Navigation locks are critical infrastructure components, and their closure for maintenance and repair can have significant impacts on the global economy. The current state of inspection and monitoring of lock components is generally to close the lock and perform a visual inspection. While structural health monitoring (SHM) of navigation locks is gaining acceptance, automation of the SHM process is lacking. This paper reports on efforts to develop an automated damage detection system for miter gates of navigation locks. The study focuses on using strain gage measurements to identify the redistribution of load throughout lock gates in the presence of damage. To eliminate the environmental variability in the data, a new damage sensitive feature is introduced, termed here as "slope" and defined as the derivative of the strain with respect to the water levels in the lock chamber. The slopes form a new, stationary time series effectively purged of environmental effects. Principal Component Analysis (PCA), a method of analyzing multivariate, stationary time series, is then used to detect significant changes in the statistics of slopes as an indication of damage. To validate the approach, damage is simulated in a finite element model, and the resulting changes in strain from the finite element model are superimposed on the measured data. The results demonstrate the potential of the proposed approach for detecting damage in navigational lock gates.