Statistical analysis of the best GIS interpolation method for bearing capacity estimation in An-Najaf City, Iraq

• Published in: Environmental earth sciences 2021-10, Vol.80 (20), Article 683
• Main Authors: Al-Mamoori, Sohaib Kareem, Al-Maliki, Laheab A., Al-Sulttani, Ahmed Hashem, El-Tawil, Khaled, Al-Ansari, Nadhir
• Format: Article
• Language: English
• Subjects: Backward elimination regression; Bearing capacity; Biogeosciences; Boreholes; Cross-validation; Datasets; Dynamic loads; Earth and Environmental Science; Earth Sciences; Environmental Science and Engineering; Field investigations; Geochemistry; Geographic information systems; Geographical information systems; Geology; Geostatistics; Geoteknik; GIS-based interpolation; Hydrology/Water Resources; Interpolation; Interpolation methods; Inverse distance weighting; Mechanical loading; Methods; Original Article; Production methods; Root-mean-square errors; Soil; Soil bearing capacity; Soil investigations; Soil Mechanics; Statistical analysis; Statistical methods; Statistics; Terrestrial Pollution
• ISSN: 1866-6280; 1866-6299; 1866-6299
• DOI: 10.1007/s12665-021-09971-2

The presence of an economical solution to predict soil behaviour is essential for new construction areas. This paper aims to investigate the ultimate interpolation method for predicting the soil bearing capacity of An-Najaf city-Iraq based on field investigation information. Firstly, the engineering bearing capacity was calculated based on the in-site N-SPT values using dynamic loading for 464 boreholes with depths of 0–2 m, using the Meyerhof formula. The data then were classified and imported to the GIS program to apply the interpolation methods. Four deterministic and two geostatistical interpolation methods were applied to produce six bearing capacity maps. The statistical analyses were performed using two methods: the common cross-validation method by the coefficient of determination ( R 2 ) and root mean square error (RMSE), where the results showed that ordinary kriging (OK) is the ultimate method with the least RMSE and highest R 2 . These results were confusing so, the backward elimination regression (BER) procedure was applied to gain the definite result. The results of BER show that among all the deterministic methods, the IDW is the optimal and most significant interpolation method. The result of geostatistical methods shows that EBK is the best method in our case than the OK method. BER also applied to all six methods and shows that IDW is the ultimate significant method. The results indicate no general ultimate interpolation method for all cases and datasets type; therefore, the statistical analyses must be performed for each case and dataset.