HYDRUS-2D simulations of water and potassium movement in drip irrigated tropical soil container cultivated with sugarcane

Abstract

Subsurface drip irrigation is increasingly being used for sugarcane production in Southeast Brazil in attempts to conserve water, especially after extreme droughts in 2014 and 2015. Correct design of a subsurface drip system is essential to provide information about water and solutes dynamics in the subsurface, leading to possible water savings. While field investigations demand much time and financial resources, numerical models can also provide much information about expected water and solute distributions in the soil profile. The objective of this study was to use HYDRUS-2D for simulations of water and potassium movement in containers packed with a subsurface drip-irrigated tropical soil cultivated with sugarcane and comparing simulated and observed data through statistical parameters. Investigations were carried out in a greenhouse during 240 days after the tillering stage. Spatial and temporal distributions of water content and potassium concentrations were collected using Time Domain Reflectometry (TDR)to calibrate and validate HYDRUS-2D. Simulations were carried out using hydraulic properties fitted directly to measured retention date, as well as estimated using pedotransfer functions (Rosetta Lite and the class pedotransfer of Carsel and Parrish). RMSE and MAE indicated a favorable capacity of the model to simulate soil water content using with measured properties and a better performance than estimated properties by pedotransfer functions. For the potassium concentrations, a less than optimal performance was obtained with HYDRUS-2D using both the measured hydraulic properties and the pedotransfer functions. A better understanding and accurate determination of solute concentrations in soil, uptake parameters, the crop root growth and distribution should be necessary for improve the simulations and observed data. The inability of HYDRUS-2D to accurately estimate potassium concentrations was not necessarily a limitation of model, but rather a limitation of knowledge about the soil and plant parameters involved.