TITLE:
A Statistical Model for Long-Term Forecasts of Strong Sand Dust Storms
AUTHORS:
Siqi Tan, Moinak Bhaduri, Chih-Hsiang Ho
KEYWORDS:
ARIMA Model, Empirical Recurrence Rate, ERR Plot, Point Process, Time Series
JOURNAL NAME:
Journal of Geoscience and Environment Protection,
Vol.2 No.3,
June
12,
2014
ABSTRACT:
Historical evidence indicates that dust
storms of considerable ferocity often wreak havoc, posing a genuine threat to
the climatic and societal equilibrium of a place. A systematic study, with emphasis
on the modeling and forecasting aspects, thus, becomes imperative, so that
efficient measures can be promptly undertaken to cushion the effect of such an
unforeseen calamity. The present work intends to discover a suitable ARIMA
model using dust storm data from northern China from March 1954 to April 2002,
provided by Zhou and Zhang (2003), thereby extending the idea of empirical
recurrence rate (ERR) developed by Ho (2008), to model the temporal trend of
such sand dust storms. In particular we show that the ERR time series is
endowed with the following characteristics: 1) it is a potent surrogate for a
point process, 2) it is capable of taking advantage of the well developed and
powerful time series modeling tools and 3) it can generate reliable forecasts,
with which we can retrieve the corresponding mean number of strong sand dust
storms. A simulation study is conducted prior to the actual fitting, to justify
the applicability of the proposed technique.