Nonparametric methods for fitting the precipitation variability applied to Dobrudja region

Abstract

Modeling the precipitation evolution at a regional scale is a topic of interest in order to predict the weather evolution and climate change, with their multiple consequences, in a region where the annual drought frequency is about 89%, as in Dobrudja. Therefore, in this article we present the result of modeling the annual and monthly precipitation evolution in the period January 1965 - December 2005, in Dobrudja, a region of situated in the South – East of Romania, between the Danube and The Black Sea. ANOVA, followed by the Tukey LSD and the Scheffe tests led us to the conclusion that between the ten studied series, one has a particular behavior, due to the geographical position of the meteorological station. Thus, the precipitation evolution in the entire region has been modeled using the data provided by the other nine series, with small loss of information. The models proposed by us used nonparametric approach (wavelets and smoothing splines) and the comparison between them has been done.