European Water 49: 43-63, 2015.
© 2015 E.W. Publications
Short-term drought forecasting combining stochastic and geo-statistical
approaches
C. A. Karavitis
1*
, C. G. Vasilakou
1
, D. E. Tsesmelis
1
, P. D. Oikonomou
2
, N. A. Skondras
1
, D. Stamatakos
1
,
V. Fassouli
1
and S. Alexandris
1
1
Department of Natural Resources Development & Agricultural Engineering, Agricultural University of Athens,
11855 Athens, Greece
2
Department of Civil and Environmental Engineering, Colorado State University, Fort Collins, Co. 8o523, USA
*
e-mail: ckaravitis@aua.gr
Abstract: The current work presents the application of the seasonal Auto Regressive Integrated Moving Average Model
(ARIMA) using the Standard Precipitation Index (SPI) as a drought indicator and then depicting the spatial
distribution through geo-statistical methods. Greece is very often facing the hazardous impacts of droughts, hence
presenting an almost ideal case for such an application. The applied methodology used precipitation data from 55
meteorological stations in two sets distributed throughout the country. To assess the drought, SPI was calculated for a
variety of temporal steps. The statistical analyses of the examined temporal step parameters were performed and the
Seasonal ARIMA model was employed. Then based on such link, forecasting attempts were derived and presented.
The effort may lead to a further understanding of drought duration, magnitude and spatial extend in semi-arid areas. It
is believed that such a methodology may provide some useful anticipatory information on the area’s vulnerability to
drought and thus portraying the system’s susceptibility to change, damages and losses. In this context, drought
contingency planning may be included in the decision making arsenal in order to also widen existing perceptions of
the area’s inherent weaknesses and limited resilience to both anthropogenic and natural hazards, serving at the same
time as an early warning mechanism.
Key words: Standardized Precipitation Index, meteorological drought, statistical analysis, seasonal ARIMA model, forecasting,
contingency planning, Greece
1. DROUGHT DEFINITIONS AND CHARACTERISTICS
The equitable water use as an integral part of the quest for integrated water resources
management confronts incessant multifaceted difficulties due to changes in human systems values,
societal structural transformations and environmental aggravations. Such basic and evolving
changes have created a context of high uncertainty and complexification as access to adequate water
resources is generating conflicts and competing interests, further intensified by historical factors,
traditional values and customs, religious considerations, and geographical peculiarities (Vlachos,
2004; Barraque et al. 2008; Karmarkar, 2012; Karavitis, 2008; Karavitis et al. 2014). Under such
circumstances, the reoccurring drought events – mainly due to changes in precipitation patterns,
duration and intensity – may accelerate the rise of water related social-ecological crises, particularly
in areas of high drought susceptibility such as the subtropics and mid-latitudes, including the
Mediterranean area. The area is already one exhibiting high vulnerability to drought due to
socioeconomic (e.g. irrigated agriculture, fragile economies) and natural conditions, such as semi-
dry environment with dry long summers (Gleick, 1993; Karavitis and Kerkides, 2002; IPCC, 2007;
2013). Therefore, drought analysis may be considered as a crucial fraction of any integrated water
resources management approach.
The pertinent literature clearly indicates that drought is a normal, recurring feature of any
climatic zone, although often perfunctorily envisaged as an exceptional and unanticipated one
(Hagman, 1984; Bruce, 1994; Grigg, 1996; Karavitis, 1999; Bordi et al. 2006; Karavitis et al.