Ibbitt, R.P. Compression of time-series data
When water-level data are stored on a computer the amount of storage space needed can be reduced by eliminating data that can be provided approximately by linear interpolation. Benefits from reduced computer time are also obtained since fewer values have to be retrieved from the file. The aim of the compression process is to filter out small and abrupt water-level fluctuations without affecting the information content of the record. To avoid over-compressing the data it is desirable to have some method by which an upper limit on the 'compression range' can be determined. A method is presented by which this limit can be obtained under different assumptions about the accuracy of a rating curve.