Projection pursuit regression in statistical downscaling model using artificial neural network for rainfall prediction

• Published in: Journal of physics. Conference series 2021-05, Vol.1872 (1), p.12021
• Main Authors: Farikha, E F, Hadi, A F, Anggraeni, D, Riski, A
• Format: Article
• Language: English
• Subjects: Agricultural production; Artificial neural networks; Forecasting; General circulation models; Neural networks; Physics; Prediction models; Rainfall; Regression models; Statistical analysis; Statistical prediction
• ISSN: 1742-6588; 1742-6596
• DOI: 10.1088/1742-6596/1872/1/012021

Rainfall prediction is important for farmers to be used in making policies, especially in areas of agricultural production, include in Indonesia. The availability of information about rainfall requires an accurate forecasting method. The General Circulation Model (GCM) is used in dynamic prediction to obtain rainfall information for one month, but with its low resolution, this model cannot be used to obtain information on a small scale so that a statistical downscaling (SD) model is needed. The Projection Pursuit Regression (PPR) used in this SD includes nonparametric and nonlinear approaches to processing large dimensional data that can describe small dimensions through a projection process. This research is further explained using a neural network-based approach, that is Artificial Neural Network (ANN) in a statistical downscaling model with applications for analysis of events related to rainfall prediction. In this case, the data will be part of the model formation of statistical downscaling. The SD prediction model uses several predictors, where some of these predictors have a physical relationship between the atmosphere and rainfall. The predictor variables are taken from the GCM output, the predictor variables used precipitation.