What is a GMC? Are observers and simulators discussing the same star-forming clouds?

As both simulations and observations reach the resolution of the star-forming molecular clouds, it becomes important to clarify if these two techniques are discussing the same objects in galaxies. We compare clouds formed in a high-resolution galaxy simulation identified as continuous structures within a contour, in the simulator's position–position–position (PPP) coordinate space and the observer's position–position–velocity space (PPV). Results indicate that the properties of the cloud populations are similar in both methods and up to 70 per cent of clouds have a single counterpart in the opposite data structure. Comparing individual clouds in a one-to-one match reveals a scatter in properties mostly within a factor of 2. However, the small variations in mass, radius and velocity dispersion produce significant differences in derived quantities such as the virial parameter. This makes it difficult to determine if a structure is truly gravitationally bound. The three cloud types originally found in the simulation in Fujimoto et al. are identified in both data sets, with around 80 per cent of the clouds retaining their type between identification methods. We also compared our results when using a peak decomposition method to identify clouds in both PPP and PPV space. The number of clouds increased with this technique, but the overall cloud properties remained similar. However, the more crowded environment lowered the ability to match clouds between techniques to 40 per cent. The three cloud types also became harder to separate, especially in the PPV data set. The method used for cloud identification therefore plays a critical role in determining cloud properties, but both PPP and PPV can potentially identify the same structures.