Rainfall Prediction Using Artificial Neural Network

Resti Salmayenti, Rahmat Hidayat, Aris Pramudia


Artificial neural network (ANN) is widely used for modelling in environmental science including climate, especially in rainfall prediction. Current knowledge has used several predictors consisting of historical rainfall data and El Niño Southern Oscillation (ENSO). However, rainfall variability of Indonesian is not only driven by ENSO, but Indian Ocean Dipole (IOD) could also influence variability of rainfall. Here, we proposed to use Dipole Mode Index (DMI) as index of IOD as complementary for ENSO. We found that rainfall variability in region with a monsoonal pattern has a strong correlation with ENSO and DMI. This strong correlation occurred during June-November, but a weak correlation was found for region with rainfall’s equatorial pattern. Based on statistical criteria, our model has R2 0.59 to 0.82, and RMSE 0.04-0.09 for monsoonal region. This finding revealed that our model is suitable to be applied in monsoonal region. In addition, ANN based model likely shows a low accuracy when it uses for long period prediction.


Artificial neural network; ENSO; IOD; Prediction; Rainfall

Full Text:


DOI: http://dx.doi.org/10.29244/j.agromet.31.1.11-21

Copyright (c) 2017 Agromet

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.