# Why do deviated wells show poor correlation to seismic data?

Well correlation and calibration is an important step in any type of seismic inversion. In order to obtain a good broadband frequency model, the synthetics should be correlated with original seismic data. The wavelet used for generating synthetic can be either statistical or from the data itself. But I'm facing a problem in obtaining a good correlation for those wells which are deviated. Why do deviated wells always show a lower correlation coefficient in well correlation process?

• "The wavelet used for generating synthetic" can be statistical or from the seismic, but not (only) from the well log. See previous answer. Commented May 18, 2015 at 12:06

Two-way time to depth calibration is a vertical problem.

How you handle deviated wells probably depends a bit on how you are tying the wells. Here are two things to watch out for:

• You should be tying to true vertical depth (TVD) anyway — make sure you're not using measured depth somehow. I expect you are using TVD — so the deviated section will be distorted (squished). It's gong to be difficult to tie, and the modeled waveform won't make much sense.
• Sometimes software tries to help you by placing wells at the midpoint between the top- and bottom-hole locations. You don't want this — make sure the well is at the top-hole location, or at the midpoint between the top-hole and kick-off point.

Because of these two things, you may want to ignore deviated wells in the tying effort, if that's an option. If it isn't, focus on tying the straight hole.

Ultimately, you'll want to build a 3D velocity model, using horizons to interpolate the velocity field, and maybe even using other velocities (like stacking velocities) to 'fill in' where there are few wells. At that point you can position everything properly, deviated or not.

One common issue is that sonic logs in deviated wells are affected by rock anisotropy (e.g. TVI in shales), and thus yield quite different velocities than are measured in vertical wells. This may affect your correlations as well as things mentioned above.

There is a good discussion of the anisotropy issue by Hornby, Howie and Ince, 2003, Anisotropy correction for deviated-well sonic logs: Application to seismic well tie, Geophysics, 68(2):464-471, doi 10.1190/1.1567212.

If you have an estimate of the anisotropy parameters it is possible to correct your sonic logs for the anisotropic effect.

Note that well ties can also suffer as a refer of sonic drift arising from the frequency dependence of wave propagation at sonic frequencies (kHz) and seismic frequencies (Hz).