Forecasting auroras from regional and global magnetic field measurements

We use the connection between auroral sightings and rapid geomagnetic field variations in a concept for a Regional Auroral Forecast (RAF) service. The service is based on statistical relationships between near-real-time alerts issued by the NOAA Space Weather Prediction Center and magnetic time derivative (dB / dt) values measured by five MIRACLE magnetometer stations located in Finland at auroral and sub-auroral latitudes. Our database contains NOAA alerts and dB / dt observations from the years 2002-2012. These data are used to create a set of conditional probabilities, which tell the service user when the probability of seeing auroras exceeds the average conditions in Fennoscandia during the coming 0-12 h. Favourable conditions for auroral displays are associated with ground magnetic field time derivative values (dB / dt) exceeding certain latitude-dependent threshold values. Our statistical analyses reveal that the probabilities of recording dB / dt exceeding the thresholds stay below 50% after NOAA alerts on X-ray bursts or on energetic particle flux enhancements. Therefore, those alerts are not very useful for auroral forecasts if we want to keep the number of false alarms low. However, NOAA alerts on global geomagnetic storms (characterized with K-p values > 4) enable probability estimates of > 50% with lead times of 3-12 h. RAF forecasts thus rely heavily on the well-known fact that bright auroras appear during geomagnetic storms. The additional new piece of information which RAF brings to the previous picture is the knowledge on typical storm durations at different latitudes. For example, the service users south of the Arctic Circle will learn that after a NOAA ALTK06 issuance in night, auroral spotting should be done within 12 h after the alert, while at higher latitudes conditions can remain favourable during the next night.Peer reviewe