Document Type: Research Paper
Graduated Master Student, Water Engineering Department, College of Agriculture, Shiraz University, Iran (Postal code: 7144165186).
Assistant Professor, Water Engineering Department, College of Agriculture, Shiraz University, Shiraz, Iran (Postal code: 7144165186).
Professor, Water Engineering Department, College of Agriculture, Shiraz University, Shiraz, Iran (Postal code: 7144165186).
Clustering was used to divide dryland farming areas in western Iran into homogeneous
sub-regions to identify dryland farming potential, considering drought impacts. Clustering
utilized eight algorithms/four indices to detect optimal number of clusters. Ward’s algorithm
validated by Silhouette index, produced the best result by detecting 7 dryland farming clusters.
Based on similar P/ETo values, four sub-regions were recognized among 7 clusters.
Northwestern sub-region was ranked first, followed by central, northeastern and southern
sub-regions. Drought impact analysis led to 6 optimal clusters by Ward’s algorithm, validated
by Silhouette index. Ranking criteria utilized drought characteristics, obtained from 3- to 12-
months SPI analysis. Northwestern sub-region and parts of central sub-region, with respectively
first and second rankings for dryland farming, are also least affected by droughts. Areas in
central sub-region with good dryland farming potential can be strongly impacted by droughts.
Northeastern and southern sub-regions respectively ranked third and fourth for dryland farming,
were severely affected by droughts. In conclusion, areas with highest dryland farming potential
were impacted minimally by drought, while areas with lowest potential were strongly affected
by droughts. However, sub-regions with good dryland farming potential were be severely
influenced by drought. Therefore, drought analysis should be considered for dryland farming