Quantitative Analysis of Soil Bearing Capacity Using Image-Based Morphology and Capacitive Moisture Sensing

Abstract

This study proposes a rapid Pedon-scale screening framework for in-situ soil bearing capacity by integrating image-based soil morphology and capacitive moisture sensing. Six disturbed soil samples were collected from KIPP IKN (Nusantara), Mandalika (Central Lombok), Pulau Seram (Central Maluku), Desa Mayang (Sukoharjo), Penajam Paser Utara, and Ciracas (East Jakarta). Morphological structure was quantified from smartphone images acquired under controlled back-illumination using grayscale conversion, adaptive histogram equalization, Otsu thresholding, and morphological opening to obtain an image-derived porosity index treated as a void-ratio proxy. Moisture content was measured using a low-cost capacitive probe and mapped to percentage water content. Laboratory references followed ASTM standards (e.g., ASTM D2216 and D854) to compute gravimetric water content and void ratio. Agreement between rapid estimators and laboratory values was assessed using RMSE/NRMSE, normality testing, and Pearson correlation. The capacitive sensor achieved RMSE = 2.4366% (NRMSE = 7.56%) with r = 0.9956, while the image-based void-ratio proxy achieved RMSE = 0.0255 (NRMSE = 18.89%) with r = 0.8670. A rule-based fusion of porosity–moisture classes aligned fully with geotechnical bearing-capacity classification across sites. The proposed workflow offers a portable, low-cost alternative for rapid preliminary subgrade and foundation screening, while acknowledging uncertainty sources (illumination variability, heterogeneity, and sensor drift) for future field-scale validation.