ATM 2009 largely improves the modelling of the atmosphere emission and absorption compared to ATM 1985 (e.g. introduction of minor atmospheric components, better (pseudo)-continuum modelling). However, the ATM 2009 version used in GILDAS shows a problem of continuum for a zero amount of water vapor. In addition, a negative amount of water is forbidden by ATM 2009, while this possibility can be useful when systematic biases dominate the measure (e.g. 3mm observations during sub-millimeter conditions). J.R. Pardo is interested by 30m skydip results to constrain the dry and wet continuum opacities under 200 GHz in ATM 2009.
From the implementation viewpoint, ATM 2009 version is robust. It can be run routinely without risk of crashing. The GILDAS interface to ATM 1985 and ATM 2009 was thought to enable an easy swap from one version to the other. When swapping from ATM 1985 to ATM 2009, variations of the calibration results are expected. In particular, the accuracy of is given by Eq. 18. It directly depends on the accuracy of the coupling factors and on the modeled opacity. It is better to underestimate the opacity than to overestimate it.