Resumen de Estudio de series climatológicas utilizando análisis discriminante lineal. Caso: Cuenca Río Guanare, Venezuela.
Yary Pérez Pérez, César Osorio Carrera, Pedro Córdova Mendoza, Teresita Barrios, Iris Córdoba, Aldo Delgado Acevedo, Tomás Rodriguez, Juan De La Cruz Lozado, Luis Ramírez
- español
El objetivo de esta investigación fue comparar series de precipitación mensual mediante el análisis discriminante con el fin de estudiar la tendencia y estacionalidad de la precipitación asociada a la cuenca del Rio Guanare en Venezuela. Se utilizaron datos de la estación meteorológica Aeropuerto-Guanare y series generadas mediante simulación de Montecarlo con base en los modelos Pearson III, Gumbel I, Log-normal, Log-Pearson III y un modelo lineal con dos armónicos y errores autorregresivos AR(1).
- English
The objective of this investigation was to compare monthly precipitation series by means of the discriminant analysis in order to study the trend and seasonality of the precipitation associated with the Guanare River basin in Venezuela. Data from the Aeropuerto-Guanare meteorological station and series generated by Montecarlo simulation based on the models were used; Pearson III, Gumbel I, Log-normal, Log-Pearson III and a linear model with two harmonics and AR(1) autoregressive errors. Outliers were detected using the Mahalanobis distance (D2 ). The parameters of the models were estimated using maximum likelihood, and for the linear model with harmonics, they were estimated by generalized least squares. The precipitation showed an annual cycle with a maximum in June, a high rainfall in May and July. The series presented AR(1) autoregressive residuals, not perfect symmetry without long tails, normal distribution, seasonality around a median of 120.5 mm, typical characteristics of tropical dry forest. Precipitation was fitted to a Pearson III model. An influence of the total sample size (12n) was observed on the discriminant analysis, therefore a more exhaustive study of the discriminant analysis was recommended for the classification of monthly precipitation series.
