Abstract
Saturated hydraulic conductivity ( K s ) plays a vital role in irrigation and drainage system design. Generally, K s is estimated in the laboratory; however, it is expensive and tedious, especially in the Himalayan ranges where soil sampling is challenging due to topographical constraints. Therefore, in this study, pedotransfer functions were generated using multiple linear regression (MLR) models for the predictability of K s in a Himalayan catchment in India. Fifty soil samples were collected and divided into two groups at a 70:30 ratio. Different soil attributes derived from 70% of samples were used for MLR generation, and attributes of the remaining 30% of samples were used for model validation. Six different MLR models constituting different independent soil attributes were generated and compared statistically. The results indicate that the MLR model comprising soil texture, bulk density, particle density, soil moisture content (MC), organic carbon content and porosity results in the highest adjusted coefficient of determination ( R 2 ; 0.93 and 0.89 during model generation and validation, respectively). Additionally, it was found that the weight basis MC ranged from 14% to 29% with a median value of 24%. These results demonstrate that simple MLR models can be used as an alternative to laborious experimental setups for K s estimation. These findings can be used as guidelines for proper irrigation planning and design in mountainous catchments.