Computer K‐index estimation by a new linear‐phase, robust, non‐linear smoothing method

Abstract

This paper describes a new method that can be used to estimate the geomagnetic K‐index values automatically with a computer. The underlying philosophy of the method together with a short description of the mathematics involved are discussed. It is shown how existing ideas in the relatively new field of data‐adaptive filtering can be modified and extended to develop a powerful Linear‐phase Robust Non‐linear Smoothing method (LRNS method). The properties of this method are ideally suited for the computer K‐index estimation problem. The method was applied to Hermanus digital geomagnetic data which extend to nearly one decade. Even with this large amount of data a 99 per cent agreement with handscaled values was still obtained when differences of ±1 in the K‐values were neglected. A 70‐80 per cent total agreement was determined with the majority of the differences occurring during quiet days. This is the result of the very small dynamic range of small K‐indices (0 and 1) at Hermanus resulting in the handscaler frequently giving a value of 0 where the computer can detect this small variation and gives a value of 1. What is more important, is that the performance of the method stays virtually the same irrespective of the day or month of the year, or the year or years used in the comparisons. This proves the method's adaptiveness to any changes in the SR pattern irrespective of day‐to‐day, seasonal or solar activity variations. This is achieved without any changes in the original input parameters. Because of its adaptiveness it is believed that the method will also adapt to data from different geographical locations thus giving us a possible global method. Copyright © 1989, Wiley Blackwell. All rights reserved