Estimación de la salinidad en suelos del delta del río Sinú en Colombia, mediante modelos de regresión lineal múltiple

Abstract

The classic statistics and geo-statistics are some of the analysis tools of soil saltiness that contribute to the identification of affected areas, monitoring the space-temporary variations, and the formulation of management strategies. The objective of this study was to generate and validate models for the electric conductivity estimation of the soil saturation extract (EC) on the area of influence of the Sinu River mouth in dry and rainy periods. The study area of 82 km2 was divided into four locations near the river mouth of the Sinu River, where 39 observation wells distributed in 25 wells in main vertices with spacing of 2 km apart and 14 auxiliary vertices with spacing 1 km of the main. For the sampling of groundwater two periods were chosen and corresponded to season of dry and rainy seasons. In the samples were evaluated the electrical conductivity (EC), pH, Ca, Mg, Na, K, HCO3, CO3 and SO4. The data obtained were subjected to simple correlation analysis between the EC and the cations and anions present in the saturation extract. The estimating the models on CE was performed by the method of stepwise regression (The Stepwise regression procedure) and its validation was performed using the chi-square test between EC values generated by the models and that obtained in soil samples. The results indicate a highly significant correlation between the EC and the contents of Cl, Na and Mg ions. The CE models calculated for the depths and periods have R2 between 0.84 and 0.96, indicating a high degree of reliability for the estimation of the EC. Similarly, the contribution of each ion on the models was determined, which predominantly were Cl, Na and Mg.