Water and Herbicide Movement in a Ghanaian Field Soil Under Varying Irrigation Frequency.

Abstract

This study reports an investigation of the influence of irrigation frequency on water and herbicide transport in a Ghanaian field soil. An attempt was also made at simulating the movements of water and atrazine [2-chloro-4-ethylamino-6-isopropylamino-s-triazine] in the soil under varying irrigation frequency. The irrigation treatments used in the study were: Treatment 1 (Tl): 150 mm of water applied in three equal amounts of 50 mm each on days 1, 9 and 17; Treatment 2 (T2): 150 mm of water applied in five equal amounts of 30 mm each on days 1, 5, 9,13 and 17; and Treatment 3 (T3): 150 mm of water applied in nine equal amounts of 16.6 mm each on days 1, 3, 5, 7, 9, 11, 13, 15 and 17 after the commencement of the experiment. Furthermore, the differences in water storage under the different irrigation treatments were also investigated. In general, T2 resulted in highest water storage at the end of the experiment, followed by Tl and then T3. Water loss from the profile was in the reverse order. Treatment 1 caused more leaching of atrazine and the centres of mass of atrazine at the end of the experiment were in the order Tl > T2 « T3. Sorption of atrazine decreased with depth. The percentage of atrazine recovered decreased linearly with time and the amounts of the herbicide recovered at the end of the experiment were in the order T2 > Tl ~ T3. Thus treatments 1 and 3 resulted in greater loss of atrazine probably due to deep percolation resulting from the relatively higher amounts of water applied and partly due to volatilization, respectively. The simulated water profiles compared favourably with the field-measured ones especially in Tl and T3 (with R2 values of 0.754 and 0.674, respectively). In all the treatments, the predicted water contents were higher in the top 0.3 m than the observed values. However, using 1:1 plots it was noted that the simulated water content values were in good agreement with the measured ones especially at the initial stages of the experiments. The deviations of the predicted values from the observed became more apparent with subsequent water applications. The inability of the model to correctly predict well was ascribed to its simplicity, neglecting the effects of hysteresis during water redistribution and the type of simplified numerical scheme used to solve the water flow equation. The patterns of the observed and predicted solute fronts were in good agreement. Using a Chi-square test, it was found that differences between the observed and predicted leaching fronts were not significantly different at 1% level.