Tesis
Análisis geoestadístico de datos funcionales de temperatura del aire en la provincia de Chimborazo
Fecha
2020-02-10Registro en:
Checa Gamarra, Marisol Carolina. (2020). Análisis geoestadístico de datos funcionales de temperatura del aire en la provincia de Chimborazo. Escuela Superior Politécnica de Chimborazo. Riobamba.
Autor
Checa Gamarra, Marisol Carolina
Resumen
The present titling work had as aim to estimate the air temperature in unsampled sites of the Chimborazo province, period 2014-2017 through the geostatistics analysis of functional data considering the 11 weather stations that the GEAA monitors. The daily, hourly and annual temperatures per day, softened by B-Splines Cubic and Fourier, with 15 and 365 bases according to the min.basis( ) command of R and a residual variance of 0.238 and 0.047 respectively, were considered as functional data. To determine the behavior of the temperature, functions were identified: mean, standard deviation and atypical (were separated from the analysis) of each of the stations. In order to define the functional data for geostatistical modeling, a FANOVA was carried out, both for the average curves per hour and per day; the null hypothesis was not rejected for the day, for this reason for the spatial modeling the average temperature of the years under study was taken. By means of cross-validation, a smaller sum of residues was obtained with the spherical model for estimating with ordinary kriging functional (OKFD), even if the more sample data the adjustment is available, the better, which is why 29 points were systematically generated, of which 4 allowed to improve the model, so it was defined with 15 georeferenced points, with a standard deviation of 3102.62 degrees Celsius. The temperature in four quinoa cultivation zones was estimated: Amulá Casaloma, Majipamba, San Pedro de Yacupamba and Columbe Grande, whose results were compared with the temperatures downloaded from NASA, obtaining sums of error squares of 355.13, 1878.12, 1465.88 and 765.05 respectively. It is recommended to apply the same methodology for the analysis of other methodological variables.