Wide area sonar such as ocean acoustic waveguide remote sensing (OAWRS) has been shown to be a useful tool for instantaneously imaging large shoals of fish distributed over continental shelf regions. Here, we show that the population density of fish groups can be accurately estimated by using incoherent averaging of the matched filtered returns. Numerical Monte-Carlo models are applied to simulate the active imaging system and determine the statistics of the received matched filtered intensity scattered off Atlantic herring shoals in the Gulf of Maine. The model includes multiple scattering of the dense fish groups and uses a range-dependent acoustic model to simulate modal dispersion in the fluctuating ocean environment. We illustrate the effects of modal dispersion on the sonar imagery for various seafloor types. We determine the conditions for when multiple scattering is significant for Atlantic herring shoals, which depends on the fish density, imaging frequency, and mean target strength. Results of the model are compared to OAWRS sonar imagery collected during the 2006 Gulf of Maine Experiment.