How to determine the number of vertical layers

What methods can be used to set the number of vertical layers? Can you recommend some references or share your experiences in determining such a number?

Some reports or papers just say they use e.g. 10 vertical layers and do not mention why. It would be nice to have vertical data to calibrate against, but what to do when calibration data are not available?

Is there a general rule from grid size and water depth?




Choosing the number of layers in the model is often a iterative process. Though you don’t have the measured data available you should try to understand more about the study area. If you are looking at a area where freshwater and salt water interaction takes place or a power plant discharge where you want to trace the temperature profiles etc. then you should use more layers. When I refer more layers, it is also relative. So, it is generally a good idea to use the same grid and make couple of runs with different vertical layers and do some sensitivity analysis. For example run a model with 6 layers and 12 layers and see whether you get similar vertical profiles or the vertical profiles in 12 layers are a little better. You should decide wisely as there is a trade off between number of layers and the computational time. I haven’t seen any rule based on grid size and water depth.

I hope it helps.

Let us know if you have any more questions.

Janesh Devkota