We used 40 high resolution, high S/N QSO spectra at 2.1<z<4.7 to search for the signature of the proximity effect in the HI Lyalpha forest. Comparing the effective optical depth near each QSO with the expected one, we clearly detect the proximity effect on the combined QSO sample and towards each individual QSO. The observed proximity effect strength distribution (PESD) is asymmetric towards a weak effect. We demonstrate that this is not simply an effect of gravitational clustering around QSOs. Comparing simulated PESDs with observations, we argue that the averaging method to determine the UVB intensity J is heavily biased towards high values because of the PESD asymmetry. Using instead the mode of the PESD provides an unbiased estimate of J. For our sample its modal value is log(J)=-21.51+/-0.15 (in units of ergcm^-2s^-1Hz^-1sr^-1) at z=2.73. We estimated the excess HI absorption attributed to gravitational clustering. On scales of ~3 Mpc, only a minority of QSOs shows overdensities of up to a factor of a few in tau_eff; these are exactly the objects with the weakest proximity effects. After removing them, we redetermined the UVB intensity arriving at log(J)=-21.46+0.14-0.21. This is the most accurate measurement of J to date. We present a new diagnostic based on the shape of the PESD which strongly supports our conclusion that there is no systematic overdensity bias for the proximity effect. This additional diagnostic breaks the otherwise unavoidable degeneracy of the proximity effect between UVB and overdensity. We estimated the redshift evolution of J and found tentative evidence for a mild decrease with increasing redshift. Our results are in excellent agreement with predictions for the evolving UVB intensity, supporting the notion of a substantial contribution of star-forming galaxies.
Download