عنوان مقاله [English]
Climate change is one of the major threats of the 21st century. One of the major data sources for studying climate change is the general circulation model (GCM), which is widely used to assess and project past and future climate change and to manage regional risk hazards. GCMs can make significant assessments of temperature and precipitation. The application of individual models has high uncertainty. Therefore, in this study, applying an ensemble approach has been considered to project future temperature changes in Iran. The purpose of this study is to create a multi-model ensemble (MME) with bias-corrected CMIP6 models to project the temperature of Iran and the warm spell duration index (WSDI) in the near future (2021-2040).
Materials and methods
To evaluate the CMIP6 models in simulating the average annual temperature for the period of 1995 to 2014 (these 20 years were considered as the historical period), we used 120 synoptic stations in Iran. In this study, five CMIP6 models (GFDL-ESM4, MPI-ESM1-2-HR, IPSL-CM6A-LR, MRI-ESM2-0, UKESM1-0-LL) with a horizontal resolution of 0.5 degrees were used. Using intermediate (SSP3-7.0) and worst-case (SSP5-8.5) scenarios, annual temperature and heat stress were projected under climate change conditions for the near future (2040-2021).
In this study, normalized root mean square error (NRMSE) and mean bias error (MBE) were used to validate the performance of the models. To correct the bias of CMIP6 models, Delta change Factor (DCF) method with 120 synoptic station was used. Then, independence weighted mean (IWM), were used to ensemble five models.
In this study, in addition to temperature anomalies, warm spell duration index (WSDI) was also projected by 2040. The warm spell duration index is the number of days per year with at least 6 consecutive days when TX > TX90th, where TX is the daily maximum temperature and TX90th is the calendar day 90th percentile.
Results and discussion
Based on the Direct Model Output (DMO) in the climates of Sea of Oman and Caspian Sea coasts, all five CMIP6 models underestimated the average annual temperature (Chabahar and Rasht stations). However, CMIP6 models in other climates of Iran overestimated the average annual temperature. The mean bias of 1.00, 0.962, 0.983, 1.001, 0.936 degrees Celsius is compute for GFDL-ESM4, IPSL-CM6A-LR, MPI-ESM1-2-HR, MRI-ESM2-0 and UKESM1-0-LL, respectively. Therefore, UKESM1-0-LL and IPSL-CM6A-LR models are efficient models among the studied models for estimating Iran temperature, respectively. The temperature bias values fluctuate from -2.27 to 2.70 degrees Celsius in Iran. The average annual temperature based on MME-CMIP6 output fluctuates between 6.27 and 27.7 degrees Celsius. The coasts of the Persian Gulf and the Sea of Oman showed the maximum temperature and the northwestern regions of Iran showed the minimum temperature. The warm spell duration index varies between 3.48 to 12.5 days during the historical period. The temperature projected for both SSP3-7.0 and SSP5-8.5 scenarios show a further increase in the interior, northwest, north, and northeast of Iran. The average annual anomaly in Iran is estimated to be 1.13 and 1.26 degrees Celsius, based on SSP3-7.0 and SSP5-8.5 scenarios, respectively for the near future during the years 2021-2040.
According to the SSP3-7.0 scenario, the minimum temperature anomaly of the country is 0.765 and the maximum is 1.227 degrees Celsius. Also, for the SSP5-8.5 scenario, temperature anomalies for minimum and maximum annual temperature are estimated to be 0.785 and 1.380 degrees Celsius, respectively. Projecting the time-series changes of the temperature in eight representative stations of Iran under the scenarios of SSP3-7.0 and SSP5-8.5 showed that the amount of warm spell duration varies in different regions of Iran. The anomaly of the warm spell duration index (WSDI) in Iran according to the SSP3-7.0 scenario will increase by at least 13.1 and at most 58.6 days. Also, the results related to the worst-case scenario (SSP5-8.5) have shown a minimum increase of 17.4 days and a maximum of 74.5 days in Iran.
Evaluation of direct output of five models from the series of Coupled Model Intercomparison project (GFDL-ESM4, MPI-ESM1-2-HR, IPSL-CM6A-LR, MRI-ESM2-0, UKESM1-0-LL) In the historical period (1995-2014) showed that although some models performed better in some climate regions of the country, the use of the direct output of individual models will increase the uncertainty in the results. For this purpose, using the independent weighted average (IWM) method, the output of the models was improved. The evaluation of the output of the ensemble model emphasizes this result.
In general, the results of model validation showed that the output of CMIP6 models has higher performance for arid and semi-arid regions of Iran, whether before or after bias-correction. However, using either individual or bias-corrected and ensemble models in very humid climates (The northern parts of Iran) should be used with more caution because even with the correction of bias and doing ensemble, the positive bias is more than 2 degrees Celsius for the mentioned areas.
The projection by using MME-CMIP6 show the annual temperature increase in near future (2021-2040). The average annual temperature anomaly will increase by 1.13 and 1.26 degrees Celsius, under the scenarios of SSP3-7.0 and SSP5-8.5 respectively, during the near future (2021-2040). The results indicate the very important role of unevenness in the heterogeneous distribution of temperature in increasing the average annual temperature of the country in the next two decades. The minimum temperature anomaly is projected in the southeast of Iran and the maximum in the northwestern and central regions.
Projecting the warm spell duration index (WSDI) indicates an increasing anomaly of that in the country. The main hotspot of WSDI is on the southern coast of Iran, especially in the Persian Gulf area, which according to the results of the worst-case scenario will increase by 74.5 days by 2040. Analysis of changes in the average annual temperature of MME-CMIP6 shows that Iran will warm up more rapidly in the near future (2021-2040) than in the historical period (1995-2014). More warming, especially in high and snowy areas, will affect natural ecosystems and limit future access to water resources. Most of the interior, east and south of Iran has arid and semi-arid climate and a sharp rise in temperature under the worst-case scenario (SSP5-8.5) leads to environmental degradation and intensification of drought on the one hand and increased desertification on the other.
Temperature projection, CMIP6 models, ensemble model, WSDI index, Iran.