draw lines when inverting for a seismic structure and press Invert. You
will often get frustrated as the lines seem to magically move back somewhere bad or you get a
message that the lines chosen will produce something impossible. First, it helps to know what
Refract does. It first figures out the velocity of the top layer by averaging the apparent velocities
of the lines for the first two layer. It then uses the difference in apparent velocity of the second
layer to get the dip on the base of the first layer. From the intercept times (for the forward picks
only in pre 1.1.1 versions of the software, from the average of the two line fits after that), the
code determines the thickness of the layer. These last two steps are repeated, albeit with
corrections for the structure above that depth, to the bottom of the model.

So if you find that you like your line slopes but you don’t like the intercepts, the problem is that
they are mutually incompatible. Try adjusting the slopes in the proper sense to get the
thicknesses under either end to work better. Or, you can try modifying the forward model
towards what you think will work better and see how the lines change.

If you get warned that your line fits produce crossing interfaces or a negative thickness
somewhere, think about how the choice of line slopes and intercepts affects the inferred model.
If, for instance, you have two lines with very different slopes but nearly identical intercepts, there
is clearly some disconnect: perhaps there is a lateral velocity variation, perhaps you are
associating picks with the wrong layers, etc. You can first try to modify your lines to something
closer to a working model, or you can try to work some in adjusting the model from the Model
Window (you might shift to the “Model layers with thicknesses at shots” mode in the
Preferences Window to help make it clear what you have to work with).
Frequently Asked Questions
"How do I pick the arrival times?"
The first break is usually fairly distinct for stations near the shotpoint. Farther out, one must
recognize that the noise before the first arrival provides a level above which a signal must
rise; in some cases this means that an inflection, rather than a first positive swing, is what is
picked. You might be using the first downswing to guide how you pick later arrivals.
"I can't find the crossover point on a profile."
Consider the following: For our simple analysis, the velocity of the first layer is identical on
both forward and reversed shots. Thus if you have a well-determined velocity in one
direction, you can use that as a guide in the other. Of course, some difference is to be
expected (in general, variations less than 5-10% are within the noise of the technique). Also,
recall the reciprocity of source and receiver applies to the travel path of energy as well as to
the total time seismic energy takes to travel from source to receiver. Thus if you clearly see
on one profile that energy is arriving from a refractor at the other shotpoint (which is just
beyond the end of the profile), then that means that the first arrival at the first shotpoint from
the second is also arriving from the refractor.
"There are lots of short, consistent linear segments--I'd like to have more than 2 [or 3] layers"
That could be your choice, but unless you see a clear shift with distance from a steeper to a
shallower dip, it is unlikely that you want to interpret these variations. While these deviations
are often larger than your pick error, they are within the errors accompanying the
assumptions of our interpretation: the refractor is probably not perfectly planar, the velocities
of the layers are probably not perfectly uniform, etc. Part of the purpose in evaluating the
uncertainties in your interpretation is to bound such variations in the Earth. For instance,
bounding the shallowest and deepest depths consistent with your picks will produce a region
within a cross section where the refractor could lie. The topography on the refractor is very
likely to lie within that region, perhaps bouncing from top to bottom of that region with
distance.
"Can I make a more distant refractor have a lower apparent velocity (steeper slope) than a nearer
refractor (or direct arrival)?"
In this context, no. Basically, the direct arrival would come in before any later refracted
arrival, thus the appearance of such a slope change cannot be from a variation in velocity
with depth. You are seeing some of the variations caused by things outside the simple
assumptions of our model, as discussed above.
"How do I choose which lines to draw for making my estimates of uncertainties?"
You want to bracket the apparent velocity and the intercept time for each shot. Thus you
want to choose two lines with different slopes, one steeper than your best-fit line (thus having
a higher apparent velocity and presumably with a smaller intercept), and one shallower than
your best-fit line (thus having a lower apparent velocity and presumably with a larger
intercept). Then you will have bracketed the intercepts and velocities, which are the values
you are then using in the equations below or which effectively you are altering through the
use of Refract.
.
Time, t
slope 1 /v 2d
= sin(i +i c)/v 1 =
c -_)/v (_
1 = 1/v = sin
up-dip 2u slope down-dip

intercept= t2u
intercept= t2d slope = 1/v1 = 2zu cos(ic)/v1
= 2zd cos(ic)/v1

distance
x

zd
ic
hd hu
zu
ic

ic = sin-1 (v1/v2)
_

Formulas for dipping layer:


ic = 1/2 [sin-1 (v1/v2d) + sin-1(v1/v2u)]
α = 1/2 [sin-1(v1/v2d) - sin-1(v1/v2u)]
v2 = v1/sin(ic)
zx = v1t1x/[2cos(ic)], where x = either u or d
hx = v1t1x/[2cos(ic)cos(α)], where x = either u or d

You might also like