Aquitard compaction is accompanied by transient changes in hydraulic parameters (i. e. , hydraulic conductivity K and specific storage Ss). The conventional “parameter optimization window” method commonly assumes that the initial condition of a physical process of interest is static, which is greatly inconsistent with the existence of an apparent time lag in aquitard drainage caused by the hydraulic response to groundwater level variations within bounding aquifers. From the perspective of highdimension parameter optimization, the transience of a parameter in the time domain is equivalent to the heterogeneity in the 1D space domain. Consequently, the geostatistical inversion approach can be used to characterize the transience of aquitard K and Ss, and this is the creative application of this methodology in hydrogeology. In this study, the longterm compaction of the lower confined aquifer system, which includes three aquitard lenses, was recorded at the extensometer site F16 in Shanghai, China. The classical quasilinear geostatistical approach was employed to interpret the observed compaction. The inversion results revealed that aquitard K varied between 1. 02×10-5 m/a and 5. 89×10-4 m/a, with an average value of 1. 38×10-4 m/a, while Ss varied between 4. 34×10-4 m-1 and 6. 02×10-3 m-1, with an average value of 2. 65×10-3 m-1. The inverse transience of aquitard K and Ss suggested that the aquitard compaction turned from elastic to inelastic behavior around the year 1994. Furthermore, the inversed K demonstrated a seasonal variation pattern that was reversed to that of the aquifer groundwater level. This abnormal phenomenon was analyzed as being partially attributed to the alternation of attachment and detachment of finegrained particles in the pore space near the contact surface between the aquifer and the aquitard lens. The proposed new parameter estimation method demonstrates the ability to capture temporal variations of model parameters within a complete monitoring period. This suggests potential application value in calculating and evaluating the temporal variations in streambed permeability as well as the permeability of permeable reactive barriers used for insitu, passive contamination remediation, and so on. Also, this newly introduced methodology can serve as a supplementary means for constructing relative constitutive models in the research areas of hydrogeology and geotechnical engineering.