Air Temperature Prediction with LSTM and LSTM-GRU: A Two-Year Meteorological Data

Air temperature is a critical parameter in me-teorology, as understanding its patterns plays a key role in identifying weather conditions, particularly extreme events, and in mitigating their potential impacts. Accurate air temperature prediction can greatly enhance preparedness for such conditions. This study develops two deep learning models, Long Short-Term Memory (LSTM) and a hybrid Long Short-Term Memory - Gated Recurrent Unit (LSTM-GRU), to predict air temperature based on weather data collected over a two-year period in the Batur Dieng region, with a 10-minute resolution. The performance of both models was evaluated using Mean Absolute Error (MAE), coefficient of determination <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$(R^{2})$</tex>, and Root Mean Squared Error (RMSE). The results indicate that the LSTM-GRU model outperforms the LSTM model, achieving an MAE of 0.01181, an <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$R^{{2}}$</tex> of 0.98902, an RMSE of 0.0159, and a running time of 5 minutes and 54 seconds. The findings suggest that LSTM-GRU provides more accurate predictions, particularly in handling rapid air temperature fluctuations, making it a more suitable model for air temperature forecasting in this region.