Field and laboratory study demonstrates accurate mathematical modeling of sandy soil permeability, indicating effective prediction through geotechnical parameters.
Key Points
To develop a mathematical model estimating in situ soil permeability by correlating field measurements with laboratory geotechnical parameters.
Gathered field and laboratory data from 20 exploration points across five representative coastal zones.
Conducted in situ permeability tests and driven-cylinder density measurements alongside laboratory tests for grain size, angularity, specific gravity, and permeability.
Evaluated 13 geotechnical variables using correlation analyses and LASSO regression in Python to address multicollinearity and select key predictors.
Laboratory-measured permeability was identified as the most influential variable for predicting in situ permeability.
The final LASSO model retained six predictors (laboratory permeability, angularity, fines content, moisture content, specific gravity, and D10), achieving strong predictive accuracy (R² = 0.81).