Visible to the eye, now in the model: Parameterizing dual porosity water retention functions in structured soils

Abstract

Soil water retention is important for the establishment and productivity of ecosystems through its role in governing the flux, depth distribution, and availability of soil moisture. With increasing application of global and regional hydrologic and climate models, there is a concomitant need to accurately predict and map soil hydraulic properties to parameterize these models and simulate soil water dynamics across spatiotemporal scales. Soil water retention functions created to fulfill this need typically assume a unimodal pore-size distribution, despite the common observation that soil pore-size distributions are multimodal due to soil structure and interpedal macropores. Existing dual porosity functions divide pores into two categories: larger pores, controlled by structure, and smaller pores, controlled by texture. Obtaining the parameters for the structural domain is difficult due to the poor characterization of large pores. Large pores cannot be characterized from water retention curves because measurement of water retention near saturation, and CT scans of soils rely on small soil sample volumes which limits the pore characterization to tens of millimeters in range, while pores may be much larger. In this study, we developed multiple PTFs to predict the van Genuchten parameters (ɑ and n) of the structural domain in dual porosity models, as well as the w coefficient, which reflects the relative abundance of these two types of pores in the dual porosity model. Our PTFs were developed from characterized pores > 180 µm from nine pedons across Kansas, USA, using recent advances of multistripe laser triangulation (MLT) scanning applications. MLT scanning pore characterization allowed us to characterize soil pores > 10 cm and was conducted on 30-cm wide soil monoliths collected from excavation walls of each pedon that were either 20 or 40 cm tall depending on the thickness of the horizon. We used ImageJ to quantify pore-size distributions that were then used to estimate the water retention curve (WRC) and hydraulic conductivity of the structural domain. We fitted van Genuchten functions to characterize the WRCs in the structural domain, and ROSETTA 3.0 was used to characterize the WRCs in the matrix domain. These WRC fits were used to develop new PTFs that predict the parameters of the dual porosity model using mixed linear models with inputs including NRCS soil structure field descriptions along with standard physical and chemical properties (clay, sand, SOM, bulk density, coefficient of linear extensibility, cation exchange capacity, horizon midpoint depth, and quantified morphological descriptions of structural type, grade, solidity, roundness, and circularity). Using the predicted parameters, we estimated water retention for each horizon and achieved high levels of correlation and accuracy when compared with the water retention derived from the MLT scans. The approach for creating PTFs can be used to improve soil hydraulic property parameterization of soils with structure in regional hydrologic and climate models by providing a framework for integrating multiple recent advances such as the characterization of large pores using MLT and use of quantified soil structure from profile descriptions. Future studies will examine performance of these PTFs in numerical hydrologic models.