Utilizing Data Mining Techniques to Analyze the Relationship Between Climatic Parameters an‎d Vegetation Cover: A Case Study of Iran

10/20/2026 12:00:00 AM

This thesis was conducted to identify an‎d regionalize contemporary climate change patterns in Iran, using data mining techniques on ERA5 reanalysis data for the period 1940 to 2024. In this regard, time series of temperature, relative humidity, an‎d wind speed were first denoised using the Wavelet Transform technique to extract the main climatic signals. Then, for dimensionality reduction, Principal Component Analysis (PCA) was applied, through which 11 principal components explained 91.5% of the total data variance. Finally, by applying the K-means clustering algorithm to these components, Iran was classified into six distinct an‎d geographically coherent climatic clusters. The results revealed a nationwide warming trend that is, however, regionally heterogeneous. the highest rate of warming was observed in high-elevation inlan‎d areas (up to +0.41°C per decade), while the lowest was in coastal regions (around +0.25°C per decade). Furthermore, a pervasive atmospheric drying trend was identified in five of the six clusters. This research also analyzed the relationship of these climatic clusters with vegetation dynamics (Leaf Area Index), showing that natural an‎d rain-fed ecosystems, such as Zagros, have a much stronger dependency on climatic variables, whereas this link is significantly weaker in arid central regions influenced by factors such as irrigation. These findings provide a data-driven framework for understan‎ding the distinct impacts of climate change an‎d for developing adaptive policies tailored to each region.

تغيير اقليم , تحليل زماني-مكاني , خوشه‌بندي K-means , روندهاي خشكي , ناحيه‌بندي اقليمي

Climate Change , Spatiotemporal Analysis , K-means Clustering , Aridity Trends , Climate Zoning

Zahra Hassanzadeh Shooshtari

Mahboubeh Molavi Arabshahi