Spatiotemporal differentiation recognition of dam deformation using Gaussian-ESDA model

The reasonable interpretation of deformation behavior is critical for the health diagnosis and safety assessment of dams. Existing studies often focus on the time-domain characteristics of single-point deformation or spatial averaging characteristics of the multi-point deformations, and fail to reveal and explain the deformation spatiotemporal differentiation. To address these issues, a deformation spatiotemporal differentiation recognition method is proposed based on Gaussian-corrected exploratory spatial data analysis. The application results of a real concrete dam show that the dam deformation share significant spatiotemporal differentiation since its Z-score exceeding 2.58 at 1% significance level, the center of gravity and hot (cold) spots changing seasonally, as well as spatial stability indices exceeding 90%, which verifies the feasibility of the proposed method for the interpretation of deformation behavior. Compared with the natural breaks(jenks) method and the classical ESDA method, the proposed method is superior in time effect recognition of dam deformation, with the 10%∼20% shrinkage in hot (cold)-spot area and 5% improvement in significance level.