Simulation of wet and dry day sequences using Binary discrete autogressive moving average models for selected areas in the Philippines

Date

1993

Abstract

An attempt to model the wet and dry sequences of rainfall in the Philippines was made using the family of the Binary Discrete Autoregressive Moving Average (BDARMA(p,q)) processes. The use of the low order BDARMA model was found to be appropriate only for thirteen stations out of the thirty-eight stations initially studied. The BDARMA (1,0) or first order binary Autoregressive (BDAR (1)) model, BDARMA (0,2) or second order binary Moving Average (BDMA (2)) model and the first order binary Autoregressive Moving Average (BDARMA (1,1)) model were found to fit well the autocorrelation function of the different rainfall series, but only for the wet season. The dry season in all 38 stations has a long persistence of dry days that will make the fitting of the models inappropriate. The binary DARMA (1.1) model was found to reproduce well the probability distributions of the length of dry runs since these are not affected by extreme values of length of run during the wet periods. The binary DMA (2) and the binary DARMA (1,1) models, on the other hand, were found to reproduce well the probability distributions of the length of wet runs since the distributions are more varied and are affected by the presence of extreme values of the length of wet runs. The Portmanteau test showed that most of these models adequately fit the probability distributions of the run length of wet and dry days. The models, however, showed poor performance in short-term forecasting even for only one-day lead forecasts. The forecast rainfall have relatively low autocorrelation coefficients showing weak relationship between succeeding rainfall events.

Document Type

Master Thesis

Degree

Master of Science in in Agrometeorology

Language

English

LC Subject

Rain and rainfall -- Philippines

Location

UPLB Main Library Special Collections Section (USCS)

Call Number

LG 995 1993 A38 D67

This document is currently not available here.

Share

COinS