This work aims to study the factors influencing dynamical changes in karst springs using Jinan area as an example. The study calculated the fractal dimension values and stability index of longterm multivariate data using time series fractal method. The database included the monitoring data of spring water level, rainfall, groundwater exploitation and artificial recharge in the Baotu and Heihu springs since the closure of groundwater supply source in 2003. The results show that the fractal dimension values of spring water level on a tenday scale are larger than that of the spring water level on a monthly scale. The fractal dimension of precipitation at rainfall stations in Ordovician limestone distribution area is smaller, but better precipitation infiltration and recharge conditions result in a closer relationship to spring water level fluctuation. Among the factors influencing the spring water level, the stability of recharge factor is the worst as it is highly vulnerable to the interference from external factors. The grey correlation degree was used to validate and analyze the degree of association between rainfall, artificial exploitation, artificial recharge with water level of the Baotu spring, which are 0. 858, 0. 647 and 0. 667 respectively, and the degree of association with water level of the Heihu spring are 0. 859, 0. 646 and 0. 668 respectively. The overall order of correlation degree is rainfall>artificial recharge>artificial exploitation, indicating that the atmospheric precipitation, not artificial exploitation, is the primary factor affecting spring water level dynamics since 2003. Based on the analysis of the evolution of spring resources, the regression analysis of fractal dimension and dynamic dimension of spring water level shows that the Ordovician limestone area is more suitable for artificial recharge of spring water. The results provide a reference for the protection of the Jinan karst spring.