Forecast of water levels in the Caspian Sea based on the sixth IPCC report

Water levels in the Caspian Sea fluctuate constantly on different time scales. In recent decades, several studies have examined the causes of fluctuations, the impact of climatic and non-climatic components, and the forecasting of water levels based on different climatic models. Considering that most of the predictions regarding hydro-climatic components and Caspian water level are based on a fixed time period. Thus, in this study, we considered the time base period for downscaling the Caspian Sea's variable level and, on the other hand, for integrating the best output of different climatic models. Accordingly, the purpose of this article is to estimate the future sea level of the Caspian Sea using the sixth IPCC report under optimistic and pessimistic scenarios.

Data for this study includes two groups of Caspian Sea water levels from 1941 to 2019, as well as the output of three climate models; INM-CM4-8, MIROC-ES2L, and MPI-ESM1-2-LR. As part of the above three models, we selected the following components: percentage of cloud cover (clt), evaporation flux (evspsbl), specific humidity (huss), precipitation flux (pr), sea surface pressure (psl), air temperature (tas), and wind components including orbital and meridional components (vas & uas) and surface stress (tauu & tauv) for the base and future periods, 1941 to 2080. These models are part of the CMIP6 international model comparison project produced under the name Phase 6. The scenarios used in this study are SSP2-4.5 and SSP5-8.5, which are based on common paths identified in the sixth IPCC evaluation report. An analysis of the Caspian Sea water level was conducted using Statisca software to model it using a multivariate regression model. In order to model the Caspian Sea water level, the selected data matrix was adjusted by including a difference in level from the previous time, which can eliminate long-term trends in the Caspian Sea water level. In this study, based on long-term statistics of the Caspian Sea water level and climatic data output, in order to model the difference in the Caspian Sea water level, we considered the base periods with five time scales, which included periods 2019-1941 in order to select the best baseline variety for modeling the Caspian Sea water level. To evaluate the efficiency of Caspian Sea water level modeling based on different models, we divided the basic modeling course into two modes, training and testing. The climate models in this study were validated using various criteria, including mean absolute error (MAE), root mean square error (RMSE), coefficient of determination (R2), Nash-Sutcliffe (NS) and Durbin Watson (DW) camera adaptation percentage (Ghanghermeh et al., 2019).

According to MIROC-ES2L climate model NS, R2, and DW statistics, there is a good match between the training and testing periods in all time scales of the basic period. In the two climate models INM-CM4-8 and MPI-ESM1-2-LR, the base time scale of 40 and 50 years, i.e. the training period of 2001-2009 and 1971-2009, the models do not perform well during testing, while in other scales, both in testing and training modes, good performance is seen, with the exception that DW has excellent consistency in all scales. The results suggest that, using two scenarios including SSP2-4.5 and SSP5-8.5, the water level will decrease by an average of 43 cm per decade in the INM-CM4-8 climate model between 2021 and 2080, resulting in the level of -30.38 meters. However, based on the SSP5-8.5 scenario, the level is increasing from 2021 to 2050 by 20 cm, and then it is decreasing with an average intensity of 61 cm, finally reaching -29.07 cm in 2080. As shown by the MIROC-ES2L climate model, both scenarios show a decreasing trend in water level, with the average decrease in SSP2-4.5 scenario being equal to 87 cm per decade, while in SSP5-8.5 scenario, the trend is slightly different. As a result, the water level will decrease by 26 cm per decade by 2050 and by 102 cm per decade by 2080 based on the SSP2-4.5 scenario and by -31.76 meters in the SSP5-8.5 scenario. In According to the MPI-ESM1-2-LR climate model, the Caspian Sea water level behavior is different from the above two models, so that the trend of changes in the Caspian Sea water level will continue to increase from 2021 through 2060 with an average of 48 cm per decade, and then the slope of decline of the Caspian Sea water level will become smoother. However, according to the SSP5-8.5 scenario, the water level of the Caspian Sea will increase with a gentler slope, with fluctuations around 10 cm per decade. Finally, the amplitude of water levels in 2080 will be the least different between the two scenarios. Considering the combined results of the above three models, it will be determined that the Caspian Sea water level will be decreasing based on the SSP2-4.5 scenario. In this case, we see that first there will be a smoother slope trend by 2060, which will reach -29 meters, and then the slope trend will be more intense and will reach -30.4 meters. But according to the SSP5-8.5 scenario, the sea level will remain stable at the focal level until 2050, and then the downward trend will reach -29.5 meters by 2080.

As a result of merging the three climate models, it was found that in the SSP2-4.5 scenario, despite the average greenhouse gas emissions remaining at current levels by 2050 and decreasing by 2100, and of course by increasing the temperature by 2 degrees Celsius for the period 2060-2041 and by 2.7 degrees Celsius for the period 2100-2081, the water level in the Caspian Sea will consistently decrease. In the SSP8-8.5 scenario, with intense greenhouse gas emissions and a tripling of carbon dioxide level by 2075, the water level of the Caspian Sea will remain stable from the beginning to 2050, but from this year on, its decline will begin. Additionally, this study's findings are consistent with those of Algioni et al. (2006 and 2007), Korich et al. (2021), Chen et al. (2017), and Hosseini et al. (2020) concerning the decrease of the Caspian Sea water level