Temporal variation in soil physical properties improves the water dynamics modeling in a conventionally-tilled soil

Temporal variations in soil physical properties are rarely recorded in field experiments or considered when modeling water and solute dynamics in agricultural soils. This study aimed at (a) quantifying the temporal variations in soil physical properties, such as the saturated hydraulic conductivity (Ks), bulk density (ρb) and soil water retention during the growing season of an irrigated maize crop conventionally tilled with a moldboard plow, and (b) modeling the observed water dynamics. For modeling, the effect of temporal variations of soil properties was explored and compared to results with constant values of soil properties during the simulation period and with results after an optimization of soil parameters by inverse modeling. Field and laboratory experiments were performed to measure the soil physical properties of five soil compartments (defined according to the position relative to crop row and the depth) at three dates during the maize season (sowing, flowering and just before harvest). During the maize season, ρb values ranged from 1.21 to 1.56gcm−3 and increased with time (by 15–25% of the initial value). Ks values, ranging from 2.9 to 56.3mmh−1, significantly decreased with time (by a factor of 3 to 6) according to the soil compartment, and were negatively correlated with ρb. In the first step to model water dynamics, the initial values of soil physical properties (measured at maize sowing) were used as constant input parameters for the model HYDRUS-2D during the maize season. This simulation led to a poor description of soil water potentials and water content dynamics, without any drainage at 100cm depth during the maize season. After an optimization of soil physical parameters, the description of the water dynamics was significantly improved, but optimized parameters, especially Ks and θs, were not within the range of field measurements. In a last modeling step, the simulation period was divided into three periods with a specific parameterization of soil physical properties for each period. The description of the water dynamics was improved compared to the simulation with constant values for soil physical properties. Such results argue for taking into account time-variable soil physical properties in modeling to correctly assess the water and solute dynamics in soils.