The extraction of noise source levels from ambient noise measurements requires accounting for the seismoacoustic propagation from the surface generated noise sources to the field measurement positions. At frequencies below 100 Hz, the waveguide nature of the environment strongly influences the distribution of ambient noise. Only when these propagation effects are considered can measurements from different experiments taken in different environments be combined to ascertain the global properties of surface noise sources. Here, such a previous analysis of a shallow water experiment [Schmidt et al., in Natural Mechanisms of Surface Generated Noise, edited by B. Kerman (Reidel, Dordrecht, The Netherlands, 1988)] is used and the results are combined with Kibble‐white and Ewans' lower‐frequency results [J. Acoust. Soc. Am. 78, 981– 994 (1985)], the latter also containing a summary of previous experimental results obtained by others. These data are treated in the same way by accounting for the environments using a wave theory model of distributed noise [W. A. Kuperman and F. Ingenito, J. Acoust. Soc. Am. 67, 1988– 1996 (1980)]. When the particular environments are used, the spread in the reprocessed noise levels is substantially reduced, showing more consistency in the frequency dependence of the noise sources, suggesting that there are only a few (or possibly only one) dominant natural noise source mechanisms that primarily contribute to ocean ambient noise below 100 Hz. The importance of appropriately including propagation factors in processing noise data for estimating source levels is therefore strongly demonstrated.