AERMOD is most commonly used air dispersion model. Can we use this model for emission sources which are very close to coastal area (say less than 500 m)?
There is a multitude of dispersion models on the market (free or not). AERMOD is described as a steady-state plume model. Applicabilits really depends on the problem you want to approach. If it is traffic emissions, AERMOD might not be the right tool. If it is an industrial Point source you're looking at, you'll be fine.
A selection of free air modeling software can be found following the link.
For tall stacks (more than 65 m), we can not use AERMOD due to coastal fumigation where we expect an elevated pollution level.
In my case, I have collected few air sample from the creek area and quantified the pollutants. Now, I want to do modeling to see the dispersion of the observed concentration in the nearby area. That's why I had asked the question.
I understand why you ask this question, as coastal areas can have internal boundary layers that can affect the plume dispersion. CERC's model, ADMS 5, has a coastline option for this very reason. You can find more information on coastline effects for your literature review in the technical documentation, as given in the link below.
This includes a list of references for coastline effects and dispersion.