Iranian Jornal of Watershed Management Science&Engineering

fa بررسی کیفیت آب سطحی و مدل سازی شاخص کیفیت آب با استفاده از SVR، SVR-GWO و SVR-PSO (منطقه مورد مطالعه: حوزه آبخیز کشکان، لرستان) Assessment of surface water quality and Water quality Index (WQI) modeling using SVR, SVR-GWO and SVR-PSO (Case study: Khorramabad, Biranshahr and Alashtar watersheds, Lorestan province) تخصصي Special پژوهشي Research <a name="_Hlk191530792">یکی از شاخص‌های مهم بررسی کیفیت آب، شاخص کیفیت آب (</a>WQI) است. مدل‌ها و روش‌های مختلفی برای تعیین و تخمین این شاخص وجود دارد که هوش مصنوعی از روش‌های جدید و امروزی است که برای تعیین شاخص کیفیت آب مورد استفاده قرار می‌گیرد. هدف این تحقیق مدل‌سازی شاخص کیفیت آب بخشی از حوزه‌آبخیز کشکان با استفاده از مدل‌های SVR، SVR-GWO و SVR-PSO بوده است. در این تحقیق در محل ایستگاه‌های هیدرومتری (خروجی حوضه‌ها) خصوصیات کیفی آب در دوره 10 ساله (1402-1393) اندازه‌گیری شد. داده‌های استفاده شده در این تحقیق برای مدل‌سازی کیفیت آب شامل TDS، EC، PH، CO3، HCO3، Cl، SO4، Ca، Mg، Na و K بودند. در ادامه برای بخش مدل‌سازی، خصوصیات کیفی آب به‌عنوان داده‌های ورودی و شاخص WQI به‌عنوان داده خروجی به مدل معرفی شد که 70 درصد کل داده‌ها برای مرحله آموزش و 30 درصد باقی‌مانده برای صحت‌سنجی مدل استفاده شدند. برای ارزیابی مدل‌ها از معیارهای سنجش خطا شامل MAE، C.C، RMSE، نمودارهای تیلور و ویولین باکس استفاده شد. نتایج این تحقیق نشان داد که مدل SVR_GWO بر اساس میزان معیارهای ارزیابی خطا C.C، MAE و RMSE در مرحله آموزش به‌ترتیب برابر 986/0، 966/0 و 792/0 و در مرحله آزمایش به‌ترتیب برابر 936/0، 871/0 و 727/3 نسبت به نتایج مدل‌های  SVRو SVR_PSO  عملکرد بهتری داشته است و به‌عنوان مدل برتر برای تخمین میزان شاخص کیفیت آب (WQI) انتخاب شد. همچنین نتایج مربوط به نمودار‌های تیلور و ویولن باکس نشان دادند که مدل SVR-GWO دارای بیشترین کارایی برای تعیین شاخص کیفیت آب (WQI) است. علاوه بر این مشخص شد که مدل SVR نسبت به سایر مدل‌ها دارای عملکرد ضعیف‌تری بود. به‌طور کلی با توجه به نتایج این تحقیق، می‌توان بیان نمود کهروش‌های هوش مصنوعی با توجه به میزان هزینه و زمان کم، در ارزیابی و پیش‌بینی کیفیت آب سطحی از قدرت بالا و عملکرد خوبی برخورداراند. همچنین می‌توان از این شاخص برای مدیریت بهینه کیفیت آب‌های سطحی و زیرزمینی در حوزه‌های‌آبخیز استفاده نمود.   Extended Abstract Introduction Water quality assessment and prediction play crucial roles in ensuring the sustainability and safety of freshwater resources. That with the rapid urbanization, industrialization and agricultural activities, large amount of river pollution from point or non-point sources have posed increasingly challenges over the world. The ongoing deterioration of water quality has put safe water supplies at risk, causing water pollution incidents and damaging aquatic ecosystems, especially in coastal cities with more prosperous economic development and intensive anthropogenic activities. Throughout human history, access to clean water has been a fundamental necessity. The water supply is essential to human well-being and is seen as a basic prerequisite for human activity and economical utility. Rivers, considered to be the most productive ecosystems and is essential for eliminating and neutralizing different types of contaminants. Thus, it is most frequently mentioned as the surface water sources that are exploited because of its accessibility and quantity, which have accelerated human development and population increase close to water channels. One of the key indicators used to assess water quality is the Water Quality Index (WQI). Various models and methods exist for calculating and estimating this index, with artificial intelligence emerging as a modern and effective approach in recent years. This study aims to model the WQI in the Kashkan watershed using three models: SVR, SVR-GWO, and SVR-PSO. Materials and Methods To campare the result of the proposed models’ performance, the Cham Anjir, Kaka Reza and Sarab Syed Ali hydrometry stations in Khorramabad, Biranshahr and Alashtar sub-watersheds (a part of Kashkan watershed) in western of Iran, is used as a case study area. The geographic coordinates of the Cham Anjir, Bahram Jo, Kaka Reza and Sarab Syed Ali are 48° 15 '34" E 33° 26' 55" N, 48° 17' 45"E 33° 34' 8" N, 48° 13' 51" E 33° 43' 39" N and 48° 12' 14" E 33° 44' 55" N, respectively. The studied area has a semiarid climate with a mean annual rainfall Less than 500 mm. The studied area has a maximum elevation of 3578 m in Alashtar watershed and the minimum elevation of 1158 m in khorramAbad watershed. Water quality parameters were collected over a ten-year period (2014–2023) at hydrometric stations located at the outlet of sub-watersheds within the Kashkan basin. The input data for the modeling process included TDS, EC, pH, CO₃, HCO₃, Cl, SO₄, Ca, Mg, Na, and K, which were used to calculate the WQI. For model development, 70% of the data was used for training and the remaining 30% for validation. The performance of the models was evaluated using error metrics such as MAE, RMSE, correlation coefficient (CC), and visual tools including Taylor and violin box plots. Results and Discussion The results showen that the SVR-GWO model outperformed the SVR and SVR-PSO models. In the training phase, it achieved MAE = 0.986, CC = 0.966, and RMSE = 0.792, while in the testing phase, the values were MAE = 0.936, CC = 0.871, and RMSE = 3.727, respectively. These findings indicate that SVR-GWO was the most accurate and reliable model for estimating the Water Quality Index (WQI). Moreover, Thus, the Taylor diagram also concluded that SVR-GWO model was the most reliable soft computing technique for the prediction of WQI. Thus, the violin cum box plot also supported the conclusion that SVR-GWO model had an edge on SVR and SVR-PSO in the prediction of WQI. Conversely, the basic SVR model showed comparatively weaker performance. In general, the SVR-GWO model, is the powerful model for the prediction of Water Quality Index (WQI). Therefore, according to the obtained results from this research, these optimal models can be used to costly and time-consuming tasks of the estimation of Water Quality Index (WQI) from river. Also, these models can be used to estimate the Water Quality Index (WQI) of nearby rivers by/without hydrometry station for the management of the quantity and quality of surface water. In such a case, soft computing techniques (SVR-GWO model) can be used to assess water quality. Conclusion The present study focused on the development of a SVR, SVR-GWO, and SVR-PSO models to estimate the water quality index(WQI). For this purpose, the quality parameters data of the Cham Anjir, Sarab Syed Ali and Kaka Reza stations in Khorramabad, Alashtar and Biranshahr sub-watersheds composed of Water Quality Index (WQI), were used. In general, The major conclusions of the study are as follows: Among those models with the highest performance, the SVR-GWO has the highest performance in both testing and training phases. -The SVR-GWO predicted data are closer to observational data compared with the other model’s output data. Besides, the SVR-GWO is the nearest predicted model with observational data. The SVR model is one of the most extensively used data driven models in the natural literature, while the usages of other data-driven models are comparatively lesser. Also, the structure of the SVR-GWO is very simple and very less time consumable. Thus, the SVR-GWO model can be useful in the Water Quality Index (WQI) modeling not only for accuracy but also for its time-saving nature and simple structure compared with other models. Overall, the findings suggest that artificial intelligence techniques, due to their cost-effectiveness and speed, offer strong capabilities for assessing and predicting surface water quality. The WQI can also serve as a practical tool for the optimal management of surface and groundwater resources in watershed areas.   استان لرستان, حوزه آبخیز کشکان, شاخص کیفیت آب (WQI), هوش مصنوعی. Lorestan Province, Kashkan Watershed, Water Quality Index (WQI), Artificial Intelligence 0 0 http://jwmsei.ir/browse.php?a_code=A-10-728-4&slc_lang=fa&sid=1 Alireza Sepahvand علیرضا سپه وند sepahvand.a@lu.ac.ir 100319475328460019013 100319475328460019013 Yes Lorestan University دانشگاه لرستان Negar Arjmand نگار ارجمند negar.arjmand87@gmail.com 100319475328460019014 100319475328460019014 No Lorestan University دانشگاه لرستان Nasrin Beiranvand نسرین بیرانوند beiranvand.n76@gmail.com 100319475328460019015 100319475328460019015 No Lorestan University دانشگاه لرستان