A Unified procedure for the probabilistic assessment and forecasting temperature characteristics under global climate change

Accurate assessment and forecasting of temperature characteristics in relation to climate change are essential for making effective climate policies. The industrial revolution is considered one of the primary causes of climate change, resulting in global warming and spatio-temporal variation in temperature around the world. This study introduces a novel, unified approach called Generalized Probabilistic Standardized Temperature Index (GPSTI) to, monitoring, forecasting and evaluate the acceleration of temperature fluctuations with consideration of climate change impact. In application, this research considered meteorological data from 41 locations across various regions of Pakistan. Additionally, different machine learning techniques that include Autoregressive Integrated Moving Average (ARIMA), TBATS, Extreme learning machine (ELM), and Artificial Neural Network – Multilayer Perceptron (MLP) are used to predict the value of the GPSTI. The results indicate that the TBATS model has been demonstrated to be the best performer among all the evaluated models by continuously achieving lower RMSE values at most of the stations (Faisalabad, 0.9307; Karachi, 0.3836; Kohat, 0.4448; Gilgit, 0.4626; and Kotli, 0.3900) during the testing stage. Outcomes associated with this research shows that the GPSTI can be used for future forecasting under various machine learning and probabilistic approach. The key advantages of GPSTI include its ability to facilitate regional comparisons and its utility for future forecasting. Overall, the computational evidence strongly supports a significant shift toward higher temperatures over time, potentially influenced by the industrial revolution and its associated factors. These results support the widely accepted scientific consensus on global warming and provide additional empirical evidence for the ongoing discussion on climate change.