Development of Local Conversion Models to Estimate the Electrical Conductivity of Saturated Soil Paste Extracts from Water Dilution Extracts (1:1, 1:2, 1:5) in the Haflez and Kambut Depressions, Eastern Tobruk, Libya

Abstract

This study aimed to develop local conversion models between ECe and extract dilution ratios (1:1, 1:2, 1:5) in three semi-arid depressions in eastern Libya. A total of 45 surface soil samples (0–20 cm) were collected from the Hafilz (HZ), Southern Kambout (K1), and Northern Kambout (K2) depressions, following a stratified–spatial sampling design. Laboratory analyses were conducted using standard methods, and linear and quadratic regression models were developed, with performance evaluated using R², RMSE, and five-fold cross-validation. The results showed significant variation in salinity between the depressions (p < 0.001), with K2 recording the highest average (9.40 dS m⁻¹). Quadratic models outperformed linear ones, especially at high salinity levels (R² ≈ 0.995 for 1:1), with high stability in cross-validation. ANCOVA analysis revealed differences in conversion coefficients between the depressions (p < 0.05), emphasizing the need for local calibration. The study recommends using the 1:1 extract ratio as the primary operational model, with quadratic models applied in high-salinity environments.