Water quality modelling software is easier to choose than it looks, because the shape of the water body rules out most of the options before price comes into it. A river reach, a stratified reservoir, an estuary and a catchment each need something different, and most of the well known models are built for one of those and awkward in the others.
What is the best water quality modelling software?
There is no single best one; the right model is the one built for your water body. QUAL2K for a river reach at steady state, CE-QUAL-W2 for a stratified reservoir, WASP for lakes, estuaries and anything that has to vary in time, a 3D package such as Delft3D or MIKE for coasts and estuaries, and SWAT or HSPF when the question is the load coming off the land.
Which water quality models are free?
Most of the standard ones. QUAL2K, WASP, CE-QUAL-W2, HSPF and SWAT are all available at no cost, and several of them are the models regulators in the United States expect to see in a TMDL. HEC-RAS, also free, includes water temperature and nutrient modelling on top of a hydraulic model you may already have built for flooding. Deltares publishes the source code of Delft3D, though its packaged suite is licensed separately. Licence terms change, so confirm the current position with each developer.
The commercial options are DHI's MIKE ECO Lab, which runs on the MIKE hydrodynamic models, and supported versions of the open codes. What they sell is integration, support and workflow rather than different chemistry.
Rivers
For the classic river question, a discharge enters and you need dissolved oxygen and nutrients downstream at low flow, QUAL2K is the fastest defensible route. When the answer has to vary in time, through a storm or a season, WASP takes over. We draw the line between those two in detail in WASP vs QUAL2K.
If a HEC-RAS model of the river already exists, its water quality module is worth checking before building anything new, because the hydraulics are the expensive half and they are already done.
Lakes and reservoirs
The deciding question is stratification. A reservoir with a thermocline behaves as two water bodies, one above and one below, and a model that averages them away will get oxygen and nutrients wrong in exactly the layer that matters. CE-QUAL-W2, from the US Army Corps of Engineers and Portland State University, is two dimensional and laterally averaged, which is the shape of a long narrow reservoir, and it is the usual choice there. The General Lake Model, an open source one dimensional model, suits smaller lakes where vertical structure is the whole question.
Estuaries and coasts
Tides, salinity and three dimensional mixing push the problem to a full hydrodynamic model with water quality attached. Delft3D's water quality module from Deltares, MIKE ECO Lab from DHI, and EFDC, a model supported by the US EPA with a commercially maintained version, all work here. The choice between them is usually made by the hydrodynamic model the project already uses and by what the client's reviewers know.
Catchments and diffuse loads
When the question is how much nitrogen, phosphorus or sediment comes off the land, it is a catchment question rather than a receiving water one. SWAT, from the USDA Agricultural Research Service and Texas A&M, and HSPF, from the US EPA, simulate loads from land use, soils and climate. In practice they are often paired with one of the receiving water models above, the catchment model producing the loads the river or lake model then routes and reacts.
Pollution in an aquifer is a different problem again, with its own codes. We cover those in groundwater contamination modeling software compared.
How to choose
Draw the water body and write down the question. A line and a steady profile means QUAL2K. A stratified reservoir means CE-QUAL-W2. Anything time varying, branching or tied to sediment means WASP. A tidal or coastal system means a 3D hydrodynamic model with water quality attached. A question about loads from the land means SWAT or HSPF, usually feeding one of the others.
What actually determines the answer
Water quality results are decided by boundary loads and by monitoring data far more than by the model. Loads are routinely estimated rather than measured, and sonde records drift and foul in ways that look like real events. Calibrating any of these models to an unscreened record produces a model that reproduces the sensor, not the river. Screening the data comes before choosing the software, not after.
Smart Bhujal delivers water quality studies in QUAL2K, WASP and CE-QUAL-W2, set out on our water quality modelling page. If you are scoping a discharge consent, a TMDL or a reservoir study and are not yet sure which model the water body needs, that is worth settling before anything is built, and where the result has to be used continuously rather than filed, it becomes a water decision support system.
