The problem of too many infiltration models: Balancing infiltration model selection and physical meaning of soil hydraulic parameters

Abstract

The crucial impact of infiltration process on many related physical and biogeochemical processes has motivated researchers to develop numerous models for assessing infiltration over the past two centuries. This posed a real challenge for researchers to decide which model to utilize when analyzing experimental infiltration data and created numerous theoretical paths for extracting soil hydraulic properties, such as saturated hydraulic conductivity $K_s$ (LT⁻¹) and sorptivity $S$ (LT^−0.5). Here, an uncertainty meta-analysis was carried out using a global infiltration database of 5023 cumulative infiltration curves. This analysis focused on assessing the variability of the estimated infiltration characteristics, $K_s$ and $S$ from eleven one-dimensional (1D) infiltration models and different extraction techniques. Results revealed notable variations in $K_s$ and $S$ derived from various models under varying extraction techniques leading to differences spanning one or two orders of magnitude in many cases for the same experimental dataset. The observed changes in $K_s$ and $S$ indicate that the characteristics related to the models and extraction methods have a significant impact on the estimation of soil hydraulic parameters. These insights are crucial for practical applications in fields such as agriculture and hydrology. They also highlight the need to standardize a more physically grounded approach that could use a consistent reference (such as the steady-state slope of cumulative infiltration curve for $K_s$) in parameter optimization to improve the reliability in estimating $K_s$ and $S$ parameters. This is especially important for large-scale studies and meta-analysis of those parameters especially when involving data from different studies, leading to potentially large uncertainties.