Vol. 10, No 1, 2013 pp. 79 - 91
UDC 55:311.3
PRECIPITATION FORECAST USING STATISTICAL APPROACHES
Mladjen Ćurić1, Stanimir Živanović2, Darko Zigar3
1Institute of Meteorology, University of Belgrade, 2Emergency Management Sector, Ministry of Interior of the RS, Belgrade, 3Faculty of Occupational Safety, University of Niš
E-mail: curic@ff.bg.ac.rs
Abstract. A statistical approach is often used in the analysis of time series in climatology and hydrology. One of its advantages is the possibility to predict future time series. This can be applied in many circumstances, such as temperature, precipitation, solar radiation and other studies of climate elements. Several models to determine the time series analysis of rainfall were used in this paper. The goal is to find an appropriate model for predicting the trend forecasting of rainfall series. Trend forecasting is one of the important methods of time series analysis (TSA). The simulation shows that the list of 50 data of elapsed values will produce optimal results. The procedures of model selection and evaluation were applied to a series of annual precipitation in the period 1961–2011 in the weather forecast station in Negotin. The obtained results show a decrease in the trend of the total annual precipitation in the period 1961-2011. The results of the analysis and rainfall forecast in the years to follow, in comparison to 2011, show an annual increase of rainfall.
Key Words: meteorological time series, the ARIMA model, trend method, the Mann-Kendall test, precipitation
PREDVIĐANJE PADAVINA POMOĆU STATISTIČKIH PRISTUPA
Statistički pristup se često koristi u analizi vremenskih serija u klimatologiji i hidrologiji. Jedna od njegovih koristi je da predvidi trend budućih vremenskih serija. Ovo može da se primeni u mnogim aplikacijama kao što su temperatura, padavine, sunčevo zračenje i druga istraživanja u vezi sa vremenskim serijama klimatskih elemenata. U ovom radu koristimo nekoliko modela za određivanje analize vremenskih serija padavina. Cilj je da se pronađe odgovarajući model u predviđanju trenda serije padavina. Metod trenda je jedan od značajnijih metoda analize vremenskih serija. Simulacija pokazuje da za listu od 50 podataka proteklih vrednosti daje optimalne rezultate. Postupak izbora i ocene modela primenjen je na seriju godišnjih suma padavina osmotrenih u periodu 1961-2011. god. na meteorološkoj stanici u Negotinu. Dobijeni rezultati pokazuju smanjenje trenda ukupnih godišnjih količina padavina u toku perioda 1961-2011. godina. Rezultati analize i predviđanja toka padavina u narednim godinama u odnosu na 2011. godinu, pokazuju povećanje količina padavina na godišnjem nivou.
Ključne reči: meteorološka vremenska serija, ARIMA model, metod trenda, Mann-Kendall test, padavine