Forecast evaluation of the coronal mass ejection (CME) geoeffectiveness using halo CMEs from 1997 to 2003

In this study we have made a forecast evaluation of geoeffective coronal mass ejections (CMEs) by using frontside halo CMEs and the magnetospheric ring current index, Dst. This is the first time, to our knowledge, that an attempt has been made to construct contingency tables depending on the geoeffectiveness criteria as well as to estimate the probability of CME geoeffectiveness depending on CME location and/or speed. For this, we consider 7742 CMEs observed by SOHO/LASCO and select 305 frontside halo CMEs with their locational information from 1997 to 2003 using SOHO/EIT images and GOES data. To select CME-geomagnetic storm (Dst < -50 nT) pairs, we adopt a CME propagation model for estimating the arrival time of each CME at the Earth and then choose the nearest Dst minimum value within the window of +/- 24 hours. For forecast evaluation, we present contingency tables to estimate statistical parameters such as probability of detection yes (PODy) and false alarm ratio (FAR). We examine the probabilities of CME geoeffectiveness according to their locations, speeds, and their combination. From these studies, we find that (1) the total probability of geoeffectiveness for frontside halo CMEs is 40% (121/305); (2) PODys for the location (L < vertical bar 50 degrees vertical bar) and the speed (>400 km s(-1)) are estimated to be larger than 80% but their FARs are about 60%; (3) the most probable areas (or coverage combinations) whose geoeffectiveness fraction is larger than the mean probability (similar to 40%), are 0 degrees < L < +30 degrees for slower speed CMEs (<= 800 km s(-1)), and -30 degrees < L < +60 degrees for faster CMEs (>800 km s(-1)); (4) when the most probable area is adopted as the new criteria, the PODy becomes slightly lower, but all other statistical parameters such as FAR and bias are significantly improved. Our results can give us some criteria to select geoeffective CMEs with the probability of geoeffectiveness depending on the location, speed, and their combination.