I agree with the previous answer. I suggest you to read this paper:
Scesi L., Gattinoni P. (2009): Study Of The Interactions Between Rivers Dynamic And Slope Stability For Geohazard Prediction: A Case In Val Trebbia (Northern Italy). International Symposium on Prediction and Simulation Methods for Geohazard Mitigation (IS-Kyoto2009), 25-28 May, Kyoto, Japan, pp. 241-246.
Yes, I agree that slope gradient is one of the criteria may be considered for landslide mapping besides other factor,specially when using remote sensing and GIS technology. It can be apply either as a single criteria or multi criteria analysis for landslide study.
This is one of several factors for the hill instability. But you could use actually, more like a 'supporting' fact though. As previously mentioned: lithology, bedding/cleavage, climate, etc. together build final conditions.
Steep hillslope is mostly susceptible for the landslide processes, but it also is affected by geology (rocktype and structure). Identification of the streams in hillslope with the help of DEM is also a difficult task. With the help of topographical maps one can easily identify the individual streams and their gradient.
You can try to fit some analytical models with the help of existing landslide events data and stream gradient. These models will be different in different geological area. So simply with the help of stream gradient you can not identify the landslide prone areas. I have attached an article. You can try to use this.
we investigate the sensitivity of landslide susceptibility assessments to several possible input parameters.
We consideredalso Topographic Wtness Index and Flow Accumulation. The results depend on the DTM resolution, however, in general, the relative importance of these parameters is very lower than other possible input parameters such as curvature or slope gradient (just to stick to parameters that may be strongly connected with hillslope hydrology).
Hope it could help!
Article Landslide susceptibility estimation by random forests techni...