LONGTERM PRECIPIDATION FORECASTING USING CLIMATE SIGNALS CLUSTERING WITH RESPECT TO PRECIPITATION VARIATIONS WITH MODIFIED K-MEANS METHOD (CASE STUDY: PRECIPIDATION FERECASTING OF SISTAN-BALOUCHESTAN PROVINCE)

Zarhrai, B.; Roozbahani, A.

LONGTERM PRECIPIDATION FORECASTING USING CLIMATE SIGNALS CLUSTERING WITH RESPECT TO PRECIPITATION VARIATIONS WITH MODIFIED K-MEANS METHOD (CASE STUDY: PRECIPIDATION FERECASTING OF SISTAN-BALOUCHESTAN PROVINCE)

Document Type : Article

Authors

B. Zarhrai ¹

A. Roozbahani ²

¹ Center of Excellence for Engineering and Management of Infrastructures University of Tehran

² School of Civil Engineering University of Tehran

Abstract

Studying climatic variations and finding different hydrological variables, such as precipitation, can be very useful in the prediction of these variables. Some recent research has tried to explain the relation between large scale climatesignals, like Sea Surface Temperature (SST), andsome hydrological variables, such as precipitation. Inthis paper, a novel method for clustering is presented,which is called \Modified K-Means". This method is based on the ordinary K-Means method, but the framework of selecting the cluster centers is modified in a way wherein clustering SST and its relation with precipitation are considered simultaneously. This model tries to find the relationship between climatic signals and precipitation variations. 20 rain gauges in the Sistan & Baluchestan province in Southeastern Iran have been considered in the case study. The use of the Modified KMeans method for showing the relation between precipitation in various seasons and temporal-spatial variations of SST is an innovative aspect of this paper.

Keywords

Clustering

climate signals

modified K-Means

GA clustering

sea surface temperature (SST)