Abstract: The deformation behaviors of an arch dam that are determined by multi-factor coupling are usually characterized by variations in space and time. Based on a spatiotemporal model of variable intercept panel data, this study makes full use of the deformation data from an array of gauging points at Jinping arch dam to examine its deformation behaviors, and solves the problem of conventional statistical models that are limited to a single data point of the time sequence in dam deformation analysis. The results show this panel model accurately detects the spatiotemporal characteristics of dam deformation responses and is validated by its good fitting accuracy and extension. In addition, it is capable of controlling the differences between gauging points and accurately evaluating the impact of each component on different zones of the dam body, thereby remedying the shortcoming of conventional models. And it manifests lower multi-factor collinearity and robustness among other excellent properties and thus helps promote the level of online dam safety monitoring.

Key words: panel data, spatiotemporal model, safety monitoring, deformation behavior, analysis and prediction