The problem with modeling water quality is the same as those attendant to air quality: It's a three-dimensional subject area changing over time. Land is so much easier. The best suggestion I can offer is for you to check with the Esri pages and look to water modeling/sampling. Since the techniques would be highly dependant on the work you're doing you need to review availability of procedures. Alternately, in the technical lliterature look to articles that use mapping as an analytic and see how theyre' incorporating statistical and cartographic approaches.
In Louisiana where our streams often have near zero bottom slope we often have long periods of stagnation. Here and elsewhere these streams are at times called "stretch lakes." I believe combining terms "stretch lake" and model or GIS might yield some relevant references for you. For example the report MERMENTAU RIVER WATERSHED TMDL discusses DO modeling in a stretch lake.
HEC RAS is a physical model. It will only predict water quality to level of sediment resuspension. It is also 2D model so will not be appropriate for estuaries.
The other model mention here - LA-QUAL is not currently available for download so Hard to use.
The problem is to assess water quality you have to do physical, chemical and biological\ecological simulations.
I dont know a gis based model that currently does it.