بررسي و برآورد تاثير عوامل مختلف بر تغييرات عمق آبخوان با استفاده از الگوريتم درختي (مطالعه موردي: دشت تيروان و كرون)

Investigation and Prediction of Impact of Different Factors on Aquifer Depth Change Using Tree Algorithm (Case Study: Tirvan and Carvan Plain

ﻫﺪف از اﯾﻦ ﭘﮋوﻫﺶ، اﺳﺘﻔﺎده از اﻟﮕﻮرﯾﺘﻢ درﺧﺘﯽ دادهﮐﺎوي، ﺑﻪﻣﻨﻈﻮر ﻣﺪﯾﺮﯾﺖ ﻣﻨﺎﺳﺐ ﺑﺮ آﺑﺨﻮان دﺷﺖ ﺗﯿﺮوان و ﮐﺮون اﺳﺖ. در اﯾﻦ ﺧﺼﻮص، از ﻫﻔﺖ ﻋﺎﻣﻞ ﻣﺨﺘﻠﻒ اﻧﺴﺎﻧﯽ و ﻃﺒﯿﻌﯽ ﺗﺄﺛﯿﺮﮔﺬار ﺑﺮ ﺗﻐﯿﯿﺮات ﻋﻤﻖ آﺑﺨﻮان اﺳﺘﻔﺎده ﺷﺪ. در اﺑﺘﺪا، ﭘﯿﺶﺑﯿﻨﯽ ﺳﻪ اﻟﮕﻮرﯾﺘﻢ درﺧﺘﯽ CHAID ،CART و MP5 در ﺗﻐﯿﯿﺮات آﺑﺨﻮان ﺑﺎ اﺳﺘﻔﺎده از ﺷﺎﺧﺺﻫﺎي آﻣﺎري ﻣﻮرد ارزﯾﺎﺑﯽ ﻗﺮار ﮔﺮﻓﺘﻨﺪ. اﻟﮕﻮرﯾﺘﻢ CAHID ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﺿﺮﯾﺐ رﮔﺮﺳﯿﻮن ﺑﺮاﺑﺮ 0/82 و ﻣﺘﻮﺳﻂ ﻣﻄﻠﻖ ﺧﻄﺎ ﺑﺮاﺑﺮ 0/12 داراي ﻋﻤﻠﮑﺮد ﺑﻬﺘﺮي ﻧﺴﺒﺖ ﺑﻪ دو اﻟﮕﻮرﯾﺘﻢ CART و MP5 اﺳﺖ. ﺑﯿﺸﺘﺮﯾﻦ ﻣﻘﺪار ﺑﺎﻻآﻣﺪﮔﯽ آﺑﺨﻮان در ﻣﺎهﻫﺎي آذر، دي، ﺑﻬﻤﻦ، اﺳﻔﻨﺪ و در زﻣﺎﻧﯽ ﮐﻪ ﻣﻘﺪار ﺣﺠﻢ ﺑﺎرﻧﺪﮔﯽ ﺑﯿﻦ 0/08 ﺗﺎ 0/72 ﻣﯿﻠﯿﻮن ﻣﺘﺮ ﻣﮑﻌﺐ و درﺻﺪ رﻃﻮﺑﺖ ﻫﻮا ﺑﯿﺸﺘﺮ از 72 درﺻﺪ ﺑﻮده و ﻫﻤﭽﻨﯿﻦ، ﺑﯿﺸﺘﺮﯾﻦ ﻣﻘﺪار اﻓﺖ آﺑﺨﻮان در ﻣﺎهﻫﺎي ﺷﻬﺮﯾﻮر، ﻣﺮداد و در زﻣﺎﻧﯽ ﮐﻪ دﻣﺎي ﻫﻮا ﺑﯿﺶ از 25 درﺟﻪ ﺳﺎﻧﺘﯽﮔﺮاد و ﺣﺠﻢ آب ﺑﺮداﺷﺘﯽ از ﭼﺎهﻫﺎي ﮐﺸﺎورزي ﺑﯿﺶ از 1/32 ﻣﯿﻠﯿﻮن ﻣﺘﺮ ﻣﮑﻌﺐ ﺑﻮده اﺳﺖ، ﺑﻪوﺳﯿﻠﻪ ﻧﻤﻮدار درﺧﺘﯽ اﻟﮕﻮرﯾﺘﻢ CHAID ﭘﯿﺶﺑﯿﻨﯽ ﺷﺪ. از ﻋﻮاﻣﻞ ﻃﺒﯿﻌﯽ، دﻣﺎي ﻫﻮا و از ﻋﻮاﻣﻞ اﻧﺴﺎﻧﯽ، ﺣﺠﻢ آب ﺑﺮداﺷﺘﯽ از ﭼﺎه ﮐﺸﺎورزي ﺑﯿﺸﺘﺮﯾﻦ ﺗﺄﺛﯿﺮ را در ﺗﻐﯿﯿﺮات ﻋﻤﻖ آﺑﺨﻮان در دﺷﺖ ﻣﺬﮐﻮر داﺷﺘﻪ اﺳﺖ. دو ﻋﺎﻣﻞ درﺻﺪ رﻃﻮﺑﺖ ﻫﻮا و ﻣﻘﺪار ﺣﺠﻢ ﺑﺎرش، ﺗﻨﻬﺎ ﻋﻮاﻣﻠﯽ ﺑﻮدﻧﺪ ﮐﻪ راﺑﻄﻪ ﻣﺴﺘﻘﯿﻢ ﺑﺎ ﺑﺎﻻآﻣﺪﮔﯽ ﻋﻤﻖ آﺑﺨﻮان داﺷﺘﻪاﻧﺪ. ﺗﺄﺛﯿﺮﮔﺬارﺗﺮﯾﻦ ﻋﻮاﻣﻞ در ﭘﯿﺶ ﺑﯿﻨﯽ ﻣﻘﺪار ﺗﻐﯿﯿﺮات ﻋﻤﻖ آﺑﺨﻮان دﺷﺖ ﺗﯿﺮوان و ﮐﺮون ﺑﻪﺗﺮﺗﯿﺐ دﻣﺎي ﻫﻮا، ﺣﺠﻢ آب ﺑﺮداﺷﺘﯽ از ﭼﺎه ﮐﺸﺎورزي و ﺣﺠﻢ ﺑﺎرﻧﺪﮔﯽ و ﺑﻘﯿﻪ ﭘﺎراﻣﺘﺮﻫﺎ ﺗﻘﺮﯾﺒﺎ ﺗﺄﺛﯿﺮﺷﺎن ﺑﺎ ﻫﻢ ﺑﺮاﺑﺮ ﺑﻮده اﺳﺖ

In this study, with using the data mining algorithm prediction, more efficient management of the aquifer of Tirvan and Karvan can be done. Seven different human and natural factors affecting aquifer depth changes were used in this manuscript. Initially, the predictions of three tree algorithms CART, CHAID and MP5 in aquifer changes were evaluated using statistical indices. The CAHID algorithm performed better than the CART and MP5 algorithms with respect to the regression coefficient of 0.82 and absolute error mean of 0.12. The highest aquifer rise in December, January, February and March, when the amount of precipitation was between 0.08 to 0.72 million cubic meters and the air humidity percentage was more than 72% and also the highest aquifer drawdown in month August and September, when air temperature more than 25 centigrade and the volume of water discharged from agricultural wells were more than 1.32 million cubic meters, were predicted by the CHAID algorithm tree diagram. From natural factors, air temperature and human factors, the volume of water harvested from agricultural wells had the greatest impact on aquifer depth changes in the plain. The two factors of air humidity percentage and precipitation volume were the only factors that had a direct relationship with the aquifer depth elevation. The most influential factors in predicting the depth changes of the aquifer of Tirvan and Karvan were air temperature, volume of water harvested from agricultural wells, and Precipitation volume and other parameters, respectively.