Introduction: Dry Spells are a defined as a sequence of days without rain. It is very important to evaluate and determine the efficiency of statistical models to estimate the start and end of dry periods and it is widely used in research related to different sectors such as water resources, agriculture, watershed management. Therefore, it is very important to identify the most appropriate models for displaying rainfall distribution to determine the beginning and end of dry periods, because it can be used in various applications such as water resources management, agricultural planning and hydrological departments. The main and important aim of this research is to modelling behavior of dry Spells and determine the appropriate model to evaluation of periods.

Materials and methods: At first, 12 rain gauge stations with a period of 1991-2020 were used to analyze precipitation and extract dry spells lasting 15 days or more. The threshold of 5 mm was used as the final threshold in the analysis because the explanation of more than 50% of the weight of the period. Using the IF function in MATLAB software, dry sequences were calculated for the two criteria of maximum and average length of periods. In the next and important step of this research, in order to determine the efficiency and Performance of the best model in fitting and estimating dry periods, it was necessary to examine the probability distribution and relationships of the models. for weight the models we used, SWARA method.

Results and discussion: The climatic behavior of dry spells based on the past and present situation showed that in the future, the average dry spells will increase from east to west and from north to south. The results showed that the average annual dry spells change between 7 and 9 days and the deviation from their average change between 9 and 12 days. In general, the decreasing trend of dry spells has a steeper slope in the direction from north to south. The negative binomial distribution has the most weight in the estimation of the eastern and northeastern regions of the province (on average 87% of the weight of courses in each station) and the combined Markov distribution model of order 1 to 10 has the most weight in the tropical regions of the south and southwest of the province (83% of the weight of the courses). And also the distribution of Weibull fit explained 91% of the weight of the courses for the central regions of the province. In the province as a whole, the negative binomial distribution between observed and estimated values has explained a good weight.

Conclusion: The results of this research, based on several statistical methods and their weighting, showed that the behavior of statistical models in estimating the length of dry spells is strongly influenced by the type of climate of the regions and topographical conditions related to it. The combined fitting of the results of the models showed that for proper and accurate evaluation of dry periods, using the results of several models and then their final weighting can provide a correct explanation of the behavior of the periods. In most of the studies about dry spells, weighting of models has not been considered. In Kohgilouyeh and Boyer Ahmad provinces, due to the different topographical conditions and uneven rainfall in the provinces, the length of long-term dry spells, especially in the central, southern and western dry regions, is a function of the rainfall weight and its behavioral fluctuations. Determining the weight of course criteria such as; The length of periods, probability of occurrence and reversibility indicate that the border of dry areas will be vaster and more northerly in the future climate of the province. In most of the studies about dry Spells, weighting of models has not been considered. In Kohgiluyeh and Boyer Ahmad province, due to the different topographical conditions and uneven rainfall in the provinces, the length of long-term dry spells, especially in the central, southern and western dry regions, is a function of the rainfall weight and its behavioral fluctuations. Determining the weight of course criteria such as; The length of spells, probability of occurrence and reversibility indicate that the border of dry areas will be extended and more northerly in the future climate of the province.