Evaluation of Aquacrop Simulation of Early Season Evaporation and Water Flux in a Semiarid Environment.

The AquaCrop model of crop growth, water use, yield and water use efficiency (WUE) is intended for use by extension personnel, farm and irrigation managers, planners and other less advanced users of simulation models in irrigation planning and scheduling. It could be useful in estimating changes in WUE due to limited or deficit irrigation, which is a potentially important practice for producing more crop revenue/yield per unit of water applied. AquaCrop simulated yield has been evaluated for various crops, but seldom has simulated evapotranspiration (ET) been evaluated using ET values derived from precision large weighing lysimeters. An important component of seasonal water use in semi-arid and arid irrigated agriculture is the evaporation (E) associated with necessary pre-plant and early-season irrigations. We parameterized and evaluated AquaCrop version 4.0 simulations of E during the pre-plant and post-plant irrigation period until 25 and 43 days after planting when transpiration (T) was a negligible component of ET so that simulations were of E. Using measured values of parameters for the maximum bare, wet, unshaded soil crop coefficient, K_ex = 1.15, and readily evaporable water, REW = 13 mm, AquaCrop overestimated E by up to 53 mm in the period from beginning of pre-plant irrigation to 43 days after planting, only correctly simulating E for K_ex = 1.05 and an unrealistically small REW = 3 mm. Using the physically measured values of K_ex and REW, the model simulated daily ET to occur at the reference ET rate, ETo, five times during this period, and it exceeded 90% of ETo another three times, despite the fact that the top layer of soil was simulated to be unsaturated, and at less than field capacity. The model failed to simulate water movement from the top 10-cm thick soil layer downward into the next lower layer, even when 129 mm of water was applied through irrigation and precipitation, instead simulating 151 mm of E. Varying the value of saturated hydraulic conductivity over a range of 66 to 276 mm d^-1 had no effect on simulated E. Furthermore, the model did not simulate downward capillary flow from the top layer to the next layer below. The overestimation of early season E caused the simulated soil water availability later in the season to be insufficient to support simulations of plant growth to match field observations.