Toward Developing a Generalizable Pedotransfer Function for Saturated Hydraulic Conductivity Using Transfer Learning and Predictor Selector Algorithm

Pedotransfer functions (PTFs) are being instrumental in saturated hydraulic conductivity (K-s) estimation. Despite various advancements, the performance of existing generic K-s predicting PTFs need augmentation. This study developed a robust K-s predicting PTF using a machine learning (ML) algorithm and exhaustive data set for 324 soils with 28 properties sampled over a tropical savanna region of India. Four ML algorithms were evaluated for this purpose, and random forest (RF) outperformed all others. A substantial improvement to the prediction by RF-based PTF was achieved through predictor selection using a hybrid wrapper-embedded algorithm. The predictor selection algorithm selected eight pertinent predictors (HID-S): S, Si, C, FSF, C-u, GMD, D-60, and D-10. The mean absolute error (MAE), root mean squared error (RMSE), coefficient of determination (R-2), and Nash-Sutcliffe efficiency (NSE) obtained as average tenfold cross-validation scores for RF algorithm training with HID-S were 0.87, 1.47, 0.94, and 0.94, respectively. The developed PTF (RF-HID-S) was evaluated alongside the recently published PTFs by Araya and Ghezzehei (2019, ), within and outside the study region. In that process, it was observed that the RF-HID-S possessed superior prediction proficiency compared to the recently published and commonly used PTFs in both cases. These findings mark RF-HID-S as the most robust generalizable PTF, which may further be evaluated in different parts of the world. Moreover, looking at the performance of the eight selected predictors within and outside the study region, they can be considered for experiment design globally to make K-s estimation accurate and cost-effective.

Automated summary

Machine learning algorithms were used to develop a robust model for predicting saturated hydraulic conductivity (Ks) based on eight selected predictors, achieving accurate and cost-effective estimation in this study. The model outperformed existing PTFs both within and outside the study region, marking it as a superior and generalizable PTF for Ks estimation in various parts of the world.