Delineating northwest of Iran based on the ratio of precipitation to evapotranspiration using the PCA and WARD methods

Classification in terms of climate and hydrological variables is one of the basic requirements in water resource planning. The same climate characteristics in the different geographical areas causes that climate scientists, separate the areas with the same climate and put the zones in one group that have similar features. Cluster analysis is one of the multivariable statistical methods that is used to reduce the data and finding homogeneous regions. Principle component analysis (PCA) is one of the multivariable statistical methods that is done by analyzing the eigenvalues of correlation or covariance matrix. The purpose of these methods regionalization or zoning and extending point information to an area which resulted in providing the information for the zones or stations without information or with insufficient information. Evaporation as output of water resources and contrary to that, precipitation, is considered as input of water resources. Then the ratio of these two variables in the given time scale (e.g., monthly) can express the climatic condition of the area in terms of dryness so whatever the rainfall is greater than evaporation, can represent the wetness of the area and success of dry farming. In addition to the total annual rainfall in dry farming, the distribution of rainfall during the period of growth is the important parameter. Existence of enough moisture in the process of budding is necessary for growth and plant establishment. Also in the spring, which coincides with the flowering and grain growth, due to the warming weather and increasing evapotranspiration, plant water requirement increased. Rainfall in these periods could have significant influence in the crop growth and production.
In this study, in order to investigate the capability of northwest of Iran for dry farming this area was delineated based on the ratio of precipitation to evapotranspiration. For this purpose, the climatic parameters of 22 synoptic and climatology stations of were used during the period from 2000 to 2006. The average evapotranspiration for each station was determined by FAO-Penman-Monteith method and by using the monthly precipitation data, the ratio of precipitation to evapotranspiration was calculated. The missing data were reconstructed by the program written in the FORTRAN programming. The principal component analysis was done by MINITAB software and correlation matrix. Then, by using the scree plot, the first up to fourth factors with high eigenvalues were selected for zoning. In the next step, the same coefficient curves for every selected component, drawn on the map of the study area and the homogeneous regions were specified based on the ratio of precipitation to evapotranspiration. Finally the delineated areas were simultaneously drawn on a map.
Then, by principle component analysis method and six selected principal components after rotation that contain 74.1 percent of data's variance, six homogenous regions based on the ratio of precipitation to evapotranspiration were introduced among which the second region with the mean ratio of precipitation to evapotranspiration equal to 0.5, has the highest priority of the dry farming. By using the WARD method and drawing dendrogam, two homogenous regions were achieved that in autumn, the second region with the value equal to 0.49 of precipitation to evapotranspiration and in spring, the first region with the value equal to 0.59 for mentioned parameter were in the priority of dry farming. By using the S-statistic method, cluster's homogeneity test was performed which was significant at the level of 5%. The F-test between the standard deviation of zones in the WARD method, showed a significant difference between them so that the ratio of the standard deviation (2.68) was obtained greater than the critical value (2.57). Therefore, with respect to the quality of the zone's separation and clustering data expansion, the WARD method was selected. The south of East Azarbaijan Province, has the lowest amount of precipitation to evapotranspiration, in case of appearing water shortage in this area, agriculture and horticulture will be in serious threats so that requires prior planning and appropriate actions. Finally, it can be said that the mentioned method can be used as a guide for proper management and optimal water resources operation on a large scale, such as in the basin or province. Moreover, by using this method, critical areas of water resources can be detected easier and the appropriate coping plans can be prepared to avoid serious damages to farmers of those areas.