deal with nans in my tutorial

[EvanBianco]

Need to deal with nans encountered in the despiking function, and in turn make sure that zeros are not encountered in the denominator when trying to compute reflection coefficient series.

[EvanBianco]

I actually see this warning message as well, however it doesn't seem to throw an error. I think what is happening here is that the despike function isn't written to deal with nan values or missing values. I should update the notebook on GitHub, especially since it has been about a year and a half since I wrote this, and it's surely not as comprehensive or elegant as it could be / should be. In the meantime, let me suggest three changes to you, that you should be able to get you back on track.

Change 1: add the following two lines of code to the despike function.

curve = np.nan_to_num(curve)
curve_sm = np.nan_to_num(curve_sm)

Change 2: we need to undo this change when we convert slowness (dt) into pvelocity. We can't divide by zero, but we can (oddly enough) divide by a nan

so add the following line of code before computing the pvelocity

dt[dt==0] = np.nan

​Change 3: modify Z before computing reflection coefficients. Do this,​

rho[rho==0] = np.nan
Z = pvelocity * rho

instead of this...
Inline image 3
​So we don't encounter zeros in the denominator (which will cause an error) when doing this ,​
Inline image 4