![Hier klicken, um den Treffer aus der Auswahl zu entfernen](images/unchecked.gif) |
Titel |
Geostatistical Study of Precipitation on the Island of Crete |
VerfasserIn |
Vasiliki D. Agou, Emmanouil A. Varouchakis, Dionissios T. Hristopulos |
Konferenz |
EGU General Assembly 2015
|
Medientyp |
Artikel
|
Sprache |
Englisch
|
Digitales Dokument |
PDF |
Erschienen |
In: GRA - Volume 17 (2015) |
Datensatznummer |
250101348
|
Publikation (Nr.) |
EGU/EGU2015-470.pdf |
|
|
|
Zusammenfassung |
Abstract
Understanding and predicting the spatiotemporal patterns of precipitation in
the Mediterranean islands is an important topic of research, which is emphasized
by alarming long-term predictions for increased drought conditionsÂ[4]. The
analysis of records from drought-prone areas around the world has demonstrated
that precipitation data are non-Gaussian. Typically, such data are fitted to the
gamma distribution function and then transformed into a normalized index, the
so-called Standardized Precipitation Index (SPI)Â[5]. The SPI can be defined
for different time scales and has been applied to data from various regionsÂ[2].
Precipitation maps can be constructed using the stochastic method of Ordinary
KrigingÂ[1]. Such mathematical tools help to better understand the space-time
variability and to plan water resources management.
We present preliminary results of an ongoing investigation of the space-time
precipitation distribution on the island of Crete (Greece). The study spans the
time period from 1948 to 2012 and extends over an area of 8 336 km2. The
data comprise monthly precipitation measured at 56 stations. Analysis of the
data showed that the most severe drought occurred in 1950 followed by 1989,
whereas the wettest year was 2002 followed by 1977. A spatial trend was
observed with the spatially averaged annual precipitation in the West measured
at about 450mm higher than in the East. Analysis of the data also revealed
strong correlations between the precipitation in the western and eastern parts
of the island. In addition to longitude, elevation (masl) was determined to be
an important factor that exhibits strong linear correlation with precipitation.
The precipitation data exhibit wet and dry periods with strong variability even
during the wet period. Thus, fitting the data to specific probability distribution
models has proved challenging. Different time scales, e.g. monthly, biannual, and
annual have been investigated. Herein we focus on annual precipitation which
are fitted locally to a three-parameter probability distribution, based on which a
normalized index is derived. We use the Spartan variogram function to model
space-time correlations, because it is more flexible than classical modelsÂ[3]. The
performance of the variogram model is tested by means of leave-one-out cross
validation. The variogram model is then used in connection with ordinary kriging
to generate precipitation maps for the entire island. In the future, we will explore
the joint spatiotemporal evolution of precipitation patterns on Crete.
References
[1]ÂÂÂP.ÂGoovaerts. Geostatistical approaches for incorporating elevation
into the spatial interpolation of precipitation. Journal of Hydrology,
228(1):113–129, 2000.
[2]ÂÂÂN.ÂB. Guttman. Accepting the standardized precipitation index: a
calculation algorithm. American Water Resource Association, 35(2):311–322,
1999.
[3]ÂÂÂD.ÂT Hristopulos. Spartan Gibbs random field models for geostatistical
applications. SIAM Journal on Scientific Computing, 24(6):2125–2162, 2003.
[4]ÂÂÂA.G. Koutroulis, A.-E.K. Vrohidou, and I.K. Tsanis. Spatiotemporal
characteristics of meteorological drought for the island of Crete. Journal of
Hydrometeorology, 12(2):206–226, 2011.
[5]ÂÂÂT.ÂB. McKee, N.ÂJ. Doesken, and J.ÂKleist. The relationship of drought
frequency and duration to time scales. In Proceedings of the 8th Conference
on Applied Climatology, page 179–184, Anaheim, California, 1993. |
|
|
|
|
|