Field-testing of models which identify soils susceptible to salinity development

• Published in: Geoderma 1989-10, Vol.45 (1), p.31-64
• Main Authors: Corwin, D.L., Sorensen, M., Rhoades, J.D.
• Format: Article
• Language: English
• Subjects: Earth sciences; Earth, ocean, space; ESTADOS DEL OESTE (EUA); ETATS DE L'OUEST (EU); Exact sciences and technology; LEACHING; LESSIVAGE DU SOL; LIXIVIACION; MODELE; MODELOS; PERMEABILIDAD; PERMEABILITE; PERMEABILITY; PROPIEDADES FISICO-QUIMICAS (SUELO); PROPRIETE PHYSICO-CHIMIQUE DU SOL; SALINIDAD; SALINITE; SALINITY; SOIL CHEMICOPHYSICAL PROPERTIES; SOILS; SOLS; SUELOS; Surficial geology; WESTERN STATES (USA)
• ISSN: 0016-7061; 1872-6259
• DOI: 10.1016/0016-7061(89)90055-4

The use of a geographic information system (GIS) and automated cartographic techniques in conjunction with multiple linear regression models resulted in final map products, that accurately delineated areas having low, medium and high potentials for the development of salinity. The vector representations of the spatial data which delineated the three salinization potentials are the result of a weighted overlay of each of four soil factors believed to influence soil salinity development on irrigated soils of the arid western United States. Three multiple regression models were formulated to determine salinization potentials from four soil salinization factors: soil permeability, depth to the groundwater, groundwater quality and leaching fraction. Each model differed in its functional form in order to define a different interrelationship between the soil salinization factors. The regression models were developed from a statistical analysis of soil salinity data for a study area of 170 square miles within the Wellton-Mohawk Irrigation District of Arizona. Data for the four regressor variables were compiled for the time period 1968–1973. The best fit to the estimation data set and best predictive quality to the validation data set were for the model which indicated that salinity development in the top 61 cm (24 inches) of the soil profile is directly related to the electrical conductivity of the groundwater, inversely related to the soil permeability and inversely related to the leaching fraction, all in a linear manner: 18.03 × (1/(soil permeability in cm/h)−0.40 × (1/(leaching fraction)) +1.32 × (electrical conductivity of the groundwater in dS/m) +2.12 = final summation value; the final summation value is classified into a low, medium or high salinization potential category. A close analysis of the three models showed that irrigation management is probably the most dominant factor in the development of salinity in the Wellton-Mohawk Irrigation District for the time period studied, 1968–1973. Predicted salinization potentials for all three models were verified against measured salinities taken from a salinity traverse conducted in November of 1973. Verification results indicated a reasonably high agreement between predicted soil salinization potentials and potentials derived from measured field salinities. Approximately 86% of the salinization potentials were correctly forecast through the use of the best of the three regression models. A regression analysis