Thesis ESTUDIO Y MODELACIÓN DE LA LIBERACIÓN DE POTASIO SOLUBLE EN CUATRO SUELOS CARACTERÍSTICOS DE CHILE.
Loading...
Date
2015-01
Journal Title
Journal ISSN
Volume Title
Program
DEPARTAMENTO DE INDUSTRIAS. INGENIERÍA CIVIL INDUSTRIAL
Campus
Campus Vitacura, Santiago
Abstract
El potasio (K) es un elemento considerado nutriente para las plantas de gran relevancia
económica y productiva en el país. La liberación de K desde el suelo a la solución (Ks),
fracción del nutriente que es absorbido por las plantas, fue estudiada en 4 suelos agrícolas
característicos de Chile clasificados, a través de extracciones secuenciales. Los suelos de
distintas características físicas y químicas fueron: S1 Andisol serie Santa Bárbara con alto
contenido de materia orgánica (MO), S2 Inceptisol serie La Rosa con alto contenido de
arcilla, S3 Alfisol serie Nilahue con alto contenido de arena y S4 Alfisol serie Parral con
textura arcillosa. El estudio se llevó a cabo con el objetivo de modelar la liberación de K
a la solución, estudiar qué variables tienen mayor incidencia en la liberación y relacionar
las características del suelo con los parámetros obtenidos de los modelos de liberación.
En el estudio se observó que en los suelos S1, S3 y S4 la entrega de potasio a la solución
disminuyó hasta llegar a un valor constante en la tercera extracción, mientras que para el
suelo S2 se requirieron más extracciones para llegar a dicho punto (séptima extracción).
Esto se explicó principalmente por el nivel inicial de potasio extractable de cada suelo:
0.51, 0.49 y 0.36 cmol/kg para S1, S3 y S4, respectivamente, y 1.75 cmol/kg para S2.
El modelo que mejor ajustó a la liberación de potasio soluble medido en el tiempo fue el
ajuste exponencial con valores de 𝑅���
2 de 0.764, 0.993, 0.780 y 0.467 para los suelos S1,
S2, S3 y S4 respectivamente. La concentración de potasio acumulado, según el criterio
AIC, se ajustó de mejor manera al modelo de difusión parabólica para el suelo S1 con un
𝑅���
2 de 0.988, el modelo función potencial ajustó a los suelos S2 y S3 con 𝑅���
2 de 0.999 en
ambos casos y finalmente el suelo S4 se ajustó de mejor manera al modelo exponencial.
Por otro lado debido a la alta correlación encontrada entre el potasio soluble y la
conductividad eléctrica medida en solución (0,930) es que se utilizó el parámetro que
explica la caída de la curva exponencial de los datos de la CE para modelar el potasio
soluble, obteniéndose que los 𝑅���
2 para los suelos S1, S2, S3 y S4 fueron respectivamente
0.753, 0.973, 0.570 y 0.942.
Se generaron modelos de regresión múltiple mediante el método stepwise para estimar el
Ks, inicialmente en base a los datos de la caracterización inicial de los suelos (obteniendo
un 𝑅���
2 = 0.283) y, posteriormente, se modeló con todo el set de variables, obteniéndose
un modelo que incorporó como variables significativas a la conductividad eléctrica (CE)
inicial del suelo, CE medida en solución (extracciones), la interacción K
intercambiable*tiempo, el porcentaje de arcilla, el K extractable inicial y el tiempo, con
un 𝑅���
2 = 0.989, en donde las variables que tuvieron un mayor peso en el modelo fueron
ambas CE y el potasio extractable inicial.
Potassium release to soil solution was studied, though a sequential extraction, in four typical agricultural soils of Chile classified as S1 Andisol Santa Barbara series, high in organic matter (OM), S2 inceptisol set Rosa series with high clay content, S3 Alfisol Nilahue series with high sand content and S4 Alfisol Parral series with high clay content, all soils with different characteristics in terms of their pH, electrical conductivity, texture, total cation exchange capacity, OM and exchangeable cations concentration. The objective of the study was to model K release after sequential extractions in saturated paste. Specific objectives were to establish the best predictors to estimate Ks, determine which variables of soil characterization and measured in solution affect potassium concentration in soil solution in order to model hat pool, and finally establish a new extraction method. It was found that in S1, S3, and S4 potassium release to the solution decreased in time until becoming constant in the third extraction, while soil S2 reached the constant release in the seventh extraction, which is mainly explained by the initial level of exchangeable potassium of each soil: for S1, S3 and S4 were 0.51, 0.49, and 0.36 Cmol /kg, respectively while for S2 it was 1.75 Cmol /kg. The model that was the best fit of K release was exponential fit with 𝑅�� 2values of 0.764, 0.993, 0.780 y 0.467 for S1, S2, S3 and S4, respectively. Relative to the accumulated K concentration, according AIC criterion, in the soil S1 the best fit was the parabolic diffusion model with a 𝑅�� 2 0.988, for the soils S2 and S3 the power function model was the best fit with a 𝑅�� 2 0,999, in both cases, and finally for the soil S4 the best fit was the exponential model with a 𝑅�� 2 0.999. On the other hand due to the high correlation found between the soluble potassium and electrical conductivity measured in solution (0.93), the latter was chosen to predict the K release using the parameter of the exponential curve from the EC fit. The 𝑅�� 2 for S1, S2, S3 and S4 were respectively 0.753, 0.973, 0.570 and 0.942. The modeling for K release was performed using stepwise multiple regression. Using only by the soil characteristics the best model showed a 𝑅�� 2 = 0.283, subsequently K release was modeled with the entire set of variables. The selected variables by StepWise were soil EC and soil solution EC, Ke*time interaction, clay content, the initial exchangeable K and time, with 𝑅�� 2 = 0.989. The variables that had a greater weight in the model were the electrical conductivitys and initial exchangeable potassium.
Potassium release to soil solution was studied, though a sequential extraction, in four typical agricultural soils of Chile classified as S1 Andisol Santa Barbara series, high in organic matter (OM), S2 inceptisol set Rosa series with high clay content, S3 Alfisol Nilahue series with high sand content and S4 Alfisol Parral series with high clay content, all soils with different characteristics in terms of their pH, electrical conductivity, texture, total cation exchange capacity, OM and exchangeable cations concentration. The objective of the study was to model K release after sequential extractions in saturated paste. Specific objectives were to establish the best predictors to estimate Ks, determine which variables of soil characterization and measured in solution affect potassium concentration in soil solution in order to model hat pool, and finally establish a new extraction method. It was found that in S1, S3, and S4 potassium release to the solution decreased in time until becoming constant in the third extraction, while soil S2 reached the constant release in the seventh extraction, which is mainly explained by the initial level of exchangeable potassium of each soil: for S1, S3 and S4 were 0.51, 0.49, and 0.36 Cmol /kg, respectively while for S2 it was 1.75 Cmol /kg. The model that was the best fit of K release was exponential fit with 𝑅�� 2values of 0.764, 0.993, 0.780 y 0.467 for S1, S2, S3 and S4, respectively. Relative to the accumulated K concentration, according AIC criterion, in the soil S1 the best fit was the parabolic diffusion model with a 𝑅�� 2 0.988, for the soils S2 and S3 the power function model was the best fit with a 𝑅�� 2 0,999, in both cases, and finally for the soil S4 the best fit was the exponential model with a 𝑅�� 2 0.999. On the other hand due to the high correlation found between the soluble potassium and electrical conductivity measured in solution (0.93), the latter was chosen to predict the K release using the parameter of the exponential curve from the EC fit. The 𝑅�� 2 for S1, S2, S3 and S4 were respectively 0.753, 0.973, 0.570 and 0.942. The modeling for K release was performed using stepwise multiple regression. Using only by the soil characteristics the best model showed a 𝑅�� 2 = 0.283, subsequently K release was modeled with the entire set of variables. The selected variables by StepWise were soil EC and soil solution EC, Ke*time interaction, clay content, the initial exchangeable K and time, with 𝑅�� 2 = 0.989. The variables that had a greater weight in the model were the electrical conductivitys and initial exchangeable potassium.
Description
Keywords
POTASIO, PLANTAS, MÉTODO STEPWISE