بررسي ويژگي‌هاي‌ خاك‌ در مناطق مستعد توليد گرد و غبار در استان خوزستان با روش طيف‌سنجي امواج مرئي و مادون‌قرمز نزديك(VNIR)

Investigating soil Properties in susceptible areas of dust production in Khuzestan province by visible and wave-near infrared spectroscopy

اﻣﺮوزه اﺳﺘﻔﺎده از ﻃﯿﻒﺳﻨﺠﯽ در ﺑﺮآورد وﯾﮋﮔﯽ ﻫﺎي ﺧﺎك ﮐﺎرﺑﺮد ﮔﺴﺘﺮده اي ﯾﺎﻓﺘﻪ اﺳﺖ. ﻫﺪف از اﯾﻦ ﭘﮋوﻫﺶ، اﺳﺘﻔﺎده از اﻣﻮاج ﻣﺮﺋﯽ و ﻣﺎدون ﻗﺮﻣﺰ ﻧﺰدﯾﮏ در ﺑﺮآورد ﺧﺼﻮﺻﯿﺎت ﺧﺎك در ﻣﻨﺎﻃﻖ ﻣﺴﺘﻌﺪ ﺗﻮﻟﯿﺪ ﮔﺮد و ﻏﺒﺎر اﺳﺘﺎن ﺧﻮزﺳﺘﺎن اﺳﺖ. ﺗﻌﺪاد 142 ﻧﻤﻮﻧﻪ ﺧﺎك از ﻣﻨﺎﻃﻖ ﻣﺴﺘﻌﺪ ﺗﻮﻟﯿﺪ ﮔﺮد و ﻏﺒﺎر ﻣﻮرد ﺗﺠﺰﯾﻪ و ﺗﺤﻠﯿﻞ ﻗﺮار ﮔﺮﻓﺖ و ﻣﻘﺪار ﮐﺮﺑﻨﺎت ﮐﻠﺴﯿﻢ ﻣﻌﺎدل، ﮔﭻ، ﮐﺮﺑﻦ آﻟﯽ و ﻧﯿﺘﺮوژن ﮐﻞ ﺧﺎك اﻧﺪازه ﮔﯿﺮي ﺷﺪ. از ﻣﺪل ﻫﺎي رﮔﺮﺳﯿﻮن ﺧﻄﯽ، ﮐﻤﯿﻨﻪ ﻣﺮﺑﻌﺎت ﺟﺰﺋﯽ )PLSR( و رﮔﺮﺳﯿﻮن ﻣﺆﻟﻔﻪ اﺻﻠﯽ )PCR( ﺑﺮاي ﺑﺮآورد اﯾﻦ ﺧﺼﻮﺻﯿﺎت در ﺧﺎك اﺳﺘﻔﺎده ﺷﺪ. ﺳﻪ روش ﺑﺎزﺗﺎب ﻃﯿﻒ اﺻﻠﯽ و روش ﻫﺎي ﭘﯿﺶ ﭘﺮدازش ﻣﺸﺘﻖ اول و ﻣﺸﺘﻖ دوم در دو ﻣﺪل رﮔﺮﺳﯿﻮﻧﯽ PCR و PLSR ﻣﻮرد ﻣﻘﺎﯾﺴﻪ ﻗﺮار ﮔﺮﻓﺖ. ﻧﺘﺎﯾﺞ ﻧﺸﺎن داد ﮐﻪ ﻣﺪل PLSR در ﺣﺎﻟﺖ اﺳﺘﻔﺎده از روش ﭘﯿﺶ ﭘﺮدازش ﻣﺸﺘﻖ دوم ﺑﺎﻋﺚ ﮐﺎﻫﺶ اﺧﺘﻼل ﻃﯿﻒ ﻫﺎي ﺑﺎزﺗﺎﺑﯽ ﻧﻤﻮﻧﻪ ﻫﺎي ﺧﺎك ﺷﺪ، ﺑﺮاي ﺳﻪ ﭘﺎراﻣﺘﺮ ﮐﺮﺑﻨﺎت ﮐﻠﺴﯿﻢ ﻣﻌﺎدل، ﮐﺮﺑﻦ آﻟﯽ و ﻧﯿﺘﺮوژن ﮐﻞ ﺑﻪ ﺗﺮﺗﯿﺐ ﺑﺎ ﺿﺮاﯾﺐ ﺗﻌﯿﯿﻦ 0/95 و 0/92 و 0/81 ﺑﯿﺸﺘﺮﯾﻦ دﻗﺖ ﺑﺮآورد را ﻧﺸﺎن داد. ﺑﺮاي ﮔﭻ، ﺑﯿﺸﺘﺮﯾﻦ دﻗﺖ در ﺣﺎﻟﺖ ﻣﺸﺘﻖ اول ﺑﺎ ﺿﺮﯾﺐ ﺗﻌﯿﯿﻦ 0/87 ﺑﻮد. ﻧﺘﺎﯾﺞ اﯾﻦ ﻣﻄﺎﻟﻌﻪ، ﺑﯿﺎﻧﮕﺮ آن اﺳﺖ ﮐﻪ اﺳﺘﻔﺎده از روش ﻃﯿﻒ ﺳﻨﺠﯽ در ﺑﺮآورد ﺧﺼﻮﺻﯿﺎت ﺧﺎك در ﻣﻨﺎﻃﻖ ﻣﺴﺘﻌﺪ ﺗﻮﻟﯿﺪ ﮔﺮد و ﻏﺒﺎر اﺳﺘﺎن ﺧﻮزﺳﺘﺎن دﻗﺖ ﻣﻨﺎﺳﺒﯽ دارد و ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ وﺳﻌﺖ اﯾﻦ ﮐﺎﻧﻮن ﻫﺎ و ﺳﺮﻋﺖ ﻋﻤﻞ و ارزان ﺗﺮ ﺑﻮدن، اﯾﻦ روش ﻣﯽﺗﻮاﻧﺪ در ارزﯾﺎﺑﯽ ﺧﺼﻮﺻﯿﺎت ﺧﺎك اﯾﻦ ﻣﻨﺎﻃﻖ ﻣﻮرد اﺳﺘﻔﺎده ﻗﺮار ﮔﯿﺮد.

Nowadays spectroscopy is used for estimating soil properties. The main objective of this research was to estimate some soil properties of Susceptible Areas of Dust Production using visible and near infrared. Therefore 142 soil sample of Susceptible Areas of Dust Production were Collected and analyzed. Equivalent calcium carbonate, gypsum, organic carbon and nitrogen of soil sample were measured and linear regression models of PCR and PLSR were used to estimate these properties .Three methods of reflection of the main spectra and pre-process methods of first derivative and second derivative were compared in two regression models of PCR and PLSR. The results showed that the PLSR method is more accurate than the PCR model for estimating soil properties. The PLSR model in the pre-processing second derivative with noise reduction, showed the highest accuracy for equivalent calcium carbonate, organic carbon and total nitrogen with the coefficient of determination as 0.95, 0.92 and 0.81, respectively. For gypsum, the highest accuracy in the first derivative with the coefficient of determination was 0.87. The results of this research revealed the use of spectroscopy in estimating soil properties of dust production-prone areas in Khuzestan province have an appropriate accuracy; and due to the extent of these areas and the speed of operation and cheapness of this method, it can be used to predict the amount of soil properties in these areas.