Accurate prediction of the leaching requirements (Lr) of crops and striving to attain them is essential for efficient irrigation water use. Solute modeling was extended to develop four Lr conceptual models that do not neglect solute reactions in the root-zone, surface evaporation, and the influence of immobile wetted pore space. The models were based on: (i) the water movement equation which included an exponential water-uptake function (-e) or the 40-30-20-10 water-uptake function (-4); (ii) the solute movement equation for a reactive salt of a linear reaction term (the Lrchem-e and Lrchem-4 models); or the employment of output (salinity of soil solution, EC vs concentration factor, CF) of the SAO comprehensive chemical model (the LrSAO-e and LrSAO-4 models); and (iii) the inclusion of an effective soil solution volume in the transport equations. The root-zone average relative effective soil solution volume νeff (L | L50, p) was of sigmoidal response to leaching fraction (L) with two adjustable parameters L50 and p; the root-zone average reduced retention coefficient decreased linearly with L; and salt concentration at soil surface was related to salt concentration of irrigation water (ECi) by the fraction of irrigation water that evaporated (∈). The resulted concentration profiles indicated the salt behaved as a conservative one down to a threshold depth (xs) below of which salt was retained and precipitated. The depth of the conservative-salt front, xs increased with L and the 40-30-20-10 water-uptake pattern overestimated the xs depth relative to the exponential pattern. Concentration profiles were integrated to compute the root-zone average salinity, which was converted to crop salt-tolerance threshold (AE). The four conceptual models were successfully calibrated using experimental AE/ECi vs. Lr data with the input parameter values: ς = 0.27, p = 1.44, L50 = 0.16, ω = 2, and ∈ = 0 or 0.1 for the exponential or the 40-30-20-10 pattern, respectively; where ς is relative root length parameter and ω is a weighing parameter. No significant difference existed between the four model correlations at the 0.05 level. The four models require ECi and AE of the crop as input for Lr prediction. Sensitivity analysis revealed predicted Lr was sensitive the least to error in ∈. For tolerant and moderately tolerant crops Lr was sensitive the most to ς, and for sensitive crops to L50 and p. Model verification and validation were discussed. In deriving the present Lr models, no osmotic adjustment was required and both the exponential and the 40-30-20-10 water uptake patterns were, equivalently, applicable.
Read full abstract