Proposal and Applicability Assessment of a Computer Vision–Based Quantification Method for Seawater Behavior in Coastal Aquifers

Complex interactions between seawater and freshwater govern intrusion and stabilization in coastal aquifers. Previous studies have primarily relied on interface-based analyses, limiting full-field spatial quantification of seawater distribution. This paper proposes a computer vision–based method for quantifying seawater behavior in laboratory sand tank experiments. Experimental videos were acquired under controlled hydraulic and lighting conditions, and an image analysis framework was established in Python using OpenCV, including preprocessing, red–green–blue (RGB) and hue–saturation–value (HSV) color-space analysis, threshold determination, binary segmentation, region extraction, area calculation, and temporal filtering. Rhodamine-B dye enabled robust segmentation, and the H channel provided the most reliable separation with reduced sensitivity to illumination variations. The extracted seawater area time series showed an initial rapid increase, a temporary decrease associated with interface readjustment, a gradual reintrusion stage, and eventual stabilization. A fourth-order Butterworth low-pass filter with a cutoff frequency of 0.001 Hz was applied to suppress the high-frequency noise and clarify the convergence toward seawater–freshwater pressure equilibrium. Stabilization was defined as the stage at which the one-hour-averaged relative area-change rate fell below 1%. These findings suggest that the proposed framework enables the non-intrusive, reproducible quantification of seawater behavior and provides a practical alternative to conventional interface-based methods.