بررسي روند، تجزيه‌و‌تحليل، مدل‌سازي و پيش‌بيني بارش ماهانه با استفاده از مدل‌هاي تصادفي سري زماني (مطالعه‌ي موردي: ايستگاه سينوپتيك اردبيل)

Stochastic Monthly Rainfall Time Series Analysis, Modeling and Forecasting ( A cas study: Ardebilcity

اﻃﻼع از ﻣﻘﺪار آب در دﺳﺘﺮس آﯾﻨﺪه، ﯾﮏ ﮔﺎم ﺑﺴﯿﺎر ﻣﻔﯿﺪ در ﺗﺼﻤﯿﻢﮔﯿﺮيﻫﺎي ﻣﺪﯾﺮﯾﺘﯽ اﺳﺖ ﮐﻪ ﮐﻤﮏ ﺑﺎﻟﻘﻮهاي در ﺑﻬﺮهﺑﺮداري ﺑﻬﯿﻨﻪ و ﭘﺎﯾﺪار از ﻣﻨﺎﺑﻊ آﺑﯽ ﻣﻮﺟﻮد ﺧﻮاﻫﺪ ﻧﻤﻮد. ﻫﺪف از اﻧﺠﺎم اﯾﻦ ﻣﻄﺎﻟﻌﻪ، ﺑﺮرﺳﯽ روﻧﺪ و ﭘﯿﺶﺑﯿﻨﯽ ﺑﺎرﻧﺪﮔﯽ ﻣﺎﻫﺎﻧﻪي اﯾﺴﺘﮕﺎه ﺳﯿﻨﻮﭘﺘﯿﮏ ﻣﻨﺘﺨﺐ اﺳﺘﺎن اردﺑﯿﻞ ﺑﺎ اﺳﺘﻔﺎده از ﺑﻬﺘﺮﯾﻦ ﻣﺪلﻫﺎي ﺳﺮيﻫﺎي زﻣﺎﻧﯽ اﺳﺖ. در اﯾﻦ ﻣﻄﺎﻟﻌﻪ، ﺑﺎرﻧﺪﮔﯽ ﻣﺎﻫﺎﻧﻪي 5 ﺳﺎل آﯾﻨﺪه )2020 ﺗﺎ 2024 ﻣﯿﻼدي( در ﻣﻨﻄﻘﻪي ﻣﻮردﻣﻄﺎﻟﻌﻪ ﺑﺎ اﺳﺘﻔﺎده از ﻣﺪلﻫﺎي ﻣﺨﺘﻠﻒ ﺳﺮيﻫﺎي زﻣﺎﻧﯽ ﺧﺎﻧﻮادهي ARIMA ﭘﯿﺶﺑﯿﻨﯽ ﺷﺪ. در اﯾﻦ ﭘﮋوﻫﺶ، از آزﻣﻮن ﻧﺎﭘﺎراﻣﺘﺮﯾﮏ ﻣﻦ-ﮐﻨﺪال ﺑﻪ ﻣﻨﻈﻮر اﻃﻤﯿﻨﺎن از وﺟﻮد روﻧﺪ و از ﻧﻤﻮدار ﺧﻮدﻫﻤﺒﺴﺘﮕﯽ )ACF( ﻧﯿﺰ ﺑﻪ ﻣﻨﻈﻮر اﻃﻤﯿﻨﺎن از وﺟﻮد ﺗﻐﯿﯿﺮات ﻓﺼﻠﯽ در ﺳﺮي زﻣﺎﻧﯽ اﺳﺘﻔﺎده ﮔﺮدﯾﺪ. ﭘﺲ از اﻧﺘﺨﺎب ﺑﻬﺘﺮﯾﻦ ﻣﺪل ﭘﯿﺶﺑﯿﻨﯽﮐﻨﻨﺪه ﺑﺮ اﺳﺎس ﻣﻘﺎدﯾﺮ ﭘﺎراﻣﺘﺮﻫﺎي ﻣﺪل، ﻣﻌﯿﺎر آﮐﺎﺋﯿﮏ و ﺿﺮﯾﺐ ﻫﻤﺒﺴﺘﮕﯽ ﻣﻘﺎدﯾﺮ ﺑﺎرﻧﺪﮔﯽ ﻣﺎﻫﺎﻧﻪي 5 ﺳﺎل آﯾﻨﺪه )2020 ﺗﺎ 2024 ﻣﯿﻼدي( ﺑﺎ اﺳﺘﻔﺎده از ﺑﻬﺘﺮﯾﻦ روش اﯾﺴﺘﺎﺳﺎزي و ﺑﻬﺘﺮﯾﻦ ﻣﺪل ﭘﯿﺶﺑﯿﻨﯽﮐﻨﻨﺪهي ﻣﺮﺑﻮط ﺑﻪ آن، ﭘﯿﺶﺑﯿﻨﯽ ﮔﺮدﯾﺪ. ﻧﺘﺎﯾﺞ ﺣﺎﺻﻞ از آزﻣﻮن ﻣﻦ-ﮐﻨﺪال ﻧﺸﺎن داد ﮐﻪ دادهﻫﺎي ﺑﺎرﻧﺪﮔﯽ ﻣﺎﻫﺎﻧﻪي اﯾﺴﺘﮕﺎه ﺳﯿﻨﻮﭘﺘﯿﮏ اردﺑﯿﻞ در دورهي آﻣﺎري ﻣﻮردﻣﻄﺎﻟﻌﻪ داراي روﻧﺪ ﮐﺎﻫﺸﯽ )0/6119-=Z( ﺑﻮده، اﻣﺎ اﯾﻦ روﻧﺪ در ﺳﻄﺢ اﻃﻤﯿﻨﺎن 95 درﺻﺪ ﻣﻌﻨﯽدار ﻧﯿﺴﺖ. ﺑﺮرﺳﯽ دادهﻫﺎي ﺑﺎرﻧﺪﮔﯽ ﻣﺎﻫﺎﻧﻪي ﻣﻮردﻣﻄﺎﻟﻌﻪ ﻧﺸﺎن داد ﮐﻪ ﺧﻮدﻫﻤﺒﺴﺘﮕﯽ ﻣﻌﻨﯽداري در ﺗﺄﺧﯿﺮﻫﺎي 36 ،24 ،12 و 48 ﻣﺎﻫﻪ وﺟﻮد دارد. ﻧﺘﺎﯾﺞ ﻣﺮﺑﻮط ﺑﻪ ﺑﺎرﻧﺪﮔﯽ ﻣﺎﻫﺎﻧﻪي ﭘﯿﺶﺑﯿﻨﯽﺷﺪه در 5 ﺳﺎل آﯾﻨﺪه )2020 ﺗﺎ 2024( ﺑﺎ اﺳﺘﻔﺎده از ﺑﻬﺘﺮﯾﻦ روش اﯾﺴﺘﺎﺳﺎزي و ﺑﻬﺘﺮﯾﻦ ﻣﺪل ﺳﺮيﻫﺎي زﻣﺎﻧﯽ در اﯾﺴﺘﮕﺎه ﺳﯿﻨﻮﭘﺘﯿﮏ اردﺑﯿﻞ ﻧﺸﺎن داد ﮐﻪ ﻣﻘﺪار ﺑﺎرﻧﺪﮔﯽ ﺳﺎﻻﻧﻪ در 4 ﺳﺎل از 5 ﺳﺎل آﯾﻨﺪه ﻧﺴﺒﺖ ﺑﻪ ﻣﯿﺎﻧﮕﯿﻦ ﺑﺎرﻧﺪﮔﯽ 20 ﺳﺎل ﮔﺬﺷﺘﻪ، ﺑﯿﻦ 3 ﺗﺎ 17 درﺻﺪ ﮐﺎﻫﺶ ﺧﻮاﻫﺪ ﯾﺎﻓﺖ ﮐﻪ ﺑﯿﺶﺗﺮﯾﻦ ﮐﺎﻫﺶ ﻣﺮﺑﻮط ﺑﻪ ﺳﺎل 2022 ﻣﯿﻼدي اﺳﺖ. ﻣﻘﺪار ﺑﺎرﻧﺪﮔﯽ ﻓﻘﻂ در ﺳﺎل 2023 ﻣﻌﺎدل 0/3 درﺻﺪ اﻓﺰاﯾﺶ ﺧﻮاﻫﺪ ﯾﺎﻓﺖ.

Rainfall is the main source of the available water for human. Predicting the amount of the future rainfall is useful for informed policies, planning and decision making that will help potentially make optimal and sustainable use of available water resources. The main aim of this study was to investigate the trend and forecast monthly rainfall of selected synoptic station in Ardabil province using the best models of stochastic time series models. In this study, monthly rainfall for the next 5 years (2020 to 2024 AD) in the study area was predicted using different models of ARIMA family time series. Non-parametric Kendall- test was used to ensure the existence of the trend and the correlation diagram (ACF) was used to ensure the existence of seasonal changes in the time series. The best precipitation forecasting model in each of the 5 methods used for stabilization, was selected based on the values ​​of the model parameters, AIC criteria and correlation coefficient. The best static method and the best predictor model were used to predicte the next 5 year monthly rainfall. The results of man -Kendal test showed that the monthly rainfall data of Ardabil Synoptic Station had a decreasing trend (Z = 0.6119), but this trend was not significant at 95% confidence level. Study of the monthly rainfall data showed that there was a significant correlation between 12, 24, 36 and 48 month delays. The results of the monthly rainfall forecasting for the next five years (2020 to 2024) using the best static method and the best time series model in Ardabil Synoptic Station showed that the annual rainfall should decrease in 4 years of the next 5 years compared to the average of the 20 past years by 3 to 17 percent, the biggest drop since 2022. Rainfall will increase by 0.3% only in 2023.