The magnitude 8 (8M) earthquake (“Wenchaun earthquake”) that hit Sichuan province on May 12, 2008 highlighted the need for the development of large-scale early warning systems in the region surrounding the Longmenshan fault. Many of the deaths from this disaster were not from the earthquake itself, but from the landslides and other hazards that it caused. To this end, real-time monitoring and early warning of mountainous hazards (e.g. landslides, debris flows) are efficient non-engineering measures for disaster prevention and mitigation. Thus, to ensure effective disaster prediction and reduction within this region, a system of real-time regional and site-specific monitoring of mountainous hazards, coupled with early warning, is necessary. We propose here a multi-scaled, real-time early warning method. It combines large-scale regional hazard calculations and site-specific hazard monitoring with subsequent early warning. Real-time monitoring and early warning systems were constructed in five mountain hazard sites along the Longmenshan fault line, including three debris flow and two landslide zones. For early disaster warning, the threshold value of precipitation needed to trigger the post-earthquake mountainous hazard should be measured. However, identification of such threshold values is one of the most difficult issues for regional and site-specific mountainous hazard monitoring and early warning systems. To account for this difficulty, this study utilized a natural coseismic landslide from the Taziping Village of Hongkou County in Dujianyan City to identify the critical values and threshold parameters that resulted. After this experimental field test, the correlation of rainfall intensity with rainfall duration was calculated. The results indicated that the experimental field test was capable of identifying the threshold factors for the rainfall-induced mountain hazards.