Abstract: Capillary action has an important impact on the deformation and stability of various geotechnical engineering structures. It is challenging to accurately monitor the dynamic changes in the capillary water rising process by traditional methods. In this study, the relationship between RGB (red, green, blue) information and water content of aeolian sand images was firstly investigated. The methodology and technology for measuring water content and obtaining centimeter-level resolution moisture fields via image RGB information were developed. According to this method, the spatial-temporal variation of water content during capillary water rising in aeolian sand was further explored, and the wetting front profile was precisely distinguished. Furthermore, the wetting front transport law was analyzed. The results show that a negative linear relationship exists between the normalized color feature ξRnor and the water content θ of aeolian sand, which brings about a good prediction for the water content of aeolian sand. The proposed technology for measuring the water content field based on image RGB information, with centimeter-level resolution and high accuracy, can visualize the spatiotemporal changes of the water content during the rising of capillary water in the aeolian sand column. An improved k-means clustering segmentation method was adopted to analyze the sand images of the capillary water rising process, allowing precise identification and quantification of the profile information of the wetting front. This method is more accurate and reliable than the visual inspection method. Under the action of surface tension and inertial force, the water content at each height of the aeolian sand in the initial capillary water rising period has obvious overshoot phenomena. After several small fluctuations, it falls back, and the stable water content is about 1.5% lower than the peak value. The rising height of the capillary water is proportional to relative density D_r. The rising process of the aeolian sand wetting front can be well fitted both by the power and double logarithmic quadratic polynomial functions. The research provides an accurate and rapid novel approach for the capillary water rise test.

Key words: aeolian sand, capillary action, digital image, moisture field, wetting front profile