What is it about?
Saturated hydraulic conductivity (Ks) is an indispensable parameter for investigating surface/ sub-surface hydrologic processes. Pedotransfer function is a reasonable alternative to the expensive and tedious direct measurement of Ks. This study augments the pedotransfer function based Ks estimation using machine learning (ML). We evaluated different machine learning algorithms and selected the Random Forest as the learning algorithm for Ks prediction. Moreover, the prediction ability of Random Forest was enhanced through the selection of pertinent predictors for Ks. A robust pedotransfer function for the reported study was developed using the selected learning algorithm and pertinent predictors. We also evaluated the developed pedotransfer function alongside the recently published and commonly used pedotransfer functions within and outside the study region and observed superior prediction proficiency by our pedotransfer function compare to others in both cases. The developed ML-based pedotransfer function may provide an excellent cost-effective estimation of Ks.
Photo by Kevin Ku on Unsplash
Why is it important?
This is the most robust and proficient pedotransfer function among the recently published and commonly used models.
Read the Original
This page is a summary of: Toward Developing a Generalizable Pedotransfer Function for Saturated Hydraulic Conductivity Using Transfer Learning and Predictor Selector Algorithm, Water Resources Research, July 2021, American Geophysical Union (AGU), DOI: 10.1029/2020wr028862.
You can read the full text:
RF-HID-S 1.0 Pedotransfer Function and Rana Dataset
This is a deposition of dataset, software developed, input template and the user guide, which is relevant to the manuscript published in Water Resources Research (AGU): Jena, S., Mohanty, B. P., Panda, R. K., & Ramadas, M. (2021). Towards Developing a Generalizable Pedo-Transfer Function for Saturated Hydraulic Conductivity using Transfer Learning and Predictor Selector Algorithm. Water Resources Research, 57, e2020WR028862. https://doi.org/10.1029/2020WR028862
The following have contributed to this page