8064

7499

كارشناسي ارشد

معدن

اصفهان: دانشگاه صنعتي اصفهان، دانشكده معدن

1391

ده،103ص.: مصور،جدول،نمودار

ص.ع.به فارسي و انگليسي

نادر فتحيان پور، علي احمدي

مجيد سرتاج، احمدرضا مختاري

4/8/92

هستي هاشمي نژاد، مرتضي طبايي

مهندسي معدن

ID7499

به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي

A Comparative study between estimation of Nitrate concentration in Arak alluvium groundwater using ANNs SVR MODFLOW and Geostatistical methods Soroush Maghsoudy soroush maghsoudy@yahoo com Date of Submission Department of Mining Engineering Isfahan University of Technology Isfahan Iran Degree M Sc Language Farsi Supervisors Dr N Fathianpour fathian@cc iut ac ir Dr A Ahmadi a ahmadi@cc iut ac ir Abstract Agriculture is the main non point polluter of groundwater in irrigated areas as fertilizers and other agrochemicals are the main contaminants in the water that drains out of the root zone to recharge the aquifer Nitrates from fertilizers dissolved in percolation losses from fields The concentration of nitrates in the percolated water depends on the distributed field water and nitrogen balances over the area Its concentration in the groundwater depends on the total recharge pollution loading groundwater flow and solute transport within the aquifer In this study application of intelligent techniques such as Neural networks ANN Support vector machines SVR and multiple regression for contaminant transport modeling is evaluated and the results compared with Modflow models and geostatistical simulations For case study select the Arak plain that placed in central of Markazi province of Iran and distribution of nitrate in groundwater in this plain is investigated We have the result of sampling of selected wells for different seasons from autumn to summer in the amount of nitrate pH electrical conductivity total dissolved solids heavy metals iron total coliform BOD and COD In summer duo to validation of the models sampling of the wells is repeated and accuracy of the modeling results is evaluated Also due to the need to identify and determine the type of aquifer and bedrock depth Geophysical Studies have been done and geoelectric sounding designed and performed With using neural networks support vector machines and multivariate regression models nitrate concentrations in the aquifer as a function of input variables was obtained Results shows that the estimates of neural networks method with more than accuracy in compared with and correlation between observation and calculation values in multiple regression method and SVR method has better alignment And this model can be used to estimate the concentration of nitrate in Arak aquifer In addition performing geostatistical simulation method SGS estimated the spatial distribution of nitrate in aquifer Modeling and prediction results shows that Nitrate concentrations in seasonal rainfall expanding from local to regional areas zone of the aquifer Also plume of contamination moved to the East and South of the plain Result of samples taken in the plain in the summer of confirmed the accuracy of the modeling results and shows the high concentration of nitrate and other pollutants such as TDS and EC that may reveal the need for greater attention to the pollution of the aquifer Keywords Nitrate pollution Contaminant transport ANNs SVR Numerical Modeling SGSIM PDF created with pdfFactory trial version www pdffactory com

نادر فتحيان پور، علي احمدي

مجيد سرتاج، احمدرضا مختاري

هستي هاشمي نژاد، مرتضي طبايي