Interpoladores determinísticos espacio-temporales, series de tiempo y análisis de datos funcionales para el estudio y la predicción de la precipitación en Cundinamarca y Bogotá D.C.

Abstract

In this research was used a methodology for the study and prediction of the precipitation in the department of Cundinamarca and the city of Bogota DC, Colombia, from 133 records of meteorological stations of the IDEAM (Institute of Hydrology, Meteorology and Environmental Studies), between january 2010 and december 2016 with a monthly frequency, it was used spatial-temporal deterministic interpolators, time series and functional data analysis, as an alternative to the most common numerical and geostatistical models, that, although they are more robust, they present limitations in terms of scale, complexity and prediction intervals. Space-time predictions are then obtained for the years 2017, 2018, 2019 and 2020 (48 months) on a local and / or regional scale, with a good level of detail and low complexity, also incorporating time series, as a complement to the interpolation process, thus generating a new competitive method with respect to the use of deterministic interpolators only, because, although it generates a higher computational cost, precipitation predictions with a lower prediction error are obtained that will be able to use as a fundamental input in planning, territorial ordering and risk management and prevention.