Abstract
South-east basins of Iran which is including Sistan and Baluchistan, Kerman, Yazd and Hormozgan provinces has an extensive desert called " Loot " and this region is one of the hottest and driest parts of Iran. Few numbers and the lack of uniformity in scattering of hydrometric stations are main characteristics of this region.
Another problem in this region is that data are short-term, so it leads to have unguaged sites.Data generation helped us to have at most 43 hydrometric stations with 20 years data plus 10 stations with 30 years data.On the other hand, in hydrology for fitting statistical distribution on rainfall and runoff, at least we need to have 30 years data and even more. Thus for analyzing different methods of flood estimation and presenting logical relationships in sub-basins, first of all we gathered meteorology, hydrology and ecological features and also morphometric characteristics were measured.Homogeneity test was done for data and they have been completed in a 20 years data.The group of homogeneous sub-basins has been determined using some methods like: Index Flood, cluster analysis and, multi-variable regression while some physiographic properties and ordinary and linear moments were used.Common statistical distributions have been tested and dominant statistical distribution of region was finally determined Log Pearson type III and based on that peak discharge with different return periods have been estimated to present mathematical models.Then, generated models have been tested using three other sub-basins which hadn't participated in presenting mathematical models.At last, the most appropriate mathematical relationships for flood discharge estimation in different return periods have been achieved in unguaged sites of south-east basins of Iran.

Keywords: Regional flood analysis, Unguaged sites, Flood index, Mathematical model, Cluster analysis, Multi-variable regression