An important result from observations of the pre-CME corona using Yohkoh/SXT is that sigmoidal loop structures are predictors of major eruptions . These sigmoidal structures are, in turn, thought to result from concentrations of magnetic helicity in the corona. CMEs also have a strong correlation with the cooler eruptive prominences that are often embedded at the base of streamers. Through it's different EUV-filters, EIT has been able to observe both these sigmoidal loop structures (coronal bandpasses) as well as the eruptive prominences (He II transition region bandpass). Observations revealed that the initiation of a CME is evident on both very small and very large spatial scales. A typical CME well-observed with the EIT shows the rapid eruption of a portion of cool prominence material accompanied by a small brightening observable as a weak X-ray flare. Running difference EIT movies frequently display the classical 3-part CME structure: a bright frontal loop surrounding a dark cavity containing a bright core .
Using advanced image recognition methods , we propose to make a systematic scanning, of the complete EIT and LASCO archives. This will result in a catalogue of all CME related events (streamers, prominences, sigmoidal loop structures, ...) that occurred since the last solar minimum. From these catalogues we can then extract statistically `typical' CME initiation scenarios.
MHD simulations of CME initiation will be performed based on the parameters and configurations derived from these typical CME initiation scenarios. The self-similar solutions of coronal transients with magnetic structures , used as starting point, will be tuned to fit the observed parameter values. The perturbations of the boundary conditions (shearing and flux emergence/cancellation) will be chosen consistent with numerically obtained solar wind solutions with partially closed magnetic fields .
Changes in the pre-CME magnetic configuration are an important aspect of the CME eruption process. Feynman and Martin  have demonstrated the role of newly emerging magnetic flux in the initiation of eruptive prominences. >From a modeling approach, magnetic helicity is almost conserved. The corona gets rid of the helicity injected by the convection zone only by ejecting part of the magnetic field. The Aly-Sturrock conjecture implies that a closed magnetic field has less energy than the equivalent fully open field, with the same photospheric boundary condition. It is however possible to open the field partially or to eject a twisted flux-rope keeping the energy of the field below the open field limit. The Meudon group is developing extrapolation codes using longitudinal magnetic field measurements (magnetograms of SOHO/MDI with a cadence of 96 min) . This group also studies the 3D topology of active regions to understand which locations have high probability of flare occurrence. The codes allow to draw the quasi-separatrix surfaces and find the separators (equivalent to the X reconnection point in 2D model). During observation campaigns, this theoretical tool is used to define the targets. It is proposed to develop a data base of these codes which could be used to forecast the location of CMEs.
After initiation, CMEs continue to have a profound effect on the solar atmosphere. Common signatures seen by EIT include the rapid motion of wispy coronal features, footpoint dimming , and coronal EIT waves  propagating across the solar disk at speeds of 200-300 km/s. Observations by LASCO show that the slow swelling of a helmet streamer (the ‘bugle’ pattern) is often a precursor, by several hours, of a CME eruption. After initiation of the CME, LASCO C1 coronagraph images suggest that strong structural changes occur within the 1.1 to 3 range . There is a large variation in time-height profiles of CMEs. LASCO observes acceleration as high as m/s2 and speeds as high as 2000 km/s.
In this project we will develop software tools that take the EIT/LASCO instrument characteristics as input, and calculate from the numerical simulations `artificial observations'. This will allow to close the circle back from simulations to observations, and check if the simulations reproduce observed signatures such as the footpoint dimming and the coronal EIT waves, and observed time-height profiles of CMEs.