Reliable noise mapping is much needed to manage noise exposure from novel drone traffic, Urban Air Mobility (UAM). Dedicated noise mapping tools, such as SAFTu, NoiseModelling, and SoundPLANnoise, require accurate source models which represent the sound emissions of the drone in operation. In relation to established civil aviation noise modelling standards, such as ECAC Doc. 29 and associated NPD data derived from ICAO Annex 14/16 certification, neither noise certificates for Unmanned Aerial Vehicles (UAVs) nor certification standards are in place to date. Until certification frameworks are in place, a methodology for establishing source data needs to be developed from voluntary test data. When developing such methodology, the different operational modes associated with VTOL propulsion systems need to be accounted for. Operation at lower altitudes necessitates consideration of entire flight trajectories, including en-route phases, rather than only take-off and landing. Recent EASA guidelines for in-situ noise measurements of UAVs up to 600 kg represent an important step toward realistic drone noise assessment; however, the omission of take-off and landing measurements, as well as the lack of frequency-dependent data, constitutes a limitation for environmental noise studies. In the present work, it is shown how field-measured noise data can be used to derive acoustic source models with controlled uncertainty. Moreover, simplified physics-based models, based on e.g. vehicle mass and rotor properties such as tip speed, diameter, and pitch are discussed in the context of extrapolating measured acoustic data at a given flight mode. Such models can also be used to support vehicle design optimization work regarding noise generation. Methodologies for converting limited experimental data sets into source models are also discussed, e.g., when directivity data is limited or missing, or when field test data are affected by ground reflections. Calculation examples are presented, highlighting how field test data, in combination with physical sound generation models, can be used to define source models to be used in environmental noise studies.