Vector Autoregressive (VAR) for Rainfall Prediction

Abstract

Weather and climate information is useful in a variety of areas including agriculture, tourism, transportation both land, sea and air. For that, up to date weather and climate data and its forecasting are essential. This study aims to create rainfall modeling with Vector Auto Regressive (VAR) using circular data and linear data. The data used comes from the station climatology Darmaga Bogor period 2006-2017. The VAR model (2) of the rainfall variables in the t-month is affected by the t-1 moisture air moisture, the t-2 moisture air and the air temperature at t-2. This VAR model (2) is used to forecast the next period. The mean absolute percentage error (MAPE) VAR (2) obtained was 42.18. The novelty of the study is (1) VAR modeling for rainfall prediction, (2) Use of circular data for wind direction data.