I recommend you to read the file attached herewith related to basic theory of Buckingham dimensionless analysis and also related to landslide modeling for your reference.

We are doing a quite similar work on Landslide, where we are using a 3ft*2ft*2ft hollow trapezoidal module for physical modelling in the field.

The selection of the slope size depends on the scale factor you select to build your model such as 1:10, 1:20 or may be 1:5.

I feel we need to arrive proper scale factor between laboratory scale to proto type field case. I suggest you to perform Buckingham phi theorem scaling analysis and arrive proper scale for simulating in the laboratory. Though literature says 1:10 or 1:20 but correct scale is to be arrived by based on scaling analysis using Buckingham phi theorem.

We are doing it maintaining a slope factor 1:1 !!!!! to maintain a much steeper slope which can represent the extreme case scenario for slope modeling in hilly area for landslides in Bangladesh. I am literally very curious to know about the dimensional analysis through "Buckingham pi theorem", can you please give a suitable link here to understand this thing, Mr. Karthigeyan. S? It will mean a lot.

I recommend you to read the file attached herewith related to basic theory of Buckingham dimensionless analysis and also related to landslide modeling for your reference.

To select the model test slope size, you can scale down your model slope by considering a particular prototype. For example, if your selected scale is 1:n then your model slope size will be equal to prototype size * 1/n.

For any scale you choose you must have to conduct a dimensional analysis in order to recognize which phisical quantities you can scale and what is their effect in other quantities magnitude. For example, if you are not considering use a centrifugue, you can only scale down 2 quantities (lenght and force, for instance), and it will produce nonlinear effects in other results you measure.

in first step, you should extract the parameters that governing your problem, write all them down, write dimension of each parameters e.g. m, m/s, .. and some thing else. the try to find dimensionless parameters by only multiplication and dividing operation. then use phi Buckingham method. you can find my thesis at tarbiat modares university library.