INTRODUCTION Stratigraphic and exploitation data processing is an increasingly prominent part of seismic data processing. Seismic data is the primary basis for the interpretation procedure. Other subsurface measures such as well logs, well velocity surveys, and vertical seismic profiles are invaluable aids in a definitive interpretation. These measures are more effective used in conjunction than on an individual basis. The potential of the individual methods in resolving earth layering is unique with the technique. Some are excellent for details vertically, but are sparsely spaced horizontally, while others are resolute laterally, but limited vertically. Combining the results of the different measures could provide an improved basis for interpretation. The use of well logs and seismic data jointly is a familiar and practiced art. The conjunction of the two is approached by working to make each into similar time functions. Calculations to transform the sonic and density logs to be comparable with the seismic trace is called the forward problem. Manipulation of the raw seismic data to match the well log synthetic is called the inverse problem. Seismic data processing must strive to remove geometrical distortions, suppress multiple energy, and present the highest frequency data allowed by noise and band limiting imposed by method and environment. Moreover, amplitudes are distorted obscuring the reflection coefficients recorded and must be corrected as much as possible. The forward process is deterministic and well formulated. Basis for generation of the synthetic trace is the sonic well log. Interval transit times are measured by a down-hole tool usually over a 3-ft interval. From the transit time, a velocity for the interval measured can be computed. This function of time and interval velocity is termed the interval velocity log. As depicted in Fig. 1, reflection coefficients are computed by (Equation available in full paper) Most often the density is not used and is considered a constant. When desired, the algorithms of Goupillard (1961), Kunetz (1962), and others can be used to add multiples such as might be present on the seismic trace. An arbitrary wavelet is convolved with the series to simulate the band limiting observed on seismic data. This derived trace is called a well log synthetic. For many reasons, the synthetic may not be too similar to the observed seismic trace. The inverse problem is to start with the seismic trace and compute the reflection coefficients. A variety of techniques has been published and used with varying degrees of success in attempting to do this inverse problem. These center around variations of the Wiener filter and popularly are called deconvolution. Lavergne and Willrn, Lindseth, and Stone proposed a different format for the inverse result. The seismic data is processed to the form of the interval velocity log rather than sharpened pulses for comparison. All the methods deconvolve the data as closely to a spike train as possible. Then, using the equation (Available in full paper) where Ri= estimated reflection coefficients, and Vi= interval velocity at i, an estimated interval velocity log is computed.