Root water uptake as simulated by three water flow models

The objective of our study was to determine to what extent four root water uptake (RWU) models differed when tested under extreme conditions. Cumulative transpiration patterns were similar, contrary to the spatial extraction patterns. The analysis showed that both soil physical and root physiological factors are important for proper deterministic modeling of RWU. We compared four root water uptake (RWU) models of different complexity that are all embedded in greater soil water flow models. The soil models used were SWAP (one-dimensional), FUSSIM2 (two-dimensional), and RSWMS (three-dimensional). Within SWAP, two RWU functions were utilized (SWAP-macro and SWAP-micro). The complexity of the processes considered in RWU increases from SWAP-macro, to SWAP-micro, to FUSSIM2, to RSWMS. The objective of our study was to determine to what extent the RWU models differed when tested under extreme conditions: low root length density, high transpiration rate, and low water content. Comparison 1 looked at the results of the models for a scenario of transpiration and uptake and Comparison 2 studied compensation mechanisms of water uptake. The uptake scenario pertained to a long dry period with constant transpiration and a single rainfall event. As could be expected, the models yielded different results in Comparison 1, but the differences in cumulative transpiration were modest due to various feedback mechanisms. In Comparison 2, the cumulative effect of different feedback processes were studied. Redistribution of water due to soil pressure head gradients generated by water uptake led to an increase in cumulative transpiration of 32%, and the inclusion of compensation in uptake by the roots resulted in a further increase of 10%. Going from one- to three-dimensional modeling, the horizontal gradients in the soil and root system increased, which reduced the actual transpiration. The analysis showed that both soil physical and root physiological factors are important for proper deterministic modeling of RW