The sources of water vapour and other gases can be the nucleus surface and the dust particles in the coma. In the second case, they are referred to as distributed sources. It is a tradition derived from ground-based observations to infer the presence of distributed sources when the gas column density is a power law of the nucleus distance with an index shallower than - 1. Rubin et al. (2011)have shown that this approach (applied to coma gas density when the index is shallower than - 2) is misleading when applied to in situ data. One of the reasons is again linked to the nucleus shape, which makes the diurnal variations of the local gas density larger than those due to the nucleus distance. In order to infer distributed sources from gas data collected in situ, we must again rely on 3D coma models. A comet where most water is coming from distributed sources is 103P/Hartley 2 (103P hereinafter). The water-loss rate observed from the ground is 9 × 10 27 mol s - 1 (Combi et al. 2011). At the same time, 3D coma models infer a total flux from the nucleus surface of 2 × 10 27 mol s - 1 and a ratio between night and sub-solar water loss of 30 per cent (Fougere et al. 2013). For 67P, the water-loss rate observed from the ground (Keller et al. 2015b) matches that inferred at the nucleus surface by 3D coma models (Fougere et al. 2016). The night-to-day water-loss ratio assumed by 3D coma models at the nucleus surface is 2 per cent. This night water-loss rate is an ad hoc assumption adopted to fit the ROSINA-COPS data collected at the terminator, and cannot come from the entire night-side nucleus surface: its temperature is too low to eject so much water (Choukroun et al. 2015). It may come from the surface just after sunset due to its thermal lag (Shi et al. 2016), or from distributed sources. Assuming a process similar to that occurring in 103P, where 80 per cent of water from distributed sources requires a night-to-sub-solar ratio of the water-loss rate at MNRAS 462, S2–S8 (2016) by guest on November 14, 2016
http://mnras.oxfordjournals.org/ Downloaded from Unexpected and significant findings in 67P S5 the surface of 30 per cent (Fougere et al. 2013), we can conclude that in 67P, at most 5 per cent of water is coming from distributed sources. This conclusion is consistent with the accuracy of the observed 67P water-loss rate. This water content is much lower than the dust-to-water mass ratio of six in the nucleus. Comet 67P ejects very dry dust, where with dry, we define dust containing less water than 15 per cent of its mass (dust is wet in the opposite case). From Earth (Fulle et al. 2010;Morenoetal. 2016) and from Rosetta (Rotundi et al. 2015; Fulle et al. 2016) we observe the same dust-loss rate, thus excluding that dust releases a significant water mass. A sublimating and/or fragmenting particle decreases its mass during its travel in space, implying a strong change in the dust size distribution from the nucleus surface to Rosetta, and from Rosetta to the outer coma observed from ground. The dust size distributions at the nucleus surface observed in the smooth plains by ROLIS (Mottola et al. 2015), inferred from Rosetta observations (Rotundi et al. 2015; Fulle et al. 2016), and inferred from ground-based observations (Fulle et al. 2010;Morenoetal.2016), do not show any systematic shift from large to small sizes. COSIMA observes dry dust (Schulz et al. 2015). Dust in 67P’s coma does neither sublimate nor fragment more than 5 per cent of its mass, confirming that most dust mass is ejected in compact particles.
!).
!
! Sunlight ionizes gas.
!
! 
