Comparative analysis of daily maximum temperature & heatwave using MLP and LSTM

Heat waves have become a recurrent issue in many Indian cities, causing adverse effects on human health and infrastructure. Through this study, we compare the performance of two deep learning models for heat wave prediction in Indian cities: a multi-layer perceptron (MLP) and a long short-term memory (LSTM) network. We use historical weather data from several Indian cities to train and evaluate the models, concentrating on the ability to forecast heat waves with various lead times. In our experiments, both models have been shown to be highly accurate when predicting heat waves in India’s cities and LSTM has generally outpaced MLP as a predicted power. This is due to the complementary strengths both models have: MLP has a stronger ability to capture temporal patterns in data, while LSTM can be more efficient at modelling longer-term dependencies. Our findings show that it is possible to predict thermal waves in Indian cities by applying advanced learning methodologies, e.g. LSTM, which could be an effective method of providing warning systems on health and safety issues. Hopefully, this study will lead to more rigorous and efficient methods for predicting and managing heat waves in Indian cities.