Statistical Analysis of weather variables of Antofagasta

TL;DR

This paper analyzes the statistical behavior of Antofagasta's weather variables (temperature, pressure, relative humidity) using daily measurements at three times of day. It employs deseasonalization, Q-Q plots, skewness and kurtosis, and Pearson correlation to characterize distribution shapes and autocorrelation. Findings show near-symmetric distributions with heavy tails (positive kurtosis) and long memory, particularly for pressure and temperature, with autocorrelation persisting up to about a year. The results highlight pronounced annual-season effects and substantial long-range dependence, informing climate interpretation and statistical modeling in coastal arid environments and related ENSO dynamics.

Abstract

The statistical behavior of weather variables of Antofagasta is described, especially the daily data of air as temperature, pressure and relative humidity measured at 08:00, 14:00 and 20:00. In this article, we use a time series deseasonalization technique, Q-Q plot, skewness, kurtosis and the Pearson correlation coefficient. We found that the distributions of the records are symmetrical and have positive kurtosis, so they have heavy tails. In addition, the variables are highly autocorrelated, extending up to one year in the case of pressure and temperature.

Figures (5)

• Figure 1: Histograms with normal fit and normal Q-Q plot of logistic distribution (top), normal distribution (center) and uniform distribution (bottom).
• Figure 2: Pressure (top), temperature (centro) and relative humidity (bottom) measured at 08:00.
• Figure 3: Residuals of pressure (top), temperature (centro) and relative humidity (bottom) measured at 08:00.
• Figure 4: Probability distribution and Q-Q plot of pressure (top), temperature (center) and relative humidity (bottom), measured at 08:00.
• Figure 5: Correlogram of pressure (top), temperature (center) and relative humidity (bottom).