Everything you never wanted to know about IBM and IEEE floating point numbers

Kevin W. Hall

The CREWES Matlab toolbox SEG-Y I/O functions have long been able to read IBM floating-point, but have not been able to differentiate between IBM and IEEE trace data. This was left to the user. In addition, it was not possible to write IBM floating point SEG-Y files using the CREWES tools. New functions have been written, tested, and are now available in the toolbox for writing IBM floats. We include new recommendations for handling both large and small numbers in IEEE format