A statistical approach to estimating runoff in center pivot irrigation with crust conditions

There have been several proposals to evaluate potential runoff in center pivot irrigation, through the integration of time varying infiltration -precipitation rate curves, involving complex iterative procedures. Some methods use empirical infiltration functions, such as the Kostiakov equation. Others use physically based infiltration functions, such as the Green-Ampt equation. Another option is to use the Richards equation, describing the one-dimensional vertical infiltration of water into the soil for a specified irrigation event. This equation is generally accepted to provide a basis for comparison between other runoff estimation methods. [P.B. Luz, J.C. Martins, M.C. Gonçalves, Reliable estimate of runoff in center pivot irrigation: statistical approach, in: Proceedings of the 16 Congress Mondial de Science du Sol, Poster 2-658, iSSS, Montpellier, France, August 19-25, 1998, pp. 577-593], developed a conceptual method of statistical nature, to estimate potential runoff in center pivot irrigation, comprising regression equations built with runoff results from a simulation computer model using the Richards equation. The procedure to simulate runoff involved a wide set of data related to water retention parameters and soil texture [W.J. Rawls, D.L. Brakensiek, Estimation of soil water retention and hydraulic properties, in: H.J. Morel -Seytoux (Ed.), Unsaturated Flow in Hydrologic Modeling, Theory and Practice, Kluwer Academic Publishers, Dordrecht, 1989, pp. 275-300], and water application. Such regression equations present a dependent variable, defined as an index of four parameters, related to the center pivot irrigation and to the soil-water system evaluation. The method had unacceptable results when a crust developed on the soil surface. Therefore, the objective of this study was to redefine the index, establishing new parameter coefficients with a trial and error approach. The model efficiency (similar to the coefficient of determination, r2) ranged from 90 to 98%, showing the results are in good agreement to those computed by Richards equation, exhibiting a strong predictive value.