Near-term climate extremes in Iran based on compound hazards analysis

Abstract

Iran, located in arid and semi-arid regions, has faced significant weather and climate extremes in recent years. This study aims to investigate the climate-related hazards associated with precipitation and temperature in Iran during the Hindcast period (1991–2019) and the Forecast period (2023–2028) using the Near-term Climate Prediction (NTCP) project. We investigate compounding climate-related hazards to assess the severity and importance of weather and climate extremes. To accomplish this, we integrated ten climate extreme indices, namely heavy precipitation (R10mm), the Simple Precipitation Intensity Index (SDII), heat wave frequency (HWF), heat wave duration (HWD), cold wave frequency (CWF), and cold wave duration (CWD), along with the Standardized Precipitation Evapotranspiration Index (SPEI-12), which further encapsulates drought frequency (DF), drought duration (DD), drought severity (DS), and drought intensity (DI). The CMIP6-DCPP models effectively simulate climate extremes and their seasonal cycle across Iran, with minor discrepancies in arid and mountainous regions due to data limitations. The result demonstrates a significant anticipated rise in drought frequency and heat wave events throughout the country within the near-term forecast period. Future projections indicate a shift in precipitation patterns, with increased heavy precipitation in the Zagros Mountains and southwest regions alongside more frequent but less intense droughts nationwide. Heat wave frequency and duration are projected to increase, particularly in southern Zagros and eastern and western Iran, with high-altitude areas experiencing accelerated warming. The results project a shift in climate risk distribution over the next decade, with low to moderate-risk areas decreasing by approximately 15.7% and high-risk areas increasing by roughly 10%, encompassing over 36% of Iran’s total area by 2028. Integrated risk maps reveal high to very high compound climate hazard levels across large parts of Iran, necessitating urgent adaptation planning, especially in western, southern Zagros, and eastern regions. Sensitivity analysis confirms that identified multi-hazard hotspots in Iran are spatially robust and statistically significant, reflecting the dominant influence of key climatic extremes.

Introduction

Iran, located in the heart of the Middle East, has a long history of exposure to a variety of natural hazards. Its location on the Iranian plateau, which lies within the Alpine-Himalayan orogenic belt and is surrounded by young, high-elevation mountain ranges of the Tertiary period^1,2,3, predisposes the region to significant geological instability. In addition, Iran’s complex terrain (Fig. 1a), extensive latitudinal range, and proximity to large bodies of water in both the north and south contribute to a highly diverse climate (Fig. 1b). The annual mean temperature varies from 10.6 °C in the northwestern mountainous regions to 28.4 °C along the southern coastal areas of the country (Fig. 1c). Precipitation in the region is also highly variable, with amounts ranging from less than 60 mm in the arid interior to over 1500 mm along the southern coasts of the Caspian Sea (Fig. 1d). The simultaneous interaction among mid-latitude, tropical, and extratropical atmospheric circulation systems further compounds this climatic diversity⁴. Consequently, the Iranian plateau has experienced a wide spectrum of natural hazards over its long history. In addition to the inherent geological hazards driven by intense tectonic activity, the country contends annually with a range of weather- and climate-related risks. These include droughts, heat waves, cold spells, heavy precipitation events, catastrophic floods, dust storms, wind storms, and wildfires. Among the recorded extreme phenomena, the 1972 Iranian blizzard stands out as the deadliest on record; in this event, snowfall reached up to 7.9 m (26 ft), ultimately burying at least 4,000 individuals^5,6,7. Notably, the country also holds the record for the highest surface temperatures ever documented on Earth. In 2005, the Lut Desert recorded a maximum land surface temperature of 70.7 °C, as measured by NASA’s Aqua satellite^8,9,10,11. Furthermore, recent updates utilizing long-term satellite-derived land surface temperature data have confirmed the Lut Desert as the hottest location on Earth, with a peak temperature of 80.8 °C recorded¹².

Natural hazards in Iran have consistently inflicted severe damage on property and human life. The World Bank’s Disaster Risk Management report (2005) estimates that nearly 15% of Iran’s land area and 47% of its population are exposed to at least two concurrent hazards over a 20-year period¹³. This high level of exposure has positioned Iran among the top 20 countries in terms of economic risk from natural disasters¹³. In particular, droughts, floods, heat waves, and dust storms are among the most prevalent hazards¹⁴. Recent studies have documented that global warming has notably expanded both the spatial extent and intensity of climate extremes in Iran^15,16,17. For example, an increase of 1.1 to 2.75 °C in maximum temperature is projected by 2049 across Iran¹⁴. Zarrin & Dadash-Roudbari¹⁶ demonstrated that the RX1day and SDII indices are projected to increase by 4 to 13% on average across Iran by the end of the century, suggesting an intensification of flash flood-inducing precipitation events in the coming decades.

Despite these alarming trends, Iran has not allocated adequate resources for hazard preparedness¹⁸. Moreover, the compound occurrence of multiple extreme events within the same geographical region further complicates risk management strategies¹⁹. A notable recent sequence includes the extensive drought experienced from 2021 to 2023, followed by anomalous heavy precipitation in July 2022, an extreme heat wave in August 2022, and an unprecedented cold wave in January 2023. Each of these events has collectively contributed to substantial damage across extensive areas. Historical expenditure on disaster-related costs in Iran has been considerable; Seddighi & Seddighi¹⁸ report that approximately US$ 29 billion has been spent over the past century on various disasters, with drought and flood alone accounting for over US$ 14 billion and US$ 6.1 billion, respectively. For instance, the catastrophic flood of 2019 resulted in damages estimated at US $2.2 billion, predominantly affecting the agricultural sector and destroying 36% of the national road network^20,21.

It is anticipated that compound hazards will increasingly impact Iran in the coming decades, particularly as global temperatures continue to rise. Therefore, this research is imperative to identify and analyze future climate-related hazards in Iran and to address their compound impacts. The primary objective of this study is to predict near-term climate extremes using the Near-term Climate Prediction (NTCP) framework. In recent decades, global climate models have substantially advanced, with simulations now capturing future climate variations on timescales ranging from seasonal to decadal^22,23. The NTCP, a key initiative of the World Climate Research Program (WCRP), delivers climate predictions at decadal (annual-to-decadal, A2D) timescales by incorporating both internal variability and external forcing²⁴. Skillful decadal predictions offer significant benefits for climate adaptation, resilience, and informed decision-making within socioeconomic and agricultural sectors. NTCP uniquely bridges the gap between initialized seasonal forecasts and longer-term uninitialized climate projections. Its success largely depends on the careful consideration of both initial and boundary conditions^25,26.

The initial state of the decadal prediction system integrates the atmosphere, ocean, land, cryosphere, and biogeochemical components, while external forcing elements such as aerosol emissions, volcanic eruptions, and greenhouse gases (GHGs) are incorporated to reflect their influence on the Earth’s climate system^27,28. The NTCP’s framework was re-evaluated under the Decadal Climate Prediction Project (DCPP), a significant initiative aimed at enhancing decadal prediction capabilities—a framework that has been utilized in our study.

The innovative aspect of this research lies in its prediction of compound hazards in a region like Iran, which is highly susceptible to natural hazards yet remains under-addressed in current literature. This study first identifies climate extreme hotspots in Iran, followed by an analysis of compound occurrences using composite mapping techniques to refine the understanding of risk areas. The findings presented here aim to improve the prediction of climate extremes and their potential to trigger hazards in the coming years, thereby contributing valuable insights for future mitigation and adaptation strategies.

Fig. 1

Study area. (a) Topography of the region and its significant geographical features. (b) Geographical distribution of the synoptic stations employed in this study. The names of the stations are provided in the accompanying table. The background color shows the Köppen-Geiger classification at a 1-km resolution²⁹. (c) 30-year (1991–2020) mean Temperature (°C). (d) 30-year (1991–2020) annual precipitation (mm).

The map in panel (a) was generated using digital elevation data from the Shuttle Radar Topography Mission (SRTM). URL link: https://www.earthdata.nasa.gov/data/instruments/srtm. Panel (b) was generated using the Köppen-Geiger climate classification data from Beck et al.²⁹. URL link: https://doi.org/10.6084/m9.figshare.6396959. The station data in panels (c) and (d) are available from IRIMO. URL link: https://data.irimo.ir. The map was created using ArcGIS Desktop (Release 10, Version number: 10.8.2) by Esri, Redlands, CA: Environmental Systems Research Institute. URL link: https://www.esri.com.

Data and methods

Data

Observation

This study utilized daily minimum and maximum near-surface air temperature and precipitation data from 70 synoptic stations, complemented by output from three models within the CMIP6-DCPP project. Prior to analysis, the observational time series underwent a rigorous quality control procedure based on the criteria established by Schneider et al.³⁰. Following quality control, 70 stations were deemed suitable for characterizing climate variability across Iran from 1991 to 2019 (Fig. 1). Furthermore, the Shapiro-Wilk test was employed to assess the normality and homogeneity of the ground station data, confirming homogeneity at the p < 0.05 significance level³¹.

CMIP6-DCPP

Using initialized predictions may improve the ability to predict climate extremes and reduce uncertainty from internal variability compared to uninitialized projections³². Thus, we implemented the Decadal Climate Prediction Project (DCPP) protocol, which attempts to bridge the gap between forecasting and climate change projections²⁴. We also followed the findings of Delgado et al.³², Konda et al.^33,34, and Asadi-RahimBeygi et al.³⁵ in using the multi-model ensemble of CMIP6-DCPP as a skillful dataset to predict climate extremes at multi-annual and seasonal timescales.

The decadal climate prediction framework consists of three primary components: dcppA (hindcast), dcppB (forecast), and dcppC (predictability, mechanisms, and case studies). This study used dcppA for the historical period (1991–2019) and dcppB for the near-future period (2023–2028). The future climate prediction of the CMIP6-DCPP framework is based on the Shared Socioeconomic Pathway 2–4.5.5, which represents the moderate anthropogenic radiative forcing. The different variant labels of CMIP6-DCPP are distinguished by the ripf Index, which denotes the realization (r), initialization (i), physics (p), and forcing (f) components of the simulation. We used nine distinct realizations, labeled r1i1p1f1 through r9i1p1f1 for each model (Table 1). In this CMIP6-DCPP study, models were initialized each November from 1991 to 2019 to enable the calculation of December to February (DJF) seasonal averages. For instance, a model initialized in November 1990 produced a hindcast spanning January 1991 to December 2000. The initial two months (November-December) served as a model spin-up. Similarly, an initialization in November 2000 would run from January 2001 to December 2010, and an initialization in November 2010 would forecast the period from January 2011 to December 2020. Consequently, the forecast initialized in November 2018 spans from January 2019 to December 2028.

This study utilized daily precipitation (pr), maximum daily temperature (tasmax), and minimum daily temperature (tasmin) data from three CMIP6-DCPP models: BCC-CSM2-MR, MPI-ESM1-2-HR, and MRI-ESM2-0 (Table 1). Each model has a horizontal resolution of approximately 100 km. These CMIP6-DCPP models incorporate significant advancements in parameterizations, process representations, and feedback mechanisms, leading to improved simulations of key climate processes. For example, BCC-CSM2-MR demonstrates advancements in simulating tropospheric temperature, East Asian circulation patterns, and climate variability across timescales³⁶. MPI-ESM1-2-HR features enhanced atmospheric resolution, improving the mean state and regional dynamics, although long-standing biases are only modestly reduced³⁷. MRI-ESM2-0 better reproduces mean climate and interannual variability and reduces biases in near-surface temperature³⁸.

Table 1 List of CMIP6-DCPP models contributing to the Multi-Model ensemble (CMIP6-DCPP-MME), employed for the prediction of extreme climate events over Iran, along with the ECS and TCR ranges of the respective CMIP6-DCPP models.

Employing a multi-model ensemble that includes climate models with high climate sensitivities to increased CO₂ concentrations can introduce bias into predictions³⁹. To reduce this, we considered the Equilibrium Climate Sensitivity (ECS) and Transient Climate Response (TCR) values of the studied models, ensuring they are within the likely range of the IPCC-assessed. The best estimate for ECS lies within the likely range of 2.5–4 °C, while TCR is estimated at 1.4–2.2 °C (high confidence) as shown in Table 1. ECS and TCR are key metrics for quantifying the climate system’s sensitivity to external forcings⁴⁰. Therefore, using models with suitable ECS and TCR values is crucial for reducing uncertainties inherent in climate modeling and producing more reliable climate change predictions⁴¹.

Methodology

Bias correction

To address systematic biases in CMIP6-DCPP models for climate impact studies, we employed the Delta Change (DC) method as it effectively adjusts the mean of target variables while preserving observed spatiotemporal patterns^42,43,44,45. This approach offers key advantages, including computational simplicity, straightforward implementation, and dependence on reliable observational data (e.g., Iran Meteorological Organization’s long-term climatology records in this study). Since our research focuses on impact assessment and long-term climate extremes, we selected the DC method because, as Madani et al.⁴⁶ emphasize, it preserves temporal structures in mean, annual, and seasonal simulations. Additionally, it captures mean shifts while maintaining natural climate variability. Given our study’s emphasis on CMIP6-DCPP and internal climate variability, the DC method allows us to retain each model’s inherent natural variability while adjusting the mean climate signal. This ensures that extreme events, seasonal cycles, and multi-year oscillations remain physically plausible while minimizing systematic biases in mean values. Furthermore, we selected the DC method because alternative bias correction techniques prove less effective for key variables, including tasmax, tasmin, and precipitation⁴⁵, which are critical inputs for calculating climate extremes using the R package.

We applied the DC method to correct biases in daily precipitation, maximum temperature, and minimum temperature from CMIP6-DCPP models (100 km resolution), ensuring more accurate climate predictions. The bias correction method is applied as follows:

$$\:\:\:\:\:\:\:\:\:\:\:\:{P}_{contr}^{BC}\left(t\right)={P}_{obs}\left(t\right)$$

$$\:{P}_{frc}^{BC}\left(t\right)={P}_{obs}\left(t\right).\left[\frac{{\mu\:}_{m}{P}_{frc}\left(t\right)}{{\mu\:}_{m}{P}_{contr}\left(t\right)}\right]$$

(5)

In the above equation, P is Variable; contr is the DCCP-GCM simulated time series during the control period, obs is the observational time series, frc is the future forecast time series to be corrected, Bc is the final bias-corrected time series, t is the time step, and $\:{\mu\:}_{m}$ is the long-term mean⁴⁷.

Generate a Multi-Model ensemble

Uncertainty is a primary limitation in climate prediction. Due to the complexity and chaotic behavior of the climate system, it is inherently impossible to achieve perfect predictions. Nevertheless, utilizing a multi-model ensemble helps to address the limitations of climate forecast systems attributed to the signal-to-noise paradox⁴⁸. The multi-model ensemble can significantly enhance the accuracy of model predictions because the variation among ensemble members of a single model gives an estimate of uncertainty. For instance, using different realizations of a single DCPP model introduces uncertainty associated with internal variability. In contrast, multi-model ensembles, which incorporate simulations from various models, also account for the impact of structural differences between models.

Two primary methods exist for generating ensemble means: the unweighted and the weighted multi-model mean. The unweighted approach adheres to the “one vote per model” principle, assigning equal weight to all models. While this method is straightforward and widely used, its major drawback is its susceptibility to the influence of less reliable models, which can degrade the overall accuracy of the ensemble prediction. To address this limitation and enhance the reliability of our results, we opted for a weighted multi-model mean. This method assigns a specific weight to each model based on its performance in simulating observed climate conditions. By giving greater influence to higher-skilled models, this approach effectively mitigates the impact of poorer-performing models and improves the precision of the final climate projections⁴⁹.

In this DCPP study, we first combine 27 members from three different models using the multi-ensemble average (MEA) method. Our objective is to minimize unpredictable noise (natural variability) and maximize signal predictability in the multi-model ensemble^48,50,51. This approach could significantly improve the accuracy of climate predictions in our analysis. As supported by Perkins & Fischer⁵² and Li et al.⁵³, while single realizations may fail to capture the full range of climate variability, averaging multiple realizations (r1i1p1f1 through r9i1p1f1) within each model: (a) Captures internal climate variability arising from perturbed initial conditions, (b) Provides a more robust estimate of extreme event probabilities, (c) Ensures each model contributes equally to the ensemble before weighting.

Then, we applied the Independent Weighted Mean (IWM) method to generate a (weighted) multi-mode ensemble. The IWM approach treats each model as a single unit, regardless of realization count, and assigns weights based solely on model performance, thereby avoiding over-representation of any individual model. The IWM method assigns a specific weight to each model based on the Mean Squared difference (MSD), aiming to find the linear combination of an ensemble of model simulations that minimizes the MSD with observational data⁵⁴. Evans et al.⁵⁵, in their examination of the Bishop & Abramowitz⁵⁴ method, found that it effectively selects optimally independent models with a high accuracy of 87%, even in the presence of noise.

Our analysis reveals that the highest weights are assigned to the MPI-ESM1.2-HR, BCC-CSM2-MR, and MRI-ESM2.0 models, respectively. This weighting reflects their superior performance and reliability in capturing the observed climate variability, thereby contributing more significantly to the overall predictive accuracy of the ensemble.

Climate extreme indices

To analyze climate extremes, we employed seven indices developed by the Expert Team on Climate Change Detection and Indices (ET-SCI). Although relative thresholds enable local contextualization, the ET-SCI indices offer a more robust, standardized, and scientifically validated framework for investigating climate extremes. Their comprehensive methodology minimizes potential biases related to base period selection, ensuring consistent and comparable results across diverse regions and providing a more reliable tool for monitoring extreme climate events.

Specifically, the indices include the Number of days with heavy precipitation (R10mm), the Simple Precipitation Intensity Index (SDII), and the Standardized Precipitation Evapotranspiration Index with a 12-month accumulation period (SPEI-12), as well as heat wave frequency (HWF) and duration (HWD), and cold wave frequency (CWF), and duration (CWD) (see Table 2). In addition, we assessed the frequency and duration of heat waves and cold waves using the Excess Heat Factor (EHF) and Excess Cold Factor (ECF)⁵⁶. The characteristics of drought- including drought frequency (DF), duration (DD), severity (DS), and intensity (DI) – were evaluated using the Run Theory⁵⁷. Drought characteristics were derived from the SPEI-12 using a threshold value of ≤ −1, whereby a drought event is defined as a period of at least two consecutive months during which the SPEI-12 value remains below this threshold^58,59.

It is imperative to consider the seasonal cycle both for evaluating model performance and for integrating projected outcomes that are contingent on seasonal variations. Consequently, the R10mm and SDII indices were analyzed on both annual and seasonal scales, whereas the remaining climate extreme indices were primarily evaluated on an annual basis. A limitation of the present study is the absence of a seasonal analysis for drought characteristics (SPEI-12), heatwaves, and coldwaves. This limitation arises from the fact that (a) drought events typically span multiple months or even years, and (b) cold and heat waves—defined respectively by the ECF and EHF—are most appropriately examined within the confines of their corresponding cold and warm seasons.

This study employs a compound hazard framework to deliver a spatially explicit climate risk assessment for Iran. Compound hazards are defined as events where multiple climate extremes either co-occur or follow in close succession, creating risks greater than the sum of their individual impacts^60,61. While these hazards can manifest through various mechanisms—such as temporal cascades, spatial co-occurrence, or multivariate dependencies—our analysis is intentionally focused on spatially compounding hazards. We define these as the simultaneous or near-simultaneous occurrence of multiple extremes (e.g., heatwaves and droughts) within a specific geographical area.

The primary rationale for this focus is to identify regional hotspots of multi-hazard exposure, which is a critical first step for effective adaptation planning. This approach is particularly pertinent for Iran, a region known for the co-occurrence of diverse climate extremes. Although cascading interactions (e.g., a heatwave intensifying a drought) are an important subject for future research, they fall beyond the scope of this analysis. Consequently, to quantify the aggregate risk from co-occurring events, we apply an equal weighting scheme to all hazard components. This methodological choice aligns with our objective to map the cumulative risk from concurrent extremes, rather than to model their complex causal interdependencies.

Table 2 ET-SCI indices are used for the analysis of individual and compound extreme events, with drought characteristics defined using the run Theory.

Sensitivity analysis of Compound-Hazards

While differential weighting could be justified for hazards with disproportionate impacts, we apply equal weighting for transparency, particularly where empirical evidence does not support prioritizing one hazard over another⁶². Our analysis focuses on spatially compounding hazards, assessing aggregated risks from concurrent events (e.g., heatwaves and droughts). To explicitly test the influence of key methodological decisions on our results, we conducted sensitivity analysis focusing on two primary aspects: hazard weighting and the statistical definition of risk hotspots.

1. Sensitivity to Hazard Weighting Scenarios.

To evaluate the sensitivity of the multi-hazard assessment to the weighting scheme, the baseline equal-weight multi-hazard map was compared with four alternative dominated-scenario maps. In each alternative scenario, the relative weight assigned to one specific hazard group was doubled relative to the others, while keeping the total sum of ranking scores across all hazards constant to ensure comparability. The four weighting scenarios were defined as follows: (i) Drought-Dominated Scenario—emphasizing drought-related indices (DI, DS, DF, DD); (ii) Coldwave-Dominated Scenario—emphasizing cold-related indices (CWF, CWD); (iii) Heatwave-Dominated Scenario—emphasizing heat-related indices (HWF, HWD); and (iv) Precipitation-Dominated Scenario—emphasizing precipitation-related indices (R10mm, SDII).

2. Sensitivity Analysis of Hotspot Definition Thresholds.

To complement the weighting analysis and assess the robustness of the spatial results, we performed a sensitivity analysis to evaluate how risk hotspots respond to varying statistical confidence thresholds. The Getis–Ord Gi* statistic was applied⁶³ at three significance levels (90%, 95%, and 99%) to determine whether the spatial clustering of high-risk areas was statistically significant and to examine how the extent and configuration of these clusters changed with increasing threshold stringency.

The Getis–Ord Gi* method, a widely used tool in spatial statistics, identifies statistically significant spatial clusters of high (hot spots) and low (cold spots) values⁶⁴. For each grid cell, the statistic computes a z-score and a corresponding p-value, which together quantify the intensity and significance of local clustering. To ensure the robustness of the results and test the sensitivity of the multi-hazard assessment, a stringent 99% confidence level was adopted. Only clusters with p-values below 0.01 were classified as statistically significant hot spots, corresponding to less than a 1% probability of random occurrence.

The integration of hazard indicators followed a three-step workflow: (i) equal weights were assigned to all climate extreme indices; (ii) each index was reclassified into ordinal risk categories ranging from Low to Very High; and (iii) the weighted categories were integrated using a spatial overlay procedure. The resulting composite multi-hazard map delineates regions susceptible to concurrent climate-related hazards, classified into five distinct risk levels.

Results and discussion

Monitoring and predicting Climate-related hazards: extreme precipitation

We investigate two indices, R10mm and SDII to examine extreme precipitation across Iran for both the hindcast period (1991–2019) and the forecast period (2023–2028), as illustrated in Fig. 2. The results show that the southern coast of the Caspian Sea and the mountainous regions of the Zagros are major hotspots of heavy precipitation events. Regarding the large amounts of precipitation received by the southern coast of the Caspian Sea, this area is recognized as a critical hotspot for heavy precipitation events across Iran. The southern coast of the Caspian Sea is a unique geographical location. It is proximate to the Caspian Sea, which has considerable extent along the latitudes and is located in the path of frequently passing extratropical cyclones and anticyclones. Hence, it has favorable conditions for extreme precipitation events throughout the year^65,66. Unlike most Middle Eastern regions, the southern coast of the Caspian Sea is influenced by the passage of extratropical anticyclones from the north, and the consequent moist northerly flows resulted in heavy precipitation over this region⁶⁶. In contrast, the heavy precipitation events in the west and southwest of Iran and the mountainous Zagros region are primarily due to the passage of Mediterranean cyclones and non-frontal Sudanese low-pressure systems, and they are limited to the cold periods of the year^67,68.

It is crucial to note that the majority of precipitation in southwestern Iran during the cold season is generated by the Sudanese low-pressure systems^{65,66,69,70,71}. These non-frontal systems frequently coincide with the strengthening of the subtropical jet stream and the formation of an atmospheric river^{72,73,74,75,76}, resulting in widespread and heavy precipitation across the Middle East due to the transportation of significant moisture from the warm southern seas to the region. Over the past four decades, research has demonstrated a discernible increase in both the spatial extent and activity of Sudanese low over Iran⁷¹. Consequently, it can be inferred that the positive anomaly in heavy precipitation and the increased intensity of daily rainfall in southwestern Iran in the near future are attributable to the heightened frequency of Sudanese lows during the cold season. In the east-northeastern parts of the country, the primary mechanism responsible for heavy precipitation is also initially Sudanese low-pressure systems. This region, when the mid-tropospheric trough in the eastern Mediterranean extends to lower latitudes in the central Red Sea, concurrently facilitates the development of the Sudanese low as a dynamical low-pressure system and the transfer of moisture from the warm southern seas (the Sea of Oman, the Arabian Sea, the Red Sea, and the Persian Gulf) into Iran. The establishment of a ridge in the mid-troposphere, along with an anticyclonic circulation in the lower troposphere over the Arabian Sea and the eastern Arabian Peninsula⁷⁷, plays a pivotal role in supplying moisture sources for heavy cold-season precipitation in Iran’s interior regions^70,78. The positive anomaly in heavy precipitation in the southwestern and eastern regions of Iran suggests a high frequency of occurrence of the aforementioned mechanism in the near future.

The comparison between hindcast and observation suggests that the DCPP-MME has successfully captured the overall spatial distribution of R10mm. However, notable differences are evident in the R10mm estimations between observation and DCPP-MME across various geographical regions of Iran (Fig. 2). For instance, while the observed maximum R10mm is approximately 39 days in the southwestern Caspian Sea, DCPP-MME values do not exceed 21 days, as illustrated in Fig. 2. A similar difference is also observed between the western and southern regions of Iran.

Fig. 2

Changes in Precipitation extreme indices (R10mm, SDII) in observation and hindcast (1991–2019), forecast (2023–2028), and anomaly (difference between hindcast and forecast period) in the top panels. The country-wide averaged interannual variability of the two precipitation indices is shown in the graphs below. In the graphs, strong and very strong El Niño years are represented by red columns, while La Niña years are shown with light green columns. Units: (R10mm: days/year), (SDII: mm/day). The data were obtained from the Earth System Grid Federation (ESGF). URL link: https://esgf-metagrid.cloud.dkrz.de/search/cmip6-dkrz/. CMIP6-DCPP data were processed using Climate Data Operators (CDO), Version number: 2.0.4, under Ubuntu 22.04 LTS. URL link: https://code.mpimet.mpg.de/projects/cdo. The SDII and R10mm indices were calculated using the Climpact2 package. URL link: https://github.com/ARCCSS-extremes/climpact2. The maps and graphs were generated using ArcGIS Desktop (Release 10, Version number: 10.8.2) by Esri, Redlands, CA: Environmental Systems Research Institute. URL link: https://www.esri.com.

The results reveal that the DCPP-MME has failed to provide exact estimates of the number of days with heavy precipitation during the hindcast period. The distinctive geographical characteristics of the southern coasts of the Caspian Sea and the complicated interactions between sea, land, and terrain along the narrow coastal strip have increased the biases of DCPP-MME in estimating precipitation values. Peer-reviewed studies using different datasets such as the agro-meteorological dataset retrieved from ERA5 reanalysis (AgERA5)⁷⁹, the Satellite Integrated Multi-Satellite Retrievals for GPM (IMERG)⁸⁰, the multi-model ensemble of NEX-GDDP-CMIP6¹⁶, and the multiple datasets based on reanalysis, satellite, and bias-correction⁸¹, have found that the precipitation amounts are generally underestimated across the southern coast of the Caspian Sea. Notably, although the DCPP-MME underestimates R10mm on the southern coasts of Iran, it outperforms precipitation across the mountainous Zagros region. The mountainous Zagros region represents the second major hotspot for heavy precipitation events in Iran. The orographic effect, along with synoptic systems such as Sudanese and Mediterranean low-pressure systems^67,68,82,83, play an important role in producing considerable amounts of precipitation across this region. As illustrated in Fig. 2, the observed maximum R10mm is 21.8 days in the southwest of Iran compared to the 15.6 days estimated by DCPP-MME. However, the comparison of the spatial pattern of the R10mm between hindcast and observational data in the mountainous Zagros region reveals a notable limitation. The lack of sufficient stations with long-term observational data (1991–2019) in this region precludes the consideration of the spatial distribution and the observed frequency of heavy precipitation days as reliable benchmarks for comparison. The DCPP-MME has appropriately estimated R10mm in the arid regions of eastern and central Iran, where heavy precipitation is infrequent. The maximum values of R10mm for DCPP-MME and observation are 0.18 and 3.2 days, respectively. Moreover, the considerable distance of arid lands from moisture sources and being surrounded by the Zagros and Alborz Mountain ranges resulted in low precipitation over the year⁸³. Whereas a considerable portion of southeastern Iran’s precipitation occurs during the summer due to local mechanisms or the influence of low-pressure systems associated with the Indian summer monsoon⁸⁴.

Regarding the substantial importance of precipitation intensity in flooding across arid and semi-arid areas, we utilized the SDII to investigate the intensity of precipitation. The analysis of SDII indicates that the maximum SDII is typically observed on the southern coasts of the Caspian Sea, the Zagros mountains, and southwestern Iran, which have a high potential risk of flooding over the year¹⁶.

Over the southern coasts of the Caspian Sea, the hindcast DCPP-MME of SDII has failed to capture the precipitation intensity compared to observation, similar to R10mm (Fig. 2). The observed maximum SDII is 15.7 mm/day, while the DCPP-MME underestimates maximum SDII values by 8.1 mm/day. Nevertheless, the difference between DCPP-MME and observation in the arid southeastern regions and the mountainous Zagros is less than 2 mm/day and 1 mm/day, respectively, demonstrating that DCPP-MME outperforms across large parts of Iran. Notably, the southeast of Iran experiences high SDII, which is due to the limited number of rainy days in this region. In fact, despite the limited number of rainy days in southeastern Iran, this region experiences significant precipitation amounts caused by the moisture-holding capacity of warmer air and the influence of the Indian Summer Monsoon (ISM) system during the warm period of the year.

Global warming is expected to intensify the hydrological cycle, which will likely result in the increased intensity of extreme precipitation events and flooding risks in the future. As observed in Fig. 2, our results show an increasing pattern of the R10mm and SDII during the near-term across large parts of Iran. In this case, the overall patterns of R10mm anomalies indicate that the arid regions of northeastern, southwestern, and western Iran (the mountainous Zagros region) will experience the maximum positive anomalies of R10mm up to 1.3 days per year. In this case, the flooding in May 2024 in Mashhad, a city in northeastern Iran, represents a consequence of an unprecedented extreme precipitation event in the area. On May 15th, this event lasted for an hour and resulted in 38.3 millimeters of precipitation at Mashhad station. Importantly, this event resulted in significant human and property damage to residents and facilities in the area. Global warming is expected to increase evaporation rates and the atmosphere’s capacity to hold water vapor, consequently leading to higher precipitation values according to the Clausius-Clapeyron equation. Therefore, an increase in both the frequency and intensity of precipitation is likely to occur in various global regions, and Iran. It will increase the risk of precipitation-related hazards. The increase in the frequency and intensity of precipitation appears to be related to global warming across large parts of Iran from the west to the southwest. In contrast, the southern coasts of the Caspian Sea will experience a slight decrease in R10mm, from 1.5 days in the east to 3.3 days in the west, according to the findings of Zarrin et al.⁸⁵. Hence, we will likely witness a decrease of R10mm in the north and west. Based on DCPP-MME, the northwest of Iran is also expected to experience negative anomalies of R10mm at a lower rate, aligning with the recent increase in the west, particularly in southwestern Iran, during the near-term.

In addition to investigating R10mm, it is imperative to predict the SDII and identify its primary hotspots in the near-term, given the high sensitivity of SDII to anthropogenic activities⁸⁶. The analysis of SDII reveals a similar spatial pattern between the R10mm and SDII in terms of anomaly distribution during the forecast period. Therefore, the maximum anomaly regions of SDII follow the exact pattern of the maximum anomalies of R10mm. In the near-term, the mountainous Zagros and southwest region are expected to experience a positive anomaly of SDII, with an increase ranging from 2 to 6 mm/day. In addition, the SDII in the northeast of Iran indicates a positive anomaly, with an expected increase of up to 1.6 mm/day. In contrast, the SDII is predicted to decrease by 3.3 mm/day in southeastern Iran.

A comparative analysis of the R10mm and SDII reveals that the regions experiencing negative anomalies have different spatial patterns during the forecast period. The southern coast of the Caspian Sea will experience a significant decrease in R10mm, and the southeast of Iran shows a noticeable decrease in SDII. These results are in agreement with the findings of Zarrin & Dadashi-Roudbari¹⁶. A future decline in heavy precipitation and daily rainfall intensity along the southern coasts of the Caspian Sea is expected, linked to a reduced frequency of northerly migratory extratropical anticyclones. Extensive research supports the role of these anticyclones, which induce heavy precipitation in the southern coasts of the Caspian Sea via northerly winds^65,66,69,70.

To provide more details on the decadal prediction of extreme precipitation in Iran, we examine the inter-annual variation of R10mm and SDII indices from 1991 to 2028 (Fig. 2). To illustrate the inter-annual variations of precipitation, we identify the years marked by very strong El Niño and La Niña events using the Niño 3.4 index, as observed in Fig. 2. The El Niño-Southern Oscillation (ENSO) plays a crucial role in influencing the variability of annual precipitation in Iran^87,88. Utilizing the ENSO index enables the analysis of extreme precipitation anomalies and inter-annual variations.

According to the graphs in Fig. 2, the R10mm and SDII indices have exhibited significant inter-annual variations during the hindcast (1991–2019) and forecast (2023–2028) periods. Notably, some of the inter-annual variations in both indices are consistent with El Niño and La Niña events. The dominance of ENSO’s warm/cold phases from 1991 to 2020 has resulted in larger/smaller values than normal for both indices, respectively. Nevertheless, there is no notable consistency between the warm and cold phases of ENSO and the R10mm and SDII indices in most of the years. Hence, regarding the inter-annual variability of extreme precipitation indices (R10mm & SDII), it is not possible to conclusively state that the occurrences of floods and droughts and the variations of these indices in Iran correspond with the El Niño or La Niña events.

For instance, the maximum values of the R10mm index exceeded the mean values in only three years (1992, 1997, 2016), while the SDII index exceeded the mean values in only two years (1997, 2016). According to previous studies, the ENSO signal accounts for approximately 26% of the variance in annual precipitation in Iran. Due to Iran’s varied climate, the impact of the ENSO signal differs across the different geographical regions during ENSO warm and cold phases^89,90. Moreover, extreme El Niño/La Niña events are not consistently associated with occurrences of extreme floods or droughts⁹¹, nor do they correspond to the highest or the lowest values of extreme precipitation indices. As reported by Li et al.⁹² and Lopez et al.⁹³, the El Niño and La Niña events have intensified in the twenty-first century compared to the twentieth century. Climate projections also suggest that extreme El Niño and La Niña events may become more frequent by the end of the 21 st century in response to global warming⁹⁴. Consequently, ENSO’s influence is of significant importance for predicting extreme precipitation events in Iran’s future. In this case, 2023, which stands as the forecast period in this research, has experienced a nearly strong El Niño event.

Despite predictions of above-normal precipitation for Iran’s fall and winter by world-renowned seasonal forecast centers, including IRI, GFS-V2, and ECMWF, the actual precipitation did not reach the normal records based on the gauges. As shown in Fig. 2, the R10mm and SDII indices were higher than the long-term annual mean (1991–2019) in 2023. The observed increase in the extreme precipitation indices (R10mm & SDII) in 2023 resulted from increased spring precipitation based on most gauge records⁹⁵. Overall, Iran is expected to experience a decrease in extreme precipitation indices during the years 2023 to 2028.

Monitoring and predicting Climate-related hazards: drought analysis

Over the recent decades, Iran has been identified as a country suffering from drought due to the increase (decrease) in dry (wet) periods⁹⁶. The drought has significantly affected water resources⁹⁷, agricultural and economic sectors^98,99, and even national security in Iran over the years. Drought, along with floods, is recognized as one of the most significant climate-related hazards, which has caused widespread and unavoidable losses and damages in Iran. This study employed the 12-month Standardized Precipitation-Evapotranspiration Index (SPEI-12) to quantify drought characteristics, including frequency (DF), duration (DD), severity (DS), and intensity (DI) across Iran.

Comparative analysis of drought indices between the hindcast and forecast periods reveals a projected significant increase in DF across all Iranian regions in the near future (Fig. 3). This augmentation is particularly pronounced in the central, eastern, and southeastern regions, as well as portions of the northwest, where DF is anticipated to exceed 42% during the forecast period. Examination of the DF anomaly map demonstrates that, with the exception of the southwest and a localized area within the eastern Zagros Mountains, which are projected to experience wetter conditions, the remainder of Iran will exhibit a positive anomaly of 12% or greater. The maximum DF anomaly, registering a 31% increase, is observed in southeastern Iran.

Conversely, the drought duration (DD), severity (DS), and intensity (DI) indices exhibit a contrasting trend when comparing the hindcast and forecast periods across Iran. Specifically, while a widespread increase in drought frequency is projected, DD, DS, and DI are anticipated to decrease significantly in the near future (Fig. 3). This suggests that droughts in the coming decade in Iran will be characterized by reduced severity and shorter durations, albeit with increased frequency. The magnitude of the negative anomaly observed in DD, DS, and DI corroborates this finding (Fig. 3). However, a region extending from southern Iran to the eastern border displays a positive anomaly in all drought indices. This area, along with a limited region in the northwest along the Iran-Turkey border, is projected to experience a positive anomaly in all drought indices in the near future.

When the frequency of drought events increases in the future, the duration and intensity of individual droughts will decline in the near future. It may be due to the changes in the hydrological cycle and the distribution of water resources. Higher drought frequency results in shorter intervals between dry periods and reduces the opportunity for hydrological systems to fully recover during wet phases¹⁰⁰. Consequently, the cumulative stress on water reserves, soil moisture, and ecosystems constrains the potential severity of individual drought events¹⁰¹. Furthermore, accelerated evaporation rates associated with global warming contribute to the rapid onset and cessation of droughts while simultaneously mitigating their intensity¹⁰².

The reduction in drought intensity is also attributable to the diminished accumulation of long-term stress within natural systems. Shorter and more frequent droughts lack the extended timeframes necessary to fully deplete critical resources over consecutive years, as is typically observed in prolonged droughts¹⁰³. It is important to note that the specific effects of increased drought frequency are contingent upon regional climatic and environmental conditions, rendering the relationship between duration and intensity subject to spatial variability and broader climate trends.

Fig. 3

Drought characteristics derived from the SPEI-12 using a threshold of ≤ −1. Panels display drought frequency (DF), duration (DD), severity (DS), and intensity (DI) for the DCPP-MME model hindcast (1991–2019) and forecast (2023–2028) periods. Lower panels illustrate anomalies for the forecast period. Units: (DF: %), (DD: month), (DS & DI: SPEI). The data were obtained from the Earth System Grid Federation (ESGF). URL link: https://esgf-metagrid.cloud.dkrz.de/search/cmip6-dkrz/. CMIP6-DCPP data were processed using Climate Data Operators (CDO), Version number: 2.0.4, under Ubuntu 22.04 LTS. URL link: https://code.mpimet.mpg.de/projects/cdo. The SPEI-12 was computed using the Climpact2 package. URL link: https://github.com/ARCCSS-extremes/climpact2. Drought characteristics were computed using Jupyter Notebook under Anaconda3, Version number: 2.14.1. The maps were generated using ArcGIS Desktop (Release 10, Version number:10.8.2) by Esri, Redlands, CA: Environmental Systems Research Institute. URL link: https://www.esri.com.

Monitoring and predicting Climate-related hazards: hot extreme temperatures

Heat waves are among the most critical extreme temperature events, substantially impacting human life. Since heat waves are recognized as threats to society, the environment, and the economy, their study is pivotal^104,105,106. A variety of indices have been developed to study and analyze the characteristics of heat waves. In this study, we focused on the frequency and duration of heat waves, which have been frequently used in many studies. Accordingly, we analyzed the HWF and HWD in the hindcast and near future periods, as illustrated in Fig. 4. Comparing the HWD observation and hindcast distribution, there is a similar spatial pattern and consistency in maximum and minimum values. However, the DCPP-MME fails to capture the maximum values of HWD on the southern coasts of the Caspian Sea, similar to its performance on extreme precipitation indices (R10mm and SDII). In the Caspian Sea regions, the DCPP-MME has estimated the maximum values of HWD to be between 4.4 and 5.6 days, while the observed values range from 7.2 to 8.2 days. The maximum values of HWD exhibit small biases in arid central and semi-arid eastern Iran during the hindcast period compared to observations. DCPP-MME has appropriately estimated the HWD in these regions, with a difference of just 0.85 days more than the observed values. Notably, Jangi et al.¹⁰⁷ have recently demonstrated a similar spatial pattern of heat waves across Iran. They also found that the arid central and semi-arid eastern regions experience the maximum values of HWD. During the hindcast period, the Dasht-e Kavir is also identified as an HWD hotspot due to the higher resolution of the DCPP-MME grid points. However, it is not seen in the observation map due to the lack of ground stations in the central arid regions of Iran.

The heat wave frequency (HWF) distribution in the hindcast period (Fig. 4) is not fully captured by the DCPP-MME, which is similar to HWD. For instance, the observed maximum HWF includes parts of eastern and southeastern Iran, ranging between 8.7 and 10.03 days. In contrast, DCPP-MME in the hindcast period depicts a specific pattern with higher HWF values. The entire arid regions of central and eastern Iran experience a maximum HWF of 12.4 to 13.9 days. In conclusion, the arid regions of central and eastern Iran have the highest values ​​of HWD and HWF. During the warm period of the year, these regions exhibit significant potential for higher temperatures and heat wave events, attributed to the influence of both local and large-scale dynamical forcings. The temperature increase in the arid regions of central Iran is significantly influenced by high long-wave radiation, lack of cloud cover and vegetation, distinctive physical characteristics, and soil types¹⁰⁸.

In addition to the local geographical characteristics that cause significant temperature rises and heat wave occurrences in central Iran’s arid regions during summer, researchers also attribute the influence of large-scale forcing by the South Asian summer monsoon system. For example, Walker¹⁰⁹ associates the higher temperatures in this region with the dominance of the tropical easterly jet and adiabatic descent in its exit area. The impact of the above factors leads to experiencing the Earth’s highest temperature in the central regions of Iran, particularly the Dasht-e Lut. Recent findings based on satellite data have confirmed this^9,10,11,12. Satellite-based findings indicate that the world’s highest recorded land surface temperature (LST) has been observed in Dasht-e-Lut, exceeding 80.8 °C. The main factors causing the highest temperatures in this region include: (1) The mountain-surrounding terrain, which leads to limited air movement and increased heat-trapping¹². (2) The blackened sand land cover causes the absorption of significant amounts of solar energy¹¹⁰. (3) The dominance and intensification of Iran’s subtropical anticyclone during the warm period of the year. Due to the adiabatic descent in the atmosphere, surface temperatures increased significantly^4,111,112.

The escalating frequency and duration of heat waves across Iran, observed over the past decade and projected for the near future, are primarily driven by the intensification of the mid-tropospheric anticyclone dominating the region. Trend analyses reveal a significant increase in both the frequency and central geopotential height of this subtropical anticyclone during the warm season, as evidenced by historical data¹¹³ and future projections¹¹⁴. Importantly, the anticyclone’s latitudinal stability¹¹², combined with its intensification, enhances atmospheric subsidence. This, in turn, leads to elevated surface temperatures and reduced precipitation, directly exacerbating the drought conditions discussed earlier. The resulting increase in evapotranspiration further depletes soil moisture across the Iranian plateau, compounding the water stress already intensified by less time for hydrological replenishment between droughts.

Furthermore, the observed decline in snow days, snow depth, and snow cover across the Zagros, Azerbaijan, and Alborz Mountain ranges^{115,116,117,118,119}, coupled with a reduction in surface albedo, contributes to a sustained rise in land surface temperature and a decrease in soil moisture. This trend, corroborated by studies highlighting elevation-dependent warming (EDW)^{117,120,121,122,123,124}, creates a positive feedback loop, amplifying the effects of the intensifying anticyclone. The synergistic effect of these atmospheric and cryospheric changes, creating a more stable and warmer atmosphere, intensifies the duration, frequency, and intensity of heat waves over Iran and the broader Middle East^113,125,126. Consequently, the projected persistence of these mechanisms suggests that heat wave activity will likely continue to intensify across a significant portion of Iran, further exacerbating drought conditions and creating a cascade of interconnected extreme atmospheric events.

In recent decades, heat waves have become a major environmental challenge for society, with notable increases in their frequency, intensity, and duration, particularly in tropical and subtropical regions. This has led to an increase in heat-related mortality in these regions^{106,107,127,128}. Furthermore, the projected rise in temperatures in the coming decades increases the concerns about the future of more frequent and severe heat waves and their associated impact^12,129. In the forecast period, the analysis of HWD reveals that the maximum values are found in the arid central and eastern parts of Iran, consistent with their identification as critical heat wave hotspots in the hindcast map (Fig. 4). These regions are predicted to experience an increase of 2.9 days in HWD compared to the hindcast period. The HWD in Dasht-e-Lut is predicted to increase by 3.4 days compared to the past three decades (1991–2019).

Fig. 4

The annual mean of temperature extreme events for heat wave duration (HWD) and heat wave frequency (HWF) is demonstrated for observation data and DCPP-CMIP6 hindcast (1991–2019), forecast (2023–2028), and anomaly (difference between hindcast and forecast period). Units: (HWD, HWF): days/year. Below is the country-wide interannual variability of HWD and HWF. In the graphs, strong and very strong El Niño years are represented by red columns, while La Niña years are shown with light green columns. The data were obtained from the Earth System Grid Federation (ESGF). URL link: https://esgf-metagrid.cloud.dkrz.de/search/cmip6-dkrz/. CMIP6-DCPP data were processed using Climate Data Operators (CDO), Version number: 2.0.4, under Ubuntu 22.04 LTS. URL link: https://code.mpimet.mpg.de/projects/cdo. The HWD and HWF indices were calculated using the Climpact2 package. URL link: https://github.com/ARCCSS-extremes/climpact2. The maps and graphs were generated using ArcGIS Desktop (Release 10, Version number: 10.8.2) by Esri, Redlands, CA: Environmental Systems Research Institute. URL link: https://www.esri.com.

Another hotspot of HWD is seen in the mountainous regions of Iran, including central Zagros and the northwest. The maximum HWD values in these regions are predicted to rise by 4.06 days. This result is in agreement with the findings of Fallah-Ghalhari et al.¹³⁰ and Usta¹³¹. The minimum values of HWD in the forecast period are observed in the southern coasts of Iran, which is in line with Abbasnia’s 2019 findings¹³². Overall, DCPP-MME for the near-term predicts a significant increase in HWD across all regions of Iran, as observed in Fig. 4. The near-term analysis of HWF indicates a distinct pattern, with the maximum values expanding across the entire western regions of Iran, including parts of the Zagros Mountains. This pattern extends along the Zagros mountains to southeastern Iran with a low frequency of heat waves.

Overall, a significant increase in HWF will be observed throughout Iran in the near future. During the forecast period, it is predicted that the HWF will exceed 20 days in western Iran, representing an anomaly of more than 15 days. The forecasted significant positive anomaly of HWF is so large that it makes it incomparable with the hindcast period.

The near-term increase in HWF across the mountainous regions of northwestern and western Iran may be attributed to Elevation-Dependent Warming (EDW). The rising temperatures in mountainous regions have been confirmed globally in recent years, caused by various reasons such as snow albedo feedbacks^133,134, changes in water vapor and latent heat release^135,136; changes in radiative flux, decrease in land surface temperature, and temperature changes and related aerosol effects^120,137. The near-term analysis suggests that the minimum values of HWF will also increase on the southern coast and parts of the central regions of Iran, with an increase ranging from 3.4 to 5.2 days compared to the hindcast. The results reveal that despite the arid central regions of Iran experiencing the lowest HWF in the near-term, the HWD is significantly predicted to increase in these regions. During the warm period of the year, the increase in HWD is likely due to adiabatic warming, which is associated with intensified anticyclone activities¹²⁵.

In addition to analyzing the spatial distribution of HWF and HWD, we also examined their inter-annual variations, as shown in Fig. 4. The results indicate that the HWD and HWF have significantly increased in Iran since 2016. The year 2016 has stood as one of the most exceptional years for climate anomalies since the beginning of climate record-keeping in the 1880s. Not only did it record the highest global average surface temperatures, but it also experienced prolonged and intense heat waves across various regions of the Earth, which may be associated with the warm phase of the Southern Oscillation (El Niño). Consequently, the heat waves of 2016 have left their impact on Russia, parts of southwestern Asia¹³⁸, and parts of North America, leading to widespread wildfires. According to the DCPP-MME, the HWD and HWF values of 2016 were 9.7 and 34.7 days, respectively, exceeding the mean values (as shown in Fig. 4, lower panels). In late July 2017, another significant heat wave event occurred in Ahvaz, a city in southwestern Iran. During this event, temperatures exceeded 54 °C on July 28th, resulting in extensive damage, including fires in electrical equipment, deformations of railway tracks, and the rapid desiccation of trees. According to the near-term DCPP-MME, the year 2023 has exhibited the maximum HWD and HWF indices. The Copernicus Climate Change Service has reported high temperatures for 2023. According to its report, the year 2023 was recognized as one of the strongest El Niño years, with the global mean temperature experiencing a positive anomaly of approximately 1.5°C¹³⁹. Furthermore, 2023 has been identified as the warmest year on Earth in recent decades, comparable to 2016. In the year 2023, Iran also experienced a warm year due to the influence of El Niño.

The DCPP-MME results are consistent with this, and the HWD and HWF values reach 7.7 and 20.6 days, respectively. Overall, the inter-annual variations of heat waves indicate a noticeable increase in HWF in Iran, with the increase expected to rise from 9.2 to 20.6 days/year during the coming decade. Similarly, the HWD is predicted to increase by 6.01 to 8.9 days yearly. Given Iran’s location in the arid and semi-arid zones, the increasing heat wave events necessitate the implementation of appropriate climate-related policies. It is imperative to consider strategic plans and efficient actions to combat the consequences of heat waves on public health, water resources, agriculture, and energy sectors.

Monitoring and predicting Climate-related hazards: cold extreme temperatures

A cold wave is an extreme temperature event characterized by a significant, rapid decline in temperature caused by the influx of extremely cold air into the region¹⁴⁰. This event may pose a severe threat to human health, especially among vulnerable populations¹⁴¹. The influence of Cold waves on subtropical and mid-latitude regions is short-term¹⁴². However, they are also extensively recognized as a critical factor in cold-related mortality, causing significant impacts in the short term¹⁴³. We employed two indices, CWF and CWD, to study cold wave events in Iran. Many studies have utilized these indices to determine the characteristics of cold waves^144,145.

Figure 5 presents the CWD and CWF indices for the hindcast period (1991–2019) and the forecast period (2023–2028). According to the hindcast analysis of CWD, cold waves continue for 5.2 to 11.8 days in different regions of Iran, with the most extended durations occurring in the northwest, west, and southeast. However, the comparison between the hindcast and the observation reveals that the spatial distribution of CWD is not entirely similar, as illustrated in Fig. 5. For instance, while northwestern Iran experiences the minimum CWD during the observation period (ranging from 5.4 to 8.2 days), it is the hotspot of maximum CWD in the hindcast period (between 9.2 and 10.3 days). The reasons that station-based observation has not captured the maximum CWD in the northwest region may be attributed to the following: Firstly, the lack of sufficient stations with long-term observational data in mountainous terrain and the high altitude of the northwest. Secondly, many of the stations used in this research are located at altitudes between 70 and 1400 m. Recently, Sarhan et al.¹²³ conducted a comprehensive analysis of Iran’s cold spots, employing daily Aqua/MODIS LST products with a 1-km resolution. They indicated that regions in northwestern Iran, particularly regions along the border of Iran and Turkey, are among the coldest regions of the country, experiencing significantly low temperatures during the cold periods of the year. Given these reasons, we can state that the DCPP-MME performed rather well in capturing CWD compared to the station-based observations. The observed low temperatures and very cold periods in northwestern Iran are attributed to the influence of the extratropical anticyclone, the Siberian high, and the dominance of cut-off lows over this region. The persistence of these systems leads to the transport of cold air and northerly flows to northwestern Iran^{146,147,148,149}.

The results indicate that in addition to northwestern Iran, the maximum CWD values are observed in the mountainous regions of the Zagros, the semi-arid regions of the northeast, and southeastern Iran, ranging from 8.4 to 10.3 days. The minimum values of CWD are also found in the arid central regions during the hindcast period, with a duration between 6.5 and 7.1 days. The analysis of CWF reveals a discrepancy between the observation and the hindcast periods, similar to what was observed within the CWD index (Fig. 5). The maximum CWF values in the hindcast period are generally much larger than the observation across Iran. Based on the observations, the northwestern-western regions of Iran experience the minimum CWF, ranging from 6.8 to 10.3 days. Whereas, the DCPP-MME estimates that the minimum values of CWF in Iran’s arid central regions will range from 11.5 to 12.7 days. In contrast, the CWF hotspot in the hindcast period has extended from the northwest mountains to the southeast regions, with values between 14.6 and 16.4 days. Moreover, the eastern and northeastern regions of Iran serve as secondary hotspots for CWF, with frequencies ranging from 13.1 to 14.3 days. Alijahan et al.¹⁵⁰ suggested that the occurrence of cold waves in northeastern Iran is caused by the extension of the Siberian high and extratropical anticyclones, expanding northerly flows from north of the Caspian Sea to northeastern Iran. The formation of extreme cold periods and cold waves in Iran is also attributed to the influence of the polar vortex, its movement in the Eastern hemisphere, and the consequent blocking system formation over Europe and western Russia^146,151.

The near-term analysis of CWD indicates that Iran will observe significant changes in the cold wave duration compared to the hindcast period. The CWD values are expected to decrease from 1.2 to 4.4 days across the northwest and Zagros mountainous regions. The decline in the CWD may be attributed to an increase in EDW in these regions, which has been identified as a consequence of global warming in recent years.

Fig. 5

The annual average of temperature extreme events for Cold Wave duration (CWD) and cold wave frequency (CWF) is demonstrated for the observation period and DCPP-CMIP6 hindcast (1991–2019), forecast (2023–2028), and anomaly (difference between hindcast and forecast period) in the top and middle panels. Unit (CWD, CWF): days/year. Below is the country-wide interannual variability of CWD and CWF. In the graphs, strong and very strong El Niño years are represented by red columns, while La Niña years are shown with light green columns.

The data were obtained from the Earth System Grid Federation (ESGF). URL link: https://esgf-metagrid.cloud.dkrz.de/search/cmip6-dkrz/. CMIP6-DCPP data were processed using Climate Data Operators (CDO), Version number: 2.0.4, under Ubuntu 22.04 LTS. URL link: https://code.mpimet.mpg.de/projects/cdo. The CWD and CWF indices were calculated using the Climpact2 package. URL link: https://github.com/ARCCSS-extremes/climpact2. The maps and graphs were generated using ArcGIS Desktop (Release 10, Version number: 10.8.2) by Esri, Redlands, CA: Environmental Systems Research Institute. URL link: https://www.esri.com.

This result is in agreement with the findings of Díaz-Poso et al.¹⁴⁵ for mountainous regions. According to their projections, the decrease in Cold Wave Days (CWD) and Cold Wave Frequency (CWF) in the mountainous areas of the Iberian Peninsula is due to the increase in the EDW. As global warming continues to rise, the complex interactions of albedo, cloud cover, water vapor, and radiative fluxes in mountainous regions accelerate the rate of temperature rise. In contrast, near-term results indicate that, despite the increase in global temperatures, an increase in CWD is predicted for northeastern Iran. We will witness an increase in CWD by 3.6 days in the forecast period relative to the hindcast period. Also, it is expected that the Persian Gulf coasts and large parts of the arid central regions of Iran will experience an increase in CWD by 1.4 to 3.6 days compared to the hindcast period.

The near-term analysis of the CWF suggests that the cold wave frequency is predicted to decrease extensively and considerably in Iran, as depicted in Fig. 5. The northwest and west of Iran exhibit the maximum negative anomaly of CWF, which is similar to the CWD index. A considerable decrease in CWF is predicted in the west of Iran, with a decrease of 3.3 to 9.1 days compared to the hindcast period. In contrast, northeastern Iran will experience a positive anomaly of CWF in the near-term, similar to the CWD index. The maximum positive anomaly of CWF is expected to occur in 3.4 days along the border of Iran and Turkmenistan. This indicates that, despite a decrease in cold waves caused by global warming, cold events are likely to continue in some regions, posing localized threats^145,152. In January 2023, a severe cold wave in Iran’s northeastern regions resulted in serious damage to various economic sectors. The occurrence of this event alerts us that, despite a significant global decline in cold waves, it is possible to experience a single extensive and severe cold wave in the coming decades^{144,145,153,154}.

The hindcast inter-annual variations of the CWF and CWD indicate that Iran has experienced many cold wave events. As illustrated in Fig. 5, a significant cold wave occurred in January 2008 during a very strong La Niña event. The temperatures dropped below − 32 °C in Iran as reported by WMO in 2008¹⁵⁵. In the forecast period, an extreme cold wave is accurately predicted by DCPP-MME for January 2023 across large parts of northeastern Iran. In January 2023, northeastern Iran experienced a sudden temperature drop due to an extreme cold wave from Siberia, which rapidly spread to the central parts of the country. Temperatures fell below 0 °C in more than 400 cities, reaching as low as −28.5 °C in Fariman, a city in northeastern Iran. This cold event occurred during a strong El Niño event. This indicates that Iran’s cold waves result not only from La Niña events but also from El Niño phases, such as the one in February 2024 that affected large areas of the country. In general, the inter-annual variations of the CWF and CWD suggest that the frequency and duration of cold waves are expected to decrease in the near-term in response to global warming.

Evaluation of monthly and seasonal cycles in DCPP models

To assess the predictive capability of DCPP models regarding temperature and precipitation extremes in Iran, we evaluated their representation of monthly and seasonal cycles. Specifically, we compared model outputs of monthly temperature and total precipitation with observational data derived from the averaged measurements of 70 meteorological stations across Iran. Analysis of the observational data from 1991 to 2019 reveals a distinct precipitation regime, characterized by widespread precipitation events from October to May, with peak precipitation in January and minimal precipitation during June, July, and August (Fig. 6). Comparison of model outputs with these observational data demonstrates that the ensemble model effectively captures the overall pattern of Iranian precipitation and its monthly variability. Strong correlations between station precipitation and model outputs further support this finding. The ensemble model exhibited the highest correlation coefficient (r = 0.94). While the MPI and BCC models also accurately represented the maximum and minimum precipitation, the MRI model overestimated monthly precipitation values despite demonstrating a reasonable correlation.

Fig. 6

Evaluation of DCPP-CMIP6 model performance in simulating monthly temperature and total precipitation over Iran (1991–2019). Model outputs are compared with observational data averaged from 70 meteorological stations. Correlation coefficients (r) between model simulations and observations are shown in the right corner of each panel. The data were obtained from the Earth System Grid Federation (ESGF). URL link: https://esgf-metagrid.cloud.dkrz.de/search/cmip6-dkrz/. CMIP6-DCPP data were processed using Climate Data Operators (CDO), Version number: 2.0.4, under Ubuntu 22.04 LTS. URL link: https://code.mpimet.mpg.de/projects/cdo. The graphs were generated using Jupyter Notebook under Anaconda3, Version number: 2.14.1.

The analysis of mean monthly temperatures across Iran demonstrates a consistent seasonal cycle. Temperatures exhibit a progressive increase from February, culminating in peak values during July and August. Conversely, a decline initiates in September, extending through the winter months, with minimum temperatures recorded in December and January (Fig. 6). In assessing model performance, we found that temperature outputs from the DCPP models exhibited higher correlations with station data compared to precipitation, reflecting the lower spatial and temporal variability of temperature in Iran. Specifically, the ensemble model achieved the highest correlation coefficient (r = 0.99). Alongside the ensemble model’s superior performance, the MPI and BCC models also yielded satisfactory results. However, consistent with the precipitation analysis, the MRI model demonstrated a comparatively weaker ability to reproduce monthly temperature variations (Fig. 6).

To further evaluate the models’ performance on a monthly scale, Taylor diagrams were employed. Analysis of the DCPP-CMIP6 models reveals a range of standard deviations for precipitation between 12 and 14, indicating significant variability in monthly precipitation outputs (except MRI with a value > 15). Notably, the multi-model ensemble (MME) exhibits a standard deviation most proximate to the observational data. Moreover, the MME demonstrates the highest correlation coefficient with observational data, coupled with the lowest Root Mean Square Error (RMSE), less than 5 mm (Fig. 7a). Therefore, the MME exhibits minimal error in estimating precipitation across Iran and surpasses the performance of all individual models. Subsequently, the MPI and BCC models demonstrate commendable performance. Conversely, the MRI model consistently manifests the least favorable metrics (r, RMSE, SD) in comparison to station data, signifying the poorest performance in estimating Iranian precipitation (Fig. 7a).

Fig. 7

Taylor diagrams illustrating the performance of the DCPP-CMIP6 models in comparison to observational data: (a) evaluation of the models for monthly precipitation and (b) evaluation of the models for monthly temperature. The Earth System Grid Federation (ESGF). URL link: https://esgf-metagrid.cloud.dkrz.de/search/cmip6-dkrz/. CMIP6-DCPP data were processed using Climate Data Operators (CDO), Version number: 2.0.4, under Ubuntu 22.04 LTS. URL link: https://code.mpimet.mpg.de/projects/cdo. Station data are available at https://data.irimo.ir. The Taylor diagrams were generated using Jupyter Notebook under Anaconda3, Version number: 2.14.1, with the SkillMetrics library. URL link: https://pypi.org/project/SkillMetrics/.

Conversely, the DCPP-CMIP6 models demonstrate superior performance in estimating monthly temperatures compared to precipitation, as evidenced by all three statistical metrics. The correlation coefficients for all models exceed 0.95, with the MME achieving a correlation coefficient of 0.99 with station data (Fig. 7b). Thus, the MME effectively estimates temperature variations across Iran. Furthermore, the MME’s standard deviation and RMSE outputs closely align with observations. Overall, the MME exhibits robust and reliable performance in estimating both precipitation and temperature variables.

This section addresses the limitations imposed by the present study’s decadal prediction framework, which, by design, focuses on decadal and annual scales, thereby constraining the resolution necessary for the identification and elucidation of intra-annual variability within the examined variables. Consequently, we aim to evaluate the capacity of DCPP-CMIP6 models to represent intra-annual variability and the seasonal cycle of precipitation extremes. Specifically, we examine two indices, heavy precipitation (R10mm) and the simple daily intensity index (SDII), across both the hindcast (1991–2019) and forecast (2023–2028) periods. It is important to note that the criteria and methodology employed for other extreme indices—heat waves, cold waves, and droughts—prohibited the investigation of their seasonal cycles and intra-annual variability within the scope of this analysis. Therefore, these indices have been excluded from this section. To maintain analytical consistency and clarity, this section will focus solely on the DCPP-MME model.

Analysis of the mean frequency of heavy precipitation (R10mm) across the hindcast and forecast periods demonstrates significant intra-annual variability in Iran. During the hindcast period, spring exhibits a markedly higher mean frequency of heavy precipitation compared to other seasons, followed by winter. Conversely, autumn and, particularly, summer show the lowest frequencies, contributing minimally to annual totals. Figure 8 confirms the spring’s dominance, revealing that it experiences the highest frequency of heavy precipitation within the seasonal cycle. Notably, in 21 years of the hindcast period, spring recorded the highest frequency of heavy precipitation. Despite the reduced number of Mediterranean and Sudanese low-pressure systems in spring compared to winter, these systems appear to exhibit an enhanced potential for heavy precipitation. This phenomenon has been attributed to increased atmospheric river activity over the past three decades, which transports substantial moisture from the warm southern seas (Red Sea, Arabian Sea, Sea of Oman, and the Persian Gulf), thereby augmenting precipitation intensity and volume during spring¹⁵⁶. Recent examples of extreme spring precipitation include the exceptionally heavy rainfall in March 2019 across Iran^72,157,158 and the torrential, short-duration rainfall in May 2024 in northeastern Iran, which caused extensive damage. Quantitatively, the mean frequency of heavy precipitation for spring during the hindcast period is 2.24 days per year, compared to 1.73, 0.88, and 0.13 days per year for winter, autumn, and summer, respectively (Fig. 8). Unexpectedly, the frequency of heavy precipitation in winter is lower than in spring during the hindcast period, despite the peak activity of low-pressure systems occurring in winter¹⁵⁹.

An examination of the frequency of heavy precipitation occurrences during the forecast period (2023–2028) reveals a clear reduction in heavy precipitation across Iran. Specifically, the total annual frequency of this event decreases from approximately 5 days per year during the hindcast period to about − 0.31 days in the forecast period (Fig. 8). This notable decline in heavy precipitation in the near future across Iran is primarily attributable to a decrease in the frequency of heavy precipitation during the spring season. Interestingly, the DCPP-MME model output for the near future indicates an increase in heavy precipitation occurrences during the winter season, amounting to 0.35 days per year.

Fig. 8

Area-averaged seasonal cycles of precipitation extremes from the DCPP-MME model for the hindcast (1991–2019) and forecast (2023–2028) periods. Upper panels: frequency of heavy precipitation (R10mm, days/year). Lower panels: simple daily intensity index (SDII, mm/day). Forecast period data are presented as anomalies. The Earth System Grid Federation (ESGF). URL link: https://esgf-metagrid.cloud.dkrz.de/search/cmip6-dkrz/. CMIP6-DCPP data were processed using Climate Data Operators (CDO), Version number: 2.0.4, under Ubuntu 22.04 LTS. URL link: https://code.mpimet.mpg.de/projects/cdo. The seasonal indices of R10mm and SDII were computed using the Climpact2 package URL link: https://github.com/ARCCSS-extremes/climpact2 and Jupyter Notebook under Anaconda3, Version number: 2.14.1. Graphs generated by Numbers, Version number: 13.1, URL link: https://apps.apple.com/us/app/numbers/id409203825?mt=12.

In synthesis, the comparative analysis of the SDII and R10mm indices highlights a significant divergence in the seasonal patterns of precipitation intensity and frequency across Iran. The hindcast period demonstrated a spring-dominated regime characterized by both high precipitation intensity and frequent heavy precipitation events. However, the forecast period projects a notable shift, with a reduction in spring precipitation intensity and frequency, coupled with a compensatory increase in intensity during winter and autumn. This transition, particularly pronounced in the southwestern Zagros region, suggests a potential reconfiguration of regional precipitation dynamics, likely influenced by alterations in atmospheric circulation patterns and the activity of Sudanese low-pressure systems or atmospheric rivers⁷¹. These findings underscore the importance of considering both precipitation intensity and frequency when assessing future climate projections and have significant implications for water resource management and hazard mitigation in Iran.

Monitoring and predicting compounding Climate-related hazards

The compound analysis of extreme events helps to detect and better understand climate risks in various regions of Iran during the hindcast and forecast periods. It also provides invaluable information for climate-related planning and risk management. In this section, we intend to assess the risk of regions prone to compound hazards originating from individual climate hazards during the hindcast and forecast periods. By converting all climate extreme indices to a standard score, a compound hazard risk map of Iran was created for both the hindcast and forecast periods. Figure 9 illustrates the critical compound climate extremes in Iran. Nearly 27% of Iran’s area has experienced high and very high risk in the hindcast period. About 34% of Iran’s area exhibited low risk in this period. In contrast, regions with high and very high risks are expected to cover over 36% of Iran’s area, with an increase of 8.5% during the forecast period. The results indicate a significant rise in areas prone to extreme hazards and high risk during the coming decade. In the forecast period, regions with low to moderate climate risk significantly decrease (~ 15.7%), whereas areas with high risk are expected to increase by approximately 10% compared to the hindcast period. The spatial distribution of compound events indicates that regions with high and very high climate risk are scattered in various geographical regions in the hindcast period. The most extensive high and very high-risk region is in western Iran, extending from the southern shores of Lake Urmia to the Persian Gulf, continuing southeastward through the Zagros Mountains. High climate risk in the mentioned region is attributed to the combination of precipitation extremes, as well as frequent and prolonged droughts and cold waves. Additionally, the DCPP-MME identifies the western Caspian Sea and eastern Iran as hotspots of high and very high climate risk. In contrast, regions with the lowest climate risk include central Iran, smaller regions in the northwest, the coastal strip of the Persian Gulf, and the Oman Sea.

The spatial analysis of compound events in the forecast period reveals that high and very high climate risk will be observed in four main geographical regions of Iran, including (1) Regions in the west, along the southern shores of Lake Urmia, close to the Iraq-Iran border. (2) Regions in eastern Iran, proximate to the Afghanistan border. (3) Regions in northeastern Iran. (4) Large regions located in the southern Zagros Mountains. Notably, all four abovementioned regions are among the mountainous parts of Iran (Fig. 1). Regions 1 to 3 were identified as having high and very high climate risk during the hindcast period. Conversely, the southern Zagros Mountains, identified as number 4, exhibited low to moderate climate risk during the same period. It seems that the combination of drought characteristics and frequent and prolonged heat waves has led to the transformation of the southern Zagros Mountains into a region with a very high risk. A combination of heat waves, drought, and heavy precipitation events causes a high and very high climate risk in western Iran (first region). The combination of heat waves and drought events leads to high climate risk in eastern Iran (the third region). In contrast, some regions of central Iran, particularly the Dasht-e Lut and the coastal strip of southern Iran, are predicted to have the lowest climate risk in the near-term (Fig. 9).

Fig. 9

The compound hazards map for Iran using ten climate extreme indices (R10mm, SDII, DF, DD, DS, DI, HWF, HWD, CWF, CWD) is demonstrated for the hindcast (1991–2019) and the forecast (2023–2028) Periods. All extreme indices are converted to standard scores to represent climate risks, which are categorized into five classes: very low, low, moderate, high, and very high. Column charts illustrate the percentage of Iran’s area affected by each class of climate risk for both the hindcast and forecast periods. The data were obtained from the Earth System Grid Federation (ESGF). URL link: https://esgf-metagrid.cloud.dkrz.de/search/cmip6-dkrz/. CMIP6-DCPP data were processed using Climate Data Operators (CDO), Version number: 2.0.4, under Ubuntu 22.04 LTS. URL link: https://code.mpimet.mpg.de/projects/cdo. The climate extreme indices were computed using the Climpact2 package. URL link: https://github.com/ARCCSS-extremes/climpact2, and Jupyter Notebook under Anaconda3, Version number: 2.14.1. All maps and graphs were created using ArcGIS Desktop (Release 10, Version number: 10.8.2) by Esri, Redlands, CA: Environmental Systems Research Institute. URL link: https://www.esri.com.

Sensitivity analysis of Multi-Hazard weighting and hotspot confidence levels

As described in the previous section, compound climate-related hazards (multi-hazards) were identified for both the Hindcast and Forecast periods. Building upon these results, the present section conducts a sensitivity analysis to evaluate the robustness of the multi-hazard maps. As noted earlier, an equal-weight approach was initially applied when integrating the climatic hazards to construct the baseline climate risk maps, reflecting the study’s main objective, the distinct characteristics of the examined extremes, and Iran’s high climatic diversity.

To assess the robustness and spatial stability of the identified hot spots, additional climate risk maps were generated for both the historical and forecast periods using unequal weights assigned to four major hazard groups—drought, heavy precipitation, heatwave, and coldwave—while maintaining the total ranking scores across all hazards constant to ensure comparability. Comparing these alternative maps with the reference (equal-weight) compound climate-related risk map allowed for an evaluation of the consistency and reliability of the spatial patterns, particularly regarding the positioning of high-risk hot spots across Iran.

The Getis–Ord Gi* spatial statistic was further applied to identify statistically significant high-risk clusters, providing a quantitative measure of confidence in the detected spatial patterns. This combined approach—incorporating both weighting variation and hotspot confidence level assessment—served as a key analytical framework for evaluating the spatial robustness of the identified hot spots under different model assumptions about the relative influence of extreme climatic indicators.

During the Hindcast period, the hot spots identified in the reference (equal-weight) map exhibited strong spatial persistence when subjected to changes in indicator weights. The alternative composite maps emphasizing specific hazards—drought, heatwave, heavy precipitation, and coldwave—were largely consistent in reidentifying the same high-risk regions, particularly across eastern Iran and the Zagros Mountains. This consistency underscores the reliability of the spatial clustering patterns and highlights the dominant climatic controls shaping multi-hazard exposure in these areas (Fig. 10).

Fig. 10

Sensitivity of the compound climate-related multi-hazard weighting across Iran under five weighting scenarios—Equal Weight, Drought-Dominated, Heatwave-Dominated, Heavy Precipitation-Dominated, and Coldwave-Dominated—for both the Hindcast (1991–2019) and Forecast (2023–2028) periods. The figure also illustrates the sensitivity of identified hot spots to varying statistical confidence thresholds (99%, 95%, and 90%) across all scenarios and timeframes, highlighting the robustness and spatial stability of the detected high-risk regions. The data were obtained from the Earth System Grid Federation (ESGF). URL link: https://esgf-metagrid.cloud.dkrz.de/search/cmip6-dkrz/. CMIP6-DCPP data were processed using Climate Data Operators (CDO), Version number: 2.0.4, under Ubuntu 22.04 LTS. URL link: https://code.mpimet.mpg.de/projects/cdo. The climate extreme indices were computed using the Climpact2 package. URL link: https://github.com/ARCCSS-extremes/climpact2, and Jupyter Notebook under Anaconda3, Version number: 2.14.1. All maps and graphs were created using ArcGIS Desktop (Release 10, Version number: 10.8.2) by Esri, Redlands, CA: Environmental Systems Research Institute. URL link: https://www.esri.com.

Furthermore, an analysis of the percentage overlap for the High and Very High climate risk zones during the Hindcast period revealed the strong influence of specific hazard groups on the composite pattern. Among the alternative weighting scenarios, the Coldwave-dominated map exhibited the highest spatial overlap with the reference (equal-weight) map, exceeding 80%, followed by the Drought-dominated map with an overlap of more than 67%. This substantial level of agreement indicates that, during the Hindcast period, coldwave and drought hazards were the principal drivers shaping the overall multi-hazard configuration. Consequently, the dominance of these two factors reinforces the robustness and validity of the equal-weight approach for this period, as it effectively captured the primary climatic controls governing the spatial distribution of high-risk hot spots across Iran.

In the Forecast period, the spatial configuration of the identified hot spots remains broadly consistent with that of the Hindcast period, though a stronger overall alignment with the reference (equal-weight) map is evident across most alternative weighting scenarios. The projected high-risk zones—concentrated primarily in eastern, central, and northwestern Iran—demonstrate remarkable spatial stability under varying weighting assumptions, indicating strong model robustness (Fig. 10).

The percentage area overlap analysis further reveals a shift in the dominant drivers of risk toward temperature-related hazards. Composite maps emphasizing heatwave and coldwave hazards display the highest degree of overlap with the reference map, exceeding 90% for areas classified as High or Very High risk. This exceptional level of spatial agreement suggests that thermal extremes are expected to become the principal determinants of the multi-hazard configuration in the forecast period. Such dominance of temperature-related hazards underscores the reliability of the projection results and highlights the growing importance of adaptive strategies to address anticipated increases and variability in future temperatures.

Analysis of hotspot confidence levels provides further evidence of the statistical robustness of these findings. The composite hotspot maps produced under different weighting scenarios—each doubling the emphasis on drought, heatwave, precipitation, or coldwave indicators—are predominantly characterized by a 99% confidence level. This high level of statistical significance is particularly evident in the cores of extensive high-risk regions, confirming the stability and reliability of the estimated intensity and spatial extent of compound climatic hazards. In essence, the identification of these principal high-risk clusters is supported by strong statistical precision and very low uncertainty, lending further confidence to the predictive framework (Fig. 10).

Conclusion

This study evaluated compounding climate-related hazards in Iran by analyzing historical climate data from 1991 to 2019 and projecting future changes for 2023 to 2028 using a near-term climate prediction framework. Our results provide detailed insights into the spatial and temporal evolution of extreme climate events across the country. The critical findings of this research are as follows:

• The analysis showed that CMIP6-DCPP successfully reproduced the overall spatial patterns of climate extremes across Iran. Some discrepancies appeared mainly in Iran’s arid and mountainous regions, where observational data are limited. These data gaps likely contributed to slight overestimations or underestimations in certain areas when compared to recorded observations.

• The DCPP-CMIP6 multi-model ensemble (MME) effectively reproduces Iran’s seasonal temperature and precipitation cycles, demonstrating a high correlation with observational data, particularly for temperature. Notably, the models accurately capture the spring peak in heavy precipitation (R10mm), attributed to atmospheric river activity. Projections suggest a shift in these seasonal patterns, with a future reduction in spring precipitation intensity and frequency and a potential increase in winter precipitation. This indicates a reconfiguration of regional precipitation dynamics, highlighting the DCPP-MME’s ability to reveal critical seasonal variations and their future changes.

• Looking forward, the Zagros Mountains and the southwest regions are expected to become major hotspots for heavy rainfall events. The model projects an increase in both the frequency of days with heavy precipitation and the intensity of rainfall. This shift in precipitation patterns is anticipated to alter water availability and could increase the risk of flash floods in these regions.

• Analysis of SPEI-12 drought indices reveals a significant projected increase in drought frequency across Iran, particularly in central, eastern, and southeastern regions, while paradoxically, drought duration, severity, and intensity are expected to decrease. This suggests a future climate characterized by more frequent but less intense and shorter-lived drought events, likely due to altered hydrological cycles and accelerated evaporation, though regional variations persist.

• The study further indicates that heat waves will become more frequent and last longer in several regions of Iran. Eastern regions, southern parts of the mountainous Zagros, and western areas are all projected to experience an increase in both the duration and frequency of extreme heat events. In particular, the extended duration of these heat waves may have more severe impacts on public health, energy consumption, and water resources compared to the increase in frequency alone. Additionally, high-altitude areas are projected to experience an accelerated warming trend. These faster warming rates in mountain regions could significantly impact local ecosystems, water resources from snowmelt, and the overall climate dynamics in these areas.

• By integrating multiple hazard indices into a single compound risk map for the forecast period, our study highlights that large parts of Iran will face high to very high levels of climate-related hazards. Regions such as the western border, along the southern shores of Lake Urmia, the southern Zagros Mountains, and the eastern semi-arid zones are particularly at risk and will likely require urgent adaptation planning. We believe that a comprehensive understanding of these emerging risks is essential for developing targeted adaptation strategies, which in turn can help mitigate future climate hazards and support effective decision-making in vulnerable regions.

• Based on the sensitivity analysis, which examined variations in indicator weighting and the confidence levels of identified hot spots in compound climate-related hazard maps, the results demonstrate that the detected high-risk regions possess strong spatial stability and robustness. This stability is not solely a consequence of the equal-weight approach but rather reflects the dominant influence of one or two key climatic hazards in each period. The persistence of these patterns, confirmed at a high statistical confidence level (predominantly 99%) within the hotspot cores, provides compelling evidence of the model’s reliability and its effectiveness in detecting statistically significant high-risk areas across Iran.