| چکیده انگلیسی مقاله |
Heavy rainfall is one of the types of environmental hazards that occur naturally and the role of humans in their aggravation is undeniable. In recent years, due to climate change, the occurrence of extreme weather events has increased. By using suitable climate models, it is possible to be prepared and reduce the harmful effects of climatic extremes through climate forecasting. In the northwest of Iran, the existence of mountainous topography provides the factor of ascent to create heavy convective rains, which is prone to flood phenomenon. In this research, two groups of observational and model data have been used daily to study heavy rainfalls. The daily rainfall data of 23 synoptic stations located in the northwest of Iran, including the provinces of East Azerbaijan, West Azerbaijan, Ardabil, North Kurdistan, and west of Zanjan province, were obtained from the Iranian Meteorological Organization (www.irimo.ir). The output of the CMIP6 models from the site https://esgf-node.llnl.gov/projects/cmip6/ for two periods 1990-2019 (historical period) and 2030-2059 (future period) based on three scenarios SSP1-2.6, SSP2 -4.5 and SSP5-8.5 were extracted as optimistic, medium and pessimistic scenarios, respectively. For this purpose, data of 8 AOGCM models (MIROC6, CANESM5, ACCESS-CM2, BCC-CSM2-MR, NORESM2-LM, IPSL-CM6A-LR, MRI-ESM2-0 and CNRM-CM6-1) from the CMIP6 model series were used. The raw rainfall output was first converted from NC to TXT format in the R software environment, and based on the coordinates of the stations in the study area, the output of the models was extracted for each station. After converting the unit to mm, the downscaling process was done by CMHyd software. The criterion of heavy rainfall in this research is the intensity of rainfall (99th percentile) and coverage of rainfall (simultaneous heavy rainfall in at least 30% of the stations). By calculating the PCC, KGE, RMSE, NSE, and R2 statistical measures, the efficiency of the models was evaluated and the ranking of the models was determined based on their performance. Also, to select the appropriate downscaling method among the three methods of Precipitation Local Intensity, Power Transformation, and Distribution Mapping, statistical indices NSE, MAE, and Taylor's diagram were used. According to the heavy rainfall criteria used in this research, 43 extreme rainfall events were identified in the observation period (1990-2019). The verification of the raw output of the studied models with the downscaled results of the models by the KGE statistical measure indicates that the results of the models are optimized after downscaling compared to model output before downscaling. According to the results of this research, the CNRM model was identified as the best and the NORESM2 model as the worst model for simulation heavy rainfall in northwestern Iran. In the CNRM model, the highest and lowest values of the KGE index are assigned to Khalkhal and Sahand stations, respectively. The maximum and minimum measures of NSE also belong to Sahand and Kalibar stations, respectively. The maximum and minimum RMSE index belong to the Kalibar and Jolfa stations, respectively, and the maximum and minimum R index belong to the Zarineh and Parsabad stations, respectively. In the NORESM2 model, the maximum and minimum KGE index belong to the Saqez and Kalibar stations, respectively. The maximum and minimum values of NSE are assigned to Sahand and Mako stations, respectively, and the maximum and minimum RMSE indices are assigned to Kalibar and Jolfa stations, respectively. The maximum and minimum R measures are assigned to Piranshahr and Kalibar stations, respectively. In 23 synoptic stations and 8 models, the lowest RMSE value, and the highest NSE value jointly in all 8 models of the studied area belonged to Jolfa and Sahand stations, respectively based on 5 statistical measures. The produced ensemble model showed better performance than individual models. The results of the Mann-Kendall test in the base period (1990-2019) and the future period (2059-2030) based on z-statistics indicate a decreasing trend of heavy rainfall in the northwest of Iran, but it is not statistically significant. The results of the heavy rainfall projection in northwest Iran using 8 GCM models presented in the CMIP6 models at 23 synoptic stations, indicate that the number of heavy rainfall events in the studied area in the future period (2059-2030), compared to the previous period (1990-2019), will increase according to two pessimistic (SSP5-8.5) and moderate (SSP2-4.5) scenarios; but in the optimistic scenario (SSP1-2.6) there will be no change in the number of extreme precipitation events. |