Modeling the spatial structure of annual temperature in Iran

Abstract

Understanding temperature and its spatial and temporal variations can prepare the ground for developing climate planning. One of the good approaches to understanding temperature variations is to use statistical information across time and space. Spatial statistical techniques are one of the practical approaches for gaining a deep and scientific understanding of the behavior of the spatial variables. The present study was an attempt to model the spatial structure of Iran’s annual temperature. For this purpose, the data gathered from 320 weather stations in Iran including synoptic and climatologic stations from their establishment time to the 2014. In line with this goal, OLS, spatial lag and spatial error models were used to determine the temperature structure and predict its spatial variation based on geographic features (elevation, slope, and aspect). The results showed that the maximum and minimum percentage distribution of Iran’s annual temperature respectively lie in the range of 90% and above in the South, Southwest, and Southeast and in the range of 50% in the Northwest and West of Iran. Moran’s I value for annual temperature was found to be 99%, which is indicative of a positive spatial autocorrelation. Scatter plot and the local Moran’s I map showed that the highest pixel values (temperature points) and their neighbors lie at the subgroups with high–high and low–low values. In other words, Iran’s annual temperature follows a cluster pattern with a high–high temperature value and a cluster pattern with low–low temperature value. Finally, spatial regression modeling on Iran’s annual temperature showed that variables such as latitude and longitude, elevation, slope and aspect show positive and negative correlations with annual temperature at different significance levels in all three methods, i.e., OLS, spatial lag and spatial error. In the OLS method, longitude and in the spatial lag method, latitude were found to have a positive and direct correlation with annual temperature. However, the results from spatial regressions showed that the spatial error is a better model for predicting Iran’s annual temperature.