| 摘要 |
Disclosed is a machine implemented method of utilizing existing log data from one or more boreholes in a field to find out what zero shift, if any, may be necessary in order to calibrate a log from a borehole in the same field. In one embodiment, the disclosed system utilizes the three porosity logs (neutron, density and sonic) and builds a statistical model of the field by counting the number of occurrences of each combination of the three porosity log measurements and by storing the counts in a three-dimensional cell memory where the three dimensions correspond to the three porosity logs and each cell location is identified by a particular combination of the three porosity measurements. To calibrate a log from the same field, data sets comprising the three porosity measurements taken at different depth levels in the borehole from which that log is derived are used sequentially to address cells, the contents of the addressed cells are added to the current contents of a central accumulator, and the cells adjacent the addressed cell along the dimension corresponding to the log which is to be calibrated are added to the current contents of respective side accumulators. For example, ten side accumulators may be used, with five of them storing the cumulative counts respectively of the five adjacent cells preceding the addressed cell, and with the other five storing the cumulative contents respectively of the five cells after the addressed cell along the selected dimension. The contents of the central and the side accumulators are used to form a correction curve whose peak indicates what zero shift, if any, should be applied to the log which is to be recalibrated. If the curve peaks at the central accumulator, then there is no need for a zero shift. If the curve peaks elsewhere, a zero shift corresponding to the distance between the peak and the center accumulator is used for calibration purposes. For example, if the curve peaks at the side accumulator immediately adjacent the central accumulator and corresponding to the first cell preceding the addressed cell, the log is calibrated by subtracting from each of its measurements a number corresponding to the dimension difference between two immediately adjacent cell locations. Disclosed is also a method of utilizing a similar technique for the purpose of reconstructing a missing log. The disclosed methods are applicable to data sets containing more than three measurements. The data sets may comprise any suitable combination of measurements; it is preferable that the measurements comprising a data set have some inherent mutual interdependence.
|
