There are two approaches for doing this. One is to generally identify landform regions, and here is an example: Article Modeling global hammond landform regions from 250-m elevatio...

The second is to identify, at a particular scale, the landforms, and here is an example of that: Article Ecologically-Relevant Maps of Landforms and Physiographic Di...

The difference is the second method finds landforms on landforms, i.e., the hills and valleys on the mountain, while the first identifies mountainous regions.

some tools offering automated landform classification based on DEMs:

GRASS

• r.param.scale: Under the "feature" method, your dem is classified into peaks, ridges, passes, channels, pits and planes. The good thing is that the window size can be adjusted so you can calculate these forms at several spatial scales. https://grass.osgeo.org/grass72/manuals/r.param.scale.html
• r.geomorphons: External module which derives a wide range of landform combinations based on a more sophisticated approach: flat, summit, ridge, shoulder, spur, slope, hollow, footslope, valley and depression

  Article Geomorphons-a pattern recognition approach to classification...

SAGA

• TPI Based Landform Classification http://www.saga-gis.org/saga_tool_doc/2.2.3/ta_morphometry_19.html
• Fuzzy Landform Element Classification http://www.saga-gis.org/saga_tool_doc/2.2.1/ta_morphometry_25.html

We have compiled these approaches in a short article and compared the outcomes for different DEMs:

Chapter Terrain analysis and landform recognition

Thank you so much for your suggestions, I was able to use the GRASS GIS software to produce 10 geomorphons.

Is it possible to reclassify the 10 geomorphons into five classes of landform namely:

I would like to create a soil landscape map by combining the landscape map of five classes with the landtype soil map. This would allow us to disintegrate the landtype soil map of South Africa into soil types which will then be used as input data in SWAT modelling.

Thank you

Geomorphons in Grass