Identifying the dominant controls on macropore flow velocity in soils: A meta-analysis

Abstract

Macropore flow is a ubiquitous hydrologic process that has not been well explained using traditional hydrologic theories. In particular, macropore flow velocity (MFV) is poorly understood with respect to its typical ranges and controlling factors. Here we conducted a meta-analysis based on an MFV dataset compiled from 243 measurements documented in 76 journal articles. The dataset includes MFV values measured using different approaches across the soil-core, field-profile, and trench scales. Our analyses show that MFV has a geometric mean of 1.08×10-3ms-1, which is about 2~3 orders of magnitude larger than the corresponding values of saturated hydraulic conductivity in the soil matrix. Using machine learning methods including classification and regression tree and random forests algorithms, we identified observation scale, travel distance, rainfall intensity and macropore diameter as the most important factors that control MFV. MFV is much larger at the trench scale than at the other two scales mainly due to abundant large macropores. Correlation analysis and multivariate regression revealed that (1) MFV and rainfall intensity have significant positive correlation, which indicates that MFV is a dynamic variable; and (2) MFV and macropore diameter also have strong positive correlation at the trench scale, which indicates macropore size as a key controlling factor. Using macropore diameter and rainfall intensity as explanatory factors, MFV can be well predicted (R2 = 0.76) by a multivariate regression equation at trench-scale, implying that rainfall intensity can be considered a proxy for the filling degree of macropores. Furthermore, both the Poiseuille and Manning equations were found to overestimate the MFV values, suggesting a parameter representing the connectivity of the macropore network is needed for providing reasonable estimates of MFV using physically-based equations.