1998A&A...333..269P

We present a non-LTE Monte-Carlo radiative transfer code with clump and interclump media complementary to the one presented by Park et al. (1996A&A...312..981P) and derived from the original homogeneous model by Bernes (1979A&A....73...67B). The main difference is that instead of representing a very few real clumps treated individually, we consider an arbitrary large number of clumps, which we treat statistically with only a radial dependance. The main advantages are i) the incomparably faster convergence time, ii) the smooth profiles closer to observations and iii) the possibility to represent clumps as small as 1 millipc or less. The main disadvantages are i) to be limited to 1D geometries and ii) to lose the excitation dependence of the interclump gas as a function of its distance from each clump. The main results are that i) we partly confirm the large-scale excitation results presented by Park et al. (1996A&A...312..981P) (that is the interdependance of the two media excitation as a function of clump filling factor, the radial excitation profile shape and absolute value and the absence of emergent line profile variation as a function of clump filling factor in the particular case they studied) but we also extend them to other situations where the presence of clumps is shown to have indeed some strong influence on the emergent line profile, whether the interclump medium is optically thick or not, ii) we show that independent, gaussian, randomly moving clumps can actually give self-reversed or saturated profiles and iii) though we could improve on the previous tentatives to model ¹²CO and ¹³CO observation profiles in some difficult cases (Pagani et al., 1993A&A...275..573P; Pagani & Breart de Boisanger, 1996A&A...312..989P) we still cannot find a coherent description for both isotopic species and the ¹²CO line profile fit is still rough. We thus believe that though this kind of models with independent, randomly moving clumps is still simplistic and has room for improvements, it is not representative of the dense and clumpy molecular medium and that other directions should be explored. For example, hydrodynamical or magnetohydrodynamical models of the gas phase now produce large data cubes to which radiative transfer calculations could be applied. We also show that the noise reduction technique presented by Bernes is not valid in some cases and also is useless before the model has converged, i.e. before the last few iterations. Its usefulness is thus questionable.