Extended Abstract
Introduction
Changes in the frequency distribution of temperature are among the most important manifestations of climate variability. Such changes not only reveal the nature of global warming but also, due to temperature’s undeniable interaction with other environmental variables, illuminate the root causes of many other environmental changes. Contrary to the common belief that changes in mean temperature are the primary indicator of warming, recent research shows that small changes in the mean stem from profound alterations in the temperature frequency distribution. Furthermore, statistical extreme value theory indicates that the frequency of temperature extremes depends more on the variance (scale parameter) of the distribution than on its mean (location parameter). Therefore, understanding changes in climate variance is as valuable as tracking changes in means, because the economic costs imposed by temperature variability far exceed those from mean temperature changes. For a vast country like Iran, with remarkable geographical and climatic diversity, examining changes in temperature frequency distribution can reveal regional and local patterns. This study aims to analyze the evolution of Iran’s minimum temperature frequency distribution over the past five decades (1975–2024).
Materials and Methods
Two data types were used: reanalysis and station data. The reanalysis dataset comprises hourly 2‑meter air temperature from ERA5‑Land, with hourly temporal resolution and 0.1° spatial resolution (200×150 cells) over a 50‑year period (21 March 1975 – 20 March 2025). Of 30351 total pixels within the framework, 28023 were data‑bearing (over land and inland water bodies) and 15577 pixels fell within Iran’s political borders. Daily minimum temperature was extracted as the minimum of a 24‑hour vector (from 21:00 UTC of the previous day to 20:00 UTC of the current day). Station data consisted of daily minimum temperatures from 389 Iranian meteorological stations with varying record lengths (10 to over 73 years); 143 stations with >30 years of overlap with reanalysis data were used for validation. For frequency distribution analysis, the temperature range -37 to +40°C was divided into 77 bins of 1‑degree interval. Annual minimum temperature frequencies per pixel per bin were counted and converted to relative frequency percentages. A centered array of relative frequencies (subtracting the spatial weighted mean of each column) with dimensions 15577×3850 was created. To identify spatial and temporal patterns, Singular Value Decomposition (SVD) was applied. Orthogonal left singular vectors represent spatial patterns, and right singular vectors represent temporal patterns of the temperature bins. Reanalysis validation was performed by calculating bias (reanalysis minus station) and comparing frequency distributions. The mean bias for daily minimum temperature was -0.2°C (slight underestimation).
Results and Discussion
The results show that over the past five decades, Iran’s minimum temperatures have mostly ranged between -15 and +30°C, with 15°C being the most frequent. The peaks of maximum frequency have intensified since 1998. Comparing the frequency distributions of 1982 and 2023 shows that 1982 had a bimodal distribution (temperatures near zero in the cold season and near 16°C in the warm season), whereas 2023 exhibits a more unimodal distribution with a 4% frequency peak, causing the 2023 mean minimum temperature to be about 2.9°C warmer than that of 1982. The first component of SVD explained nearly 42% of the total data variance. Its scores partition Iran into two main parts: Cold Iran (positive scores, including the mountainous northwest, Zagros, and Alborz) and Warm Iran (negative scores, including plains, lowlands, and southern coasts). The first eigenvector revealed two important temperature ranges: a cold range (-7 to +3°C) and a warm range (20–29°C). The best representative of the dominant pattern is Mount Hezar southwest of Rayen (Kerman) with a bimodal distribution on -9 and -3°C, while the opposite pattern is best represented by the Gulf of Oman coast northeast of Chabahar with a bimodal distribution on 26 and 17°C. Trend analysis of the first eigenvector shows that in Cold Iran, the frequency of temperatures below -5°C has decreased over the past five decades, replaced by increased frequency of temperatures from -1 to +8°C. In Warm Iran, the frequency of temperatures above 27°C has increased and that of temperatures 15–25°C has decreased. Comparing the second half of the period (2000–2024) with the first half (1975–1999) shows that the reduction in <‑5°C temperatures is stronger in the Sefidrud, Urmia, Aras, Karun, and Gavkhuni basins, while the increase in >27°C temperatures has occurred along the southern coasts and central Iran.
Conclusion
The frequency distribution of minimum temperature in Iran has evolved over the past five decades. This evolution, in both the cold and warm parts of the country, means a rightward shift of the frequency distribution curve: decreased frequency of low temperatures and increased frequency of high temperatures. An 1.8% decrease in the frequency of temperatures below -5°C and a 1.1% increase in temperatures above 27°C have respectively raised Iran’s mean minimum temperature by 0.19°C and 0.33°C. Changes in other temperature bins (-1 to 8°C and 15–25°C) have contributed -0.04°C and +0.42°C, respectively. Overall, these changes have increased Iran’s mean minimum temperature in the second half of the period by 0.94°C relative to the first half. The largest increase in minimum temperature (more than 0.7°C) occurred in the Sefidrud, Urmia, Aras, Karun, and Gavkhuni basins. This research demonstrates that analyzing temperature frequency distribution changes is a more fundamental approach than merely studying means or extremes, and can provide environmental planners and water resource managers with a more precise picture of Iran’s climate evolution.