Numerical processing of sunspot images using the digitised long-term archives

We present the results of applying our numerical algorithms for identification and parametrisation of sunspots from a long series of photographic observations made in white light during the period of time from 1918 to 1972 by the Royal Greenwich Observatory (RGO) and Kislovodsk astronomical station from 1954 to 2014. The main purpose of this work was creating a catalogue of individual sunspot parameters, which is a step forward with respect to the previous catalogues of this kind. The RGO archive contains about 26000 photographic plates, and Kislovodsk more 18500 plates. All plates digitised with the resolution that corresponds to the Sun's image radius having approximately 1600 16-bit pixels. We have developed special numerical algorithms and implemented them in our image analysis software. The first stage of processing each photographic plate consists of identifying the solar disc image and projecting onto this image the heliographic coordinate grid corresponding to the date and time of observation. Then, we calculate the local Quiet Sun Level (QSL) brightness, which is used for determining the local contrast threshold for sunspot identification and localisation. Our software includes also manual screening of the automatically localised sunspots, which was used at the last stage of the solar image processing to filter or remove unavoidable photographic plate artefacts. In total, we have localized about 240 thousand ind ividual sunspots in RGO archive and about 200 thousand sunspots in Kislovodsk data. The calculated areas of these spots were compared with the results of manual measurements of these areas (www.ngdc.edu) and showed high correlation (R~0.98). We performed the analysis of the distributions of the leading and trailing sunspots. In the distribution of the leading sunspots have a maximum area of S ~ 120-150 mhm. In the distribution trailing sunspots such maximum is absent.