Statistical methods applied to log normalization

Abstract

The response of well-logging toais can be affected by numerous problems deriving from their miscalibration. These errors need to be eliminated so that the petrophysical properties of subsurface rocks can be determined without distortion. White eliminating these errors, the analyst must also be aware of lateral facies variations that may fikewise ínf!uence log readings_ Statistical methods based on frequency distribution and trend surface anafysis using a fithologic unit with well-defined log readings serve as hefpful toais in loca­ting, identifying, and correcting these errors. Trend surface analysis is used to determine a polynomiaf function that reffects the regional structure of the field and íts local facies variations. The frequency distribution of the residuais, obtained by subtracting the raw data from the vafues computed using the polynomial function, ser­ves in quantifying the correction needed to normalíze different log curves.