<![CDATA[Forecasting the weather conditions is very important to support all community activities. An accurate weather forecast requires knowledge and experience from weather forecasters and is supported by weather modelling technology. In this study, a rainfall intensity modelling was carried out using an artificial neural network (ANN) at the Kemayoran Meteorological Station. In the process of making ANN models, data training is required using past weather conditions. The data used for training in making ANN models are daily weather data for January 2011 until December 2019, which was then tested using a case study from January until August 2020. Model variations are made based on the type of input and the number of hidden layers to determine differences in the use of the predictor data. The ANN model was then created using 3 layers consisting of input layer, hidden layer, and output layer. Furthermore, the model’s comparison is tested using the correlation coefficient (R) and mean absolute error (MAE) to determine the best model. Based on the research results, rainfall prediction using input parameter data for daily weather conditions consist of temperature, humidity, and sunshine has a correlation coefficient (R) is 0.3 – 0.5 and a mean absolute error (MAE) is 9.7 – 9.8 mm. Meanwhile, if the model is made with the rainfall input parameter in the previous days, the correlation coefficient (R) is only 0.1 – 0.3 with the mean absolute error (MAE) is 11.3 – 12.3 mm. This condition indicates that a better predictor to predict daily rainfall using an artificial neural network is to use the input parameter of surface weather conditions.]]>
Copyrights © 2020
