Abstract: Due to the characteristics of sudden occurrence and diversity of influencing factors, the forecasting and prediction of karst ground collapse can only be defined in the area of dangerous areas, and it is impossible to predict when and where the dangerous area will occur.The existing investigation results of karst surface collapse show that the occurrence of karst surface collapse is precursory, that is, the water level fluctuation and turbidity of nearby wells will occur ahead of time. Based on this, this paper proposes to establish a sound groundwater observation network and carry on the real-time dynamic monitoring of karst water level, water volume, water turbidity and main chemical components, and can be combining with karst water flow parameters to achieve the higher precision karst ground collapse prediction for short time (hours to days) in small range (less than the monitoring network spacing) .It is illustrated by the Mengjiazhuang area of Laiwu City, Shandong Province.Due to the characteristics of sudden occurrence and diversity of influencing factors, the forecasting and prediction of karst ground collapse can only be defined in the area of dangerous areas, and it is impossible to predict when and where the dangerous area will occur.The existing investigation results of karst surface collapse show that the occurrence of karst surface collapse is precursory, that is, the water level fluctuation and turbidity of nearby wells will occur ahead of time. Based on this, this paper proposes to establish a sound groundwater observation network and carry on the real-time dynamic monitoring of karst water level, water volume, water turbidity and main chemical components, and can be combining with karst water flow parameters to achieve the higher precision karst ground collapse prediction for short time (hours to days) in small range (less than the monitoring network spacing) .It is illustrated by the Mengjiazhuang area of Laiwu City, Shandong Province.