Document Type : Research Article
Authors
1
Postdoctoral Researcher in Climatology, Faculty of Social Sciences, University of Mohaghegh Ardabili, Ardabil, Iran
2
Professor in Climatology, Department of Physical Geography, Faculty of Social Sciences, University of Mohaghegh Ardabili, Ardabil, Iran
10.22067/geoeh.2025.93054.1565
Abstract
Climate change has led to alterations in the temporal and spatial distribution of precipitation across the globe. The present study aimed to identify trends and project future precipitation in selected cities of northern Iran. To achieve this goal, the 30-year precipitation trend (1994–2023) for each city was first examined using the Mann-Kendall test and Sen’s slope estimator. In the second phase, by employing the LARS-WG8 weather generator, the latest version of this software, precipitation in the study area was projected for the next 20 years under two scenarios: SSP2-4.5 and SSP5-8.5, using the ACCESS-ESM1-5 and CNRM-CM6-1 models. The outputs of both models were then evaluated and compared using validation indices (R², MAPE, and RMSE) to select the most suitable model. Trend analysis results indicated that over the past 30 years (1994-2023), the precipitation pattern in all stations in the region has changed and has had a significant decreasing trend at the 99% confidence level. This decrease has been greater in the cities of Qazvin, Ardabil, and Zanjan and less in Rasht and Gorgan. According to the prediction made based on the scenarios of the two models ACCESS-ESM1-5 and CNRM-CM6, over the next 20 years (2040-2060), the entire study area will face a decrease in precipitation in most months of the year. The decrease in precipitation in the northwest will be more than 60% in some months of the year. Also, some cities will experience an increase in precipitation of the warm seasons. This increase will be more evident in cities with low precipitation such as Semnan. Finally, the comparison of the total annual precipitation of the observed and forecast periods confirmed a significant decrease in precipitation in the entire study area compared to the base period. The research findings indicate temporal and spatial anomalies in precipitation, confirm the effects of climate change on the region, and emphasize the need to adopt efficient management policies to deal with, optimize water resource consumption, and develop methods for adapting to climate change.
Introduction
Climate change has caused changes in the temporal and spatial distribution of precipitation and its intensity throughout the world. Precipitation description and modeling are among the challenging issues in climate research and indicate its complex and irregular temporal and spatial structure. In this regard, several modeling methods have been developed to deal with this complexity (Sánchez et al, 2011). Reducing the number of rainy days and increasing the intensity of precipitation during limited days can lead to an increase in the probability of flooding in areas such as coastal cities of the Caspian Sea, soil erosion and reduced water infiltration into groundwater resources.
Methods and Material
In the present study, the trending and forecasting of precipitation in selected cities in northern Iran (Ardabil, Tabriz, Bojnourd, Mashhad, Gorgan, Semnan, Zanjan, Tehran, Urmia, Gorgan, Rasht, Sanandaj) was investigated. In order to trend the total annual precipitation, the data of the last 30 years (1994-2023) of the precipitation of the selected cities were received from the Iranian Meteorological Organization and the age slope test, which is a non-parametric method for estimating the slope of the regression line in data trend analysis, was used. This test includes the following steps:
(1) Qi=(Xt -Xs)/(t-s),i=1,2,…N
Where xt and xs are the data at times t > s and N= (n (n-1))/2, respectively.
(2) C α=Z_(1-α/2)=√(Var (A))
In the above equation, Z is the quantile of the standard normal distribution (Salmi et al, 2002).
(3) M1=(N-Cα)/2
M2=(N+Cα)/2
Where N is the number of slopes calculated in section a.
To determine the significance of the trend, the Mann-Kendall test was used. In this test, the null hypothesis implies randomness and the absence of a trend. In this method, first the difference between each observation and all subsequent observations is calculated and the parameter S is obtained based on equation (4):
(4) S=∑_(k-1)^(n-1)▒∑_(j-k+1)^n▒〖sgn〗^((x_(j+x_k )))
In the above equation n is the number of observations in the series Xj and Xk are the kth and jth data of the series, respectively.
Next step, the variance s was calculated using equation (5).
(5) var(s)=(n(n-1)(2n+5))/18
Where n is the number of sequences in which at least one data is repeated. Finally, the z statistic was obtained based on equation (6).
(6) S=(S-1)/(√var(s))
Assuming two ranges of the trend test, the null hypothesis is accepted if the following condition (relation 7) holds:
(7) ǀZǀ<Z_(α⁄2)
In the above relationship, Z is the significance level and Z∝ is the standard normal distribution statistic at the significance level. In the present study, this test has been used for the 99% and 95% confidence levels, and if the Z statistic is positive, the trend of the data series is upward and if it is negative, the trend is downward.
In order to examine the confidence limits, in equation (1), which is arranged from small to large, M1th and M2+1th slopes are obtained. If the number zero is in the range between the two obtained slopes, the null hypothesis is confirmed and the time series is without trend (ghorbani et al, 2012). In the next part of the research, in order to predict future precipitation, first the base period data during the years (1980-2010) were simulated in the LARS-WG8 microcomputer. Then, the precipitation forecast for the next 20 years (years 2041 to 2060) was carried out using the two ACCESS-ESM1-5 and CNRM-CM6-1 models under the scenarios (SSP2-4.5) and (SSP5-8.5). Finally, to examine the accuracy of the two mentioned models in predicting future precipitation of the stations, three error indicators were used; (R2), (RMSE) and (MAPE) were used (relations 8 to 10):
(8) R^2=(∑_(t=1)^n▒〖A_t F_t 〗)/√(∑_(t=1)^n▒〖A_t^2 ∑_(t=1)^n▒F_t^2 〗)
(9)
RMSE=√(∑_(t=1)^N▒(A_t-F_t )^2 )/n
(10) MAPE=(∑_(t=1)^n▒|(A_t-F_t)/A_t | )/n×100
In the above equations, At is the observational data, Ft is the simulated data and N is the number of data.
Results and Discussion
According to the results of trend analysis, during the statistical period under study (1994-2023) in all stations, the Q statistic (Sen’s slope estimator) and Z (Mann-Kendall test) were negative and the total annual precipitation in all 12 cities had a decreasing trend. The highest Q was in Qazvin -81.8, Ardabil -63.8 and Zanjan -45.8, and the highest Z values were in Sanandaj, Zanjan and Qazvin cities -4.62, -4.57 and -4.42, respectively, and the lowest value was in Rasht with -2.98. Was negative and the total annual precipitation in all 12 cities had a decreasing trend. The highest Z values were related to the Sanandaj, Zanjan and Qazvin cities and were 4.62, -4.57, -4.42 respectively, and the lowest value was related to the Rasht station with the figure of -2.98; however, considering that the Z score in all cities is less than -2.58, therefore, the existence of a significant downward trend in precipitation in the last 30 years is accepted in the entire study area at a confidence level of 99%. In the second part of the study, after evaluating the accuracy of the LARS-WG micro-rotor software, in simulating the precipitation of the base period (1980-2010), two ACCESS-ESM1-5 and CNRM-CM6-1 models were used to predict the future precipitation of the cities under the (SSP2-4.5) and (SSP5-8.5) scenarios. According to the results, over the next 20 years, Urmia city will experience a decrease in precipitation in all months of the year under both model scenarios. The cities of Tehran, Tabriz, Ardabil, Gorgan, and Rasht will experience a significant decrease in precipitation in most months of the year under both model scenarios, and a very slight increase in precipitation (less than 20 percent) in some months under one of the scenarios. The cities of Zanjan, Sanandaj, Qazvin, Mashhad, Bojnourd, and Semnan will experience a 50 percent increase in precipitation in at least one of the final months of spring and summer under one of the scenarios. Overall, the predictions made based on the scenarios of both ACCESS-ESM1-5 and CNRM-CM6-1 models show that all 12 cities studied will witness a significant decrease in precipitation compared to the long-term average in most months of the year over the next 20 years.
Finally, the accuracy of the two models studied in predicting precipitation was examined based on three error indicators: R2, MAPE and RMSE. According to the results, although the error rate was relatively high in both models, each of the cities had the lowest error and highest accuracy in one model, which was considered as the selected model for that station. Therefore, the CNRM model was selected for the four cities of Sanandaj, Ardabil, Tehran, and Gorgan, and the ACCESS model was selected for the other cities.
Conclusion
The aim of the present study was forecasting and trending precipitation in selected cities in northern half of Iran using GCMs models to reveal the extent of the impact of precipitation in different climatic regions of this part of the country on climate change. For this purpose, the precipitation of the last 30 years (1994-2023) of selected cities in the northern half of Iran was trended based on the Sen’s slope estimator and Mann-Kendall tests, and its possible changes over the next 20 years (2041-2060) were predicted using the two ACCESS and CNRM models. Based on the results of the two Mann-Kendall tests and Sen’s slope estimator, over the past 30 years (1994-2023), the precipitation pattern in all cities in the region has changed and has had a significant decreasing trend at the 99% confidence level. The decreasing trend has been greater in the cities of Qazvin, Ardabil, and Zanjan, which are considered mountainous cities in the northwest, and less in Rasht and Gorgan, which are rainy and humid coastal areas. According to the predictions made based on the scenarios of the two ACCESS and CNRM models, the precipitation patterns in the study area will undergo significant changes. Over the next 20 years (2040-2060), the entire study area will face a decrease in precipitation in most months of the year. The decrease in precipitation in cities located in the northwest will be more than 60% in some months of the warm season. Also, some cities in the region will experience an increase in precipitation in the warm months of the year (late spring to late winter), at least based on the prediction of one of the two model scenarios, which will be more evident in cities with low precipitation such as Semnan. According to the explanations provided and considering that the highest percentage of increase and decrease in precipitation in the entire region is related to the warm months of the year. Therefore, it is expected that during the forecast period (2040-2060) the precipitation in the warm seasons will undergo more changes and fluctuations. Finally, a comparison of the total annual precipitation of the observation period and the forecast based on the scenarios of both models confirmed a significant decrease in precipitation in all 12 cities studied compared to the base period, with the highest predicted decrease in precipitation compared to the observation period in Rasht and the lowest in Semnan. The aforementioned results indicate temporal and spatial anomalies in the precipitation of the region.
Keywords
Main Subjects
Send comment about this article