TY  - JOUR
T1  - A Comparison of Low-Flow Clustering Methods: Streamflow Grouping
AU - , Ercan Kahya AU - , M. Cuneyd Demirel 
JO  - Journal of Engineering and Applied Sciences
VL  - 2
IS  - 3
SP  - 524
EP  - 530
PY  - 2007
DA  - 2001/08/19
SN  - 1816-949x
DO  - jeasci.2007.524.530
UR  - https://makhillpublications.co/view-article.php?doi=jeasci.2007.524.530
KW  - Cluster analysis
KW  -low-flow
KW  -homogeneous region
KW  -Ward`s method
KW  -Turkey
KW  -thematic maps
AB  - In this study, three clustering algorithms which use agglomerative clustering procedure to identify groups of similar catchments are investigated to determine their effectiveness in low flow clustering scheme. These hierarchical clustering algorithms are single linkage, complete linkage and Ward`s algorithms. The effectiveness of the cluster analysis algorithms is investigated by using monthly minimum streamflow data recorded from watersheds in Turkey with the period of 1964-1994. Furthermore one of cluster validity indices (namely, cophenet correlation coefficient index) is used to strengthen our results. The hierarchical cluster analysis is found to be useful in minimizing efforts needed to identify homogeneous clusters. Ward`s algorithm with Euclidean metric is the one to decrease the variance in each group and appears to be favourable in the applications of low-flow. The results and comparison with our previous studies are also presented in the thematic maps.
ER  -