• cluster analysis;
  • dry spells;
  • Iran;
  • L-moments;
  • regionalization;
  • spatial analysis


This paper presents a methodology for regional frequency analysis and spatial pattern features of the Annual Maximum Dry Spell Length (AMDSL) as an indicator of drought conditions using the well-known L-moments approach and statistical-based methods. Applying Ward's cluster-analysis method identifies eight regions with distinctive AMDSL behaviours for Iran. Homogeneity testing indicates that most of these regions are homogenous. The goodness-of-fit test ZDist shows that Generalized Logistic; Generalized Extreme Value and Pearson type III, distributions fit best for most regions. The spatial pattern of L-Moment statistics demonstrates that although the northwestern and northern parts of the country experience short dry spells, these periods are inconstant, and extreme dry spell events may happen in these areas. Almost all spatial mapping of AMDSLs at different probabilistic levels demonstrates that dry spells increase gradually from west to east and from north to south, and the southern parts (especially along the Persian Gulf and Oman Sea) and central areas, including most agricultural lands, stand out as the most sensitive to soil moisture deficits, because of longer lasting droughts. © 2013 Royal Meteorological Society