Forecasting of Maximum Temperature Using ETS, ARIMA and Random Forest Models: A Case Study for Karachi, Pakistan

Climate mitigation and rise in average temperature are caused by global warming which is extreme in some regions like Pakistan. Due to this reason, accurate extreme temperature predictions are essential for managing water resources, landatmosphere interactions, and supporting agricultural activities. However, forecasting extreme temperatures remains difficult due to its nonlinear trend. In Pakistan, due to unavailability of data and less frequency of the data collection by limited number of meteorological data collection center it is difficult to estimate accurate values. In Pakistan, Karachi is the backbone of the country and supports the economy. Karachi is facing the most frequent heat waves for the last 5 years. To deal with the situation it is important to set a model which accurately predicts the maximum temperature based on historical data. Several deep learning techniques have been proposed over the last few decades to forecast maximum temperature in different countries. This study provides a comprehensive review of techniques utilized for maximum temperature forecasting. In this paper, we have utilized a maximum temperature data for Karachi from 2015 to 2023 and we will compare Autoregressive integrated moving average (ARIMA), Exponential time series (ETS), and Random Forest models for forecasting temperature, utilizing monthly average maximum temperature data from Karachi, Pakistan, spanning 2015 to 2023. The experimental findings show that the ETS model outperformed both the traditional ARIMA and Random Forest models, demonstrating a lower RMSE. The model attained an RMSE of 1.005265, reflecting a high level of accuracy within the standard range for maximum temperature forecasting. This strong error metric highlights the model's precision in predicting temperature variations, which is especially important for planning and environmental management. Forecasting maximum temperature trends is essential for enhancing resilience and quality of life in smart, sustainable cities, enabling informed decision-making and proactive urban planning