This is the Jupyter Notebook, an interactive coding and computation environment. For this lab, you do not have to write any code, you will only be running it.
To use the notebook:
Instructions as to how to set up Python and the iPython notebook on your personal computer are attached in the appendix of the lab
In [1]:
%pylab inline
from geoscilabs.seismic.NMOwidget import ViewWiggle, InteractClean, InteractNosiy, NMOstackthree
from SimPEG.Utils import download
# Define path to required data files
synDataFilePath = 'http://github.com/geoscixyz/geosci-labs/raw/master/assets/seismic/syndata1.npy'
obsDataFilePath = 'https://github.com/geoscixyz/geosci-labs/raw/master/assets/seismic/obsdata1.npy'
timeFilePath= 'https://github.com/geoscixyz/geosci-labs/raw/master/assets/seismic/time1.npy'
# Download the data
synData = download(synDataFilePath,overwrite=True,verbose=False)
obsData = download(obsDataFilePath,overwrite=True,verbose=False)
timeData = download(timeFilePath,overwrite=True,verbose=False)
We have two CMP gathers generated from different geologic models. One data set is clean and the other is contaminated with noise. The seismic data were adapted from SeismicLab (http://seismic-lab.physics.ualberta.ca/).
In this notebook, we will walk through how to construct a normal incidence seismogram from these data sets.
We will do this in the following steps:
As you can see from clean CMP gather, you can recognize that we have only have one reflector, meaning there is a single interface seperating two geologic units visible in these data. (Note: The direct and any refracted arrivals have been removed).
It is difficult to distinguish any reflectors in the noisy data. However, there is a single reflector in these data, and we will perform normal moveout (NMO) and stacking operations to construct a normal-incidence seismogram where this reflector is visible.
In [2]:
# Plot the data
ViewWiggle(synData, obsData)
where $x$ is offset between source and receiver, $v_{stacking}$ is stacking velocity, and $t_0$ is the intercept time:
$$ t_0 = \sqrt{\frac{4d^2}{v^2_{stacking}}}$$where $d$ is the thickness of the first layer.
For each reflection event hyperbola, perform a velocity analysis to find $v_{stacking}$. This is done by first choosing $t_o$. Then choose a trial value of velocity.
Calculate the Normal Moveout Correction: Using the hyperbolia corresponding to $v_{stacking}$, compute the normal moveout for each trace and then adjust the reflection time by the amount $\triangle T$: $$ \triangle T = t_0-t(x) \\ $$
Estimate $t_0$, and a plausible $v_{stack}$ by altering t0 and v using below widget. This hyperbola will be drawn as red hyperbola on the middle panel. On the right panel we apply stack with the velocity that you fit, and provice stacked trace.
Parameters of the below widget to fit observed reflection event are:
In [3]:
# Fit hyperbola to clean data
clean = InteractClean(synData,timeData)
clean
Compared to the previous data set, this one is quite noisy. There is a reflector in the data, and your goal is to construct a stacked trace where this reflection is visible.
Estimate $t_0$, and a plausible $v_{stack}$ by altering t0 and v using below widget. This hyperbola will be drawn as red hyperbola on the middle panel. On the right panel we apply stack with the velocity that you fit, and provice stacked trace.
In [4]:
noisy = InteractNosiy(obsData,timeData)
noisy
In the previous step, you chose an intercept time (t0) and a stacking velocity (v). Running below cell will generate trhee stacked traces:
In [5]:
NMOstackthree(obsData, noisy.kwargs["t0"], noisy.kwargs["v"]-200., noisy.kwargs["v"], noisy.kwargs["v"]+200.,timeData)
In [ ]: