4.2 Agricultural Return Flow Salinity as a Function of Irrigation Loading and Soil Type
Agricultural return flow salinity was modelled using 1-D reactive transport with an evaporative loading applied to the top 3 cells as described earlier. Arabian Gulf seawater was selected as the irrigation feed water and model layers were populated with mineral composition using site-specific information from the Abu Dhabi Soil Survey. AD134 was a strongly gypsic soil (Table 3(a) ) with high gypsum occurrence in all cells or layers, whereas AD146 (Table 3(b) ) was a carbonate-rich soil with almost no gypsum presence. Finally, soil type AD157 (Table 3(c) ) was a mildly gypsic soil with high gypsum content only in cells 3 and 4 from the ground surface. Three different annual irrigation water loadings, 2.5 m, 3m, and 4 m, were applied to each soil column model corresponding to leachate fractions of 0.25, 0.50, and 1.0, respectively.
The results from the reactive transport modelling for the three soil types after one calendar year of continuous flow are presented inFigure 5 . Increased irrigation loading was found to generally correlate with a decrease in return water salinity. This result appears to corroborate the “salt penalty” effect described by Edward P. Glenn, J. Jed Brown, and Eduardo Blumwald (1999), which essentially attributes this effect to a combination of lower dissolution of minerals from the soil at faster flowrates and an averaging of the mineral dissolution over the larger volumetric flow. The highest return water salinity (i.e., 56,500 ppm TDS and higher) was observed for the high gypsum containing AD134 soil type using the B-dot activity coefficient model. The discrepancy in salinity prediction using the two different models was most pronounced for this soil type, with PHRQPITZ predicting a salinity in the range of only 53,500 - 54,000 ppm TDS. Meanwhile soil types AD146 and AD157, the carbonate-rich and gypsum-deficient soils, displayed a close match between the two different activity coefficient models employed. These results suggest that unlike the PHRQPITZ model, the B-dot activity coefficient model appears to estimate a much higher solubility for gypsum at higher ionic strengths. This result would require field verification using soil core analysis during initial pilot studies.