A critical review and presentation of the complete, historic series of K-indices as determined at Norwegian Magnetic Observatories since 1939

Abstract. The complete, existing, time series of K-indices from Norwegian observatories in Tromsø (TRO), Dombås (DOB) and Bear Island (BJN) has been digitized. The digitized time series are continuous spanning from 1939 (DOB) and 1947 (TRO) until 1998. Today, Tromsø Geophysical Observatory manages geomagnetic observations throughout Norway and K-indices are calculated in real-time with a fully automatic, in-house method. In this paper, the old, hand-scaled, and new, automatic, time series of K-indices are reviewed and compared for the intervals were they overlap. Our analysis confirms that the digital K-index series is a valid continuation of the old series. Since 1939, three K-index derivation methods have been applied to Norwegian magnetic observatory data. These are traditional hand-scaling, the method developed by the Finnish Meteorological Institute and an in-house method. Here, we compare the tree methods. It becomes clear that each method both have strengths and weaknesses. Importantly, differences arise when calculating the quiet-day variation, especially during periods of consecutive disturbed nights at auroral latitudes. By analysis of the K-index frequency distributions for six stations in mainland Norway and on Svalbard, it arises that the lower limit for K = 9 of 2000 nT is too high for TRO and K = 9 of 750 nT possibly too low at DOB. The assumption that X > Y, which makes it possible to calculate K for only the magnetic X component is investigated, and it is shown that the assumption is indeed only valid for auroral stations. In total, this paper presents all K-indices derived from Norwegian observatories since the nineteen-thirties until today, the used derivation methods and the long, historic time-series as a whole, and thus, enables a critical use of the indices for future scientific work.