پهنه‌بندی اقلیمی استان اردبیل با استفاده از روش‌های چند متغیره

Abstract
Climatic zoning and knowledge of effective climatic factors and elements can be very effective in determining climatic potentials. The purpose of this research is to pay attention to this principle in order to know the environmental capabilities and limitations and to deal with natural disasters. In this regard, the climatic zoning of Ardabil province was discussed using modern methods. First, for 11 synoptic stations of this province, 22 important meteorological parameters for the 12-year statistical period (2009-2021) were received from the General Meteorological Department. In the next step, the annual average of these variables was obtained for each station, and since their measurement scale was different, they were standardized in the Spss software environment. Then, in the same software, factor analysis was performed on these data and the output results were 5 main factors, which are: the first thermal factor with 31.89% variance, the second pressure-steam factor with 21.08% variance, the third moisture factor with 18.61 variance, the fourth precipitation factor with 14.05 percent variance and the fifth wind-radiation factor with 7.48 percent variance, which in total explained 93.12 percent of the climatic behavior of the region. In the next step, according to the output of the 5 factors, three similar climatic regions were distinguished from each other using Ward's hierarchical cluster analysis method. Also, to show the scope of activity of each of these factors, climate zoning was done in Gis software environment and using Idw interpolation method. According to the final map obtained, most of these factors were concentrated in the cities of Parsabad, Beilesawar, Nemin, Serain and Givi. According to Dumarten's climate classification, it was also recognized that the climate of Ardabil province is part of the semi-arid climate.
1. Introduction
Climate is a complex condition and a summary of the atmospheric condition of a region that changes in relation to geographical factors. In the classification of different climates of the earth, there is no general method that is accepted by all scientists, and the classifications are based on different goals. To know the climate, a set of rules is used that places regions with similar climates in one group and it is referred to as climate classification (Yarmoradi and Abbasnejad, 2019: 47). In other words, a region of the earth's surface where the combined effects of climatic factors cause the establishment of homogeneous conditions is called a type of climate or a climatic region. To investigate the climate of an area, it is necessary to pay attention to climatic elements and factors (Vahdani, 2012: 405). Zoning phenomena according to place has a very long history in geography. The result of this study is the separation of geographical areas and the emergence of regional geography. Each geographical area is a part of the earth's surface that has a significant internal consistency in terms of the phenomena and processes in it; Therefore, the main goal of climate grouping is to discover the existing order in the weather conditions of the entire earth's surface and to know the main weather phenomena accurately and comprehensively (Kaviani and Alijani, 2013: 364).
Today's climatologists use various methods for classification, and in all these methods, two things are considered. The first is to determine the necessary criteria for classification, and the second is to determine the boundary between two groups or climate zones. Climate classification methods are mainly divided into two groups, which are: 1- Experimental methods 2- Genetic methods. The first classification is based on the actual observations of the effect of climate on changes in environmental phenomena and is based on two or three climatic variables. This classification has a very long history, and the terms used to name different climate zones in this type of system are derived from the names of regions and elements of the natural environment on the surface of the planet, such as savannah or steppe. Since in this type of classification system, all climate factors and elements are not taken into account and it is considered a weakness for it, other methods for climate classification were created, the basis and origin of which are the factors that create climate, such as solar radiation, characteristics of air masses, general circulation. atmosphere, etc., and this type of climate classification is called genetics (Jaafarpour, 2017: 8).

Research Methods
In this research, the annual average of 22 meteorological parameters of 11 synoptic stations located in Ardabil province, for the twelve-year statistical period (2009-2021), was used to perform statistical analysis. Due to the difference in data measurement scale in SPSS software, these data were converted into standard Z score and factor analysis method was used to reduce the number of variables. This method was popularized by Lorenz in 1956 in climatology research and called the name of this method uncorrelated empirical function. The factor analysis method combines all dependent variables as a new factor. The product of this method is used in clustering or grouping based on distance. The factor analysis mechanism itself is based on the correlation between variables, and its most important feature is to be able to express the relationship between the primary variables and the created factors in a simple way, and the created factors from a scientific point of view. can be interpreted or justified (Alijani, 1381:180). Using the factor analysis method, the factor load of each studied station was obtained, and in order to show the scope of activity of each climatic factor in the Gis software environment, the IDW interpolation method was used. In the following, ward hierarchical cluster method was used to determine similar climatic regions. This method, which was presented in 1939, can be performed on quantitative data. Hierarchical analysis and average cluster analysis are the most widely used algorithms. The difference between these two algorithms is that in average cluster analysis, the number of groups or classes is determined in advance and each parameter is placed in these classes, but in the method of hierarchical analysis, the number of clusters or the groups are unclear and the method identifies these groups based on multiple techniques and shows them in the form of a tree branch diagram. In this method, two issues are of interest, one is the integration of parameters and the other is measuring the distance between variables. In the discussion of merging groups, Ward's method is more applicable and uses the analysis of variance approach to evaluate the distance between clusters, but in the second discussion, the most common method of calculating distances is the Euclidean distance, which measures the geometric distance in multidimensional space (Farajzadeh, 2014: 99). . In the last step, the Dumarten method was used to perform the classification in an experimental way. The relationship of this method is as follows:
Relationship 1: I=P/(T+10)
In the above formula, I is the dryness coefficient, P is the average annual rainfall of each station, and T is the average annual temperature of each station.