TITLE:
Estimating Soil Hydraulic Parameters Characterizing Rainwater Infiltration and Runoff Properties of Dryland Floodplains
AUTHORS:
Sabelo S. W. Mavimbela, Leon D. van Rensburg
KEYWORDS:
Floodplain Soil Types, Hydrus-1D Model, Inverse Parameter Optimisation, Objective Function, Parameter Sensitivity, Sedimentary Crust, Rainfall Simulation
JOURNAL NAME:
Computational Water, Energy, and Environmental Engineering,
Vol.8 No.1,
January
29,
2019
ABSTRACT:
The two-layered (0 - 50 and 50 - 250 mm)
surface horizon hydraulic parameters of three dryland floodplain soil-types
under aquafer water management in Postmasburg, Northern Cape Province of South
Africa were estimated with HYDRUS-1D model. Time dependent water infiltration
measurements at 30 and 230 mm depths from simulated rainfalls on undisturbed 1
m2 small plots with
intensities of 1.61 (high), 0.52 (medium) and 0.27 (low) mm·min-1,
were minimised using a two-step inversion. Firstly, separate
optimisation of the van Genuchten-Mualem model parameters for the two
surface-horizon layers and secondly, simultaneous
optimisation for the joint two-layered horizon with first step optimal
parameters entered as initial values. The model reproduced transient
water-infiltration data very well with the Nash-Sutcliffe model efficiency
coefficient (NSE) of 0.99 and overestimated runoff (NSE; 0.27 to
0.98). The upper surface horizon had highly optimised and variable parameters
especially θs and Ks. Optimal Ks values from higher soil surface bulk-density (≥1.69 g·cm-3)
were lower by at least one order of magnitude to double ring infiltrometers and water infiltration properties were different (P α and n parameter values corresponded
well with texture of the Addo (Greysols), Augrabies (Ferralsols) and Brandvlei (Cambisols)
soil types. However, θs and Ksshowed greater sensitivity to
model output and exerted greater
influence on dryland floodplain water-infiltration and runoff characteristics.
Increasing rainfall simulation period to attain near-surface saturated
conditions and inclusion of surface ponding data in the inverse problem could
considerable improve model prediction of hydro-physical parameters controlling
surface-subsurface water distribution in fluvial environments.