What is the spatial distribution of magnetic helicity injected in a solar active region?

Context. Magnetic helicity is suspected to play a key role in solar phenomena such as flares and coronal mass ejections. Several investigations have recently computed the photospheric flux of magnetic helicity in active regions. The derived spatial maps of the helicity flux density, called G(A), have an intrinsic mixed-sign patchy distribution. Aims. Pariat et al. ( 2005) recently showed that GA is only a proxy of the helicity flux density, which tends to create spurious polarities. They proposed a better proxy, G.. We investigate here the implications of this new approach on observed active regions. Methods. The magnetic data are from MDI/SoHO instrument and the photospheric velocities are computed by local correlation tracking. Maps and temporal evolution of GA and G. are compared using the same data set for 5 active regions. Results. Unlike the usual GA maps, most of our G. maps show almost unipolar spatial structures because the nondominant helicity flux densities are significantly suppressed. In a few cases, the G. maps still contain spurious bipolar signals. With further modelling we infer that the real helicity flux density is again unipolar. On time-scales larger than their transient temporal variations, the time evolution of the total helicity fluxes derived from GA and G. show small differences. However, unlike GA, with G. the time evolution of the total flux is determined primarily by the predominant-signed flux while the nondominant-signed flux is roughly stable and probably mostly due to noise. Conclusions. Our results strongly support the conclusion that the spatial distribution of helicity injected into active regions is much more coherent than previously thought: on the active region scale the sign of the injected helicity is predominantly uniform. These results have implications for the generation of the magnetic field ( dynamo) and for the physics of both flares and coronal mass ejections.