Nonlinear dynamic analysis of meteorological variables for Ha'il region, Saudi Arabia, for the period 1990-2022

Abstract

The study applies diverse methods of chaos detection to meteorological variable data (air temperature, relative humidity, surface pressure, precipitation, and wind speed for Ha'il, Saudi Arabia) to understand the nonlinear dynamics and to classify their nature. Additionally, Random Forest Algorithm model is used to predict the precipitation and wind speed. The results obtained by classical and modern approaches are compared. All the variables are found to be chaotic based on correlation dimension, approximate entropy, and 0-1 test. The chaos decision tree algorithm diagnoses air temperature, relative humidity, and wind speed as chaotic, while precipitation and surface pressure are identified as stochastic. This shows that the classical methods are well-validated with the modern methods. Nevertheless, some of them contradict modern methods. The analysis for 32 years of data showed no precipitation for 92% of the time during the entire period based on the Random Forest algorithm.

Keywords

• nonlinear dynamic analysis
• chaos theory
• climate variables
• time series
• chaos detection
• recurrence analysis, random forest algorithm