Seasonal Variability and Transition Dynamics of Temperature and Windspeed Regimes in Southwestern Nigeria

Abstract

This study investigated the seasonal variability and transition dynamics of temperature and windspeed regimes in South-west, Nigeria using Fuzzy Markov Model. Daily Temperature and Windspeed data for the South-west States (Lagos, Osun, Ogun, Ondo, Ekiti and Oyo) in Nigeria from 1991 to 2020 (30 years) was obtained from National Aeronautic and Space Administration (NASA) meteorological center. Atmospheric variables often exhibit uncertainty, gradual transitions, and overlapping climatic states that cannot be adequately represented using conventional deterministic classification methods. To address this limitation, fuzzy logic was employed to quantify the degree of belonging of temperature and windspeed conditions to different atmospheric regimes, while transition structures were analysed using fuzzy transition matrices. The climatic variables were categorized into Low/Moderate/High windspeed regimes and Cool/Normal/Hot temperature regimes. The Markov Chain Model result reveals that the high diagonal temperature and windspeed transition probabilities (0.96–0.99) across the South-west regions are highly persistent, with limited switching between regimes while the Fuzzy Model transition analysis (0.77–0.88) of temperature and windspeed across study area reveals strong overlapping atmospheric behaviour during the dry and wet seasons. This implies that climatic conditions do not exist as rigid or isolated states but instead possess varying degrees of belonging to multiple regimes simultaneously. The total elimination of moderate regimes in the Fuzzy transition temperature and windspeed matrices, where the moderate rows had zero membership values, was one of the study's key conclusions. This implies that the South-west climate system has weak, unstable, or transient intermediate atmospheric conditions. The fuzzy logic captured gradual regime transitions and overlapping state memberships that were not evident in the classical Markov framework.