Assessment of daily suspended sediment laod in Forest Watershed

معتمدنيا, محبوبه and كريمي زارچي, كمال (2018) Assessment of daily suspended sediment laod in Forest Watershed. In: The 13th National Conference on Watershed Management Science & Engineering of Iran and The 3rd National Conference on Conservation of Natural Resources and Environment, 3-4 October 2018, Ardabil, Iran.

[img] Text (برآورد بار رسوب معلق روزانه در حوزة آبخيز جنگلي کجور)
wms97_UMA2018P55T12_FullPaperSubmission.pdf

Download (580kB)

Abstract

Erosion and sediment transport phenomenon is one of the most complicate problems in hydrodynamics which is very important in water projects. In this regard, there are many methods and equations with enormous errors and inefficincy in estimation of sediment load because of which, the researchers got encouraged to use artificial neural network. In other words, this method is cheaper, faster and easier than other methods compared to existing methods. For this reason, in this search used artificial neural network in two methods viz. multi layer perceptoron (MLP) with back propagation algorithm and radial basis function to estimate suspended sediment load in Kojour Forest Watershed in Mazandaran Province. The effective water discharge on sediment load, five past successive, individual and continous daily discharge and their combined effects were then determined after normalizing data sets. The results showed that the range of RMSE in MLP were 0.70 and 1.63 with average of1.24 while it was 0.98 and 1.77 with of 1.40 in RBF. The MAE was also 0.73 and 0.76 in MLP and RBF, respectively. The results ultimately indicated that the MLPwas more efficient and superior than RBF. KeyWords: Artificial neural network, Kojour watershed, Soil erosion, Suspunded sediment load

Item Type: Conference or Workshop Item (Poster)
Persian Title: برآورد بار رسوب معلق روزانه در حوزة آبخيز جنگلي کجور
Persian Abstract: پديده¬ فرسايش و انتقال رسوب يكي از پيچيده¬ترين مسايل هيدروديناميك مي¬باشد كه در طرح¬هاي آبي اهميتي بسيار دارد. در اين راستا روش¬ها و معادلات زيادي وجود دارد كه به دليل خطاهاي زياد و عدم كفايت براي تخمين رسوب انتقالي توسط رودخانه محققان را به استفاده از شبكه ترغيب ساخته است. به خاطر همين از شبكه عصبي به دو روش شبكه پرسپترون چند لايه و با الگوريتم پس انتشار خطا و تابع پايه شعاعي به منظور تخمين بار رسوب معلق در حوزه آبخيز جنگلي كجور در استان مازندران استفاده شد. در اين تحقيق به منظور بررسي و تعيين بيش¬ترين تاثير دبي آب بر رسوب، تاثير روزهای متوالي دبي آب بر رسوب را به صورت هر روز جداگانه و به صورت مشترک و تاثيرات هم¬زمان مورد بررسی قرار گرفت. نتايج نشان داد که دامنه تغييرات مجموع ميانگين خطا در روش MLP برابر با 70/0 و 63/1 و ميانگين آن 24/1 مي¬باشد. در حالي كه دامنه تغييرات مجموع ميانگين خطا در روش RBFبرابر با 98/0 و 77/1 و ميانگين آن 40/1 مي¬باشد. در مورد ميانگين مطلق خطا در روش MLPو RBF به ترتيب 73/0 و 76/0 مي¬باشد كه تمامي آماره¬ها دلالت بر كارايي و برتري روش MLP بر روش RBF دارد. واژگان كليدي: بار رسوب معلق، شبكه عصبي، حوزه آبخيز كجور، فرسايش خاک
Subjects: Divisions > Conferences > The 13th National Conference on Watershed Management Science & Engineering of Iran and The 3rd National Conference on Conservation of Natural Resources and Environment
Conferences > The 13th National Conference on Watershed Management Science & Engineering of Iran and The 3rd National Conference on Conservation of Natural Resources and Environment
Divisions: Conferences > The 13th National Conference on Watershed Management Science & Engineering of Iran and The 3rd National Conference on Conservation of Natural Resources and Environment
Subjects > Conferences > The 13th National Conference on Watershed Management Science & Engineering of Iran and The 3rd National Conference on Conservation of Natural Resources and Environment
Date Deposited: 17 Oct 2018 16:27
Last Modified: 17 Oct 2018 16:27
URI: http://repository.uma.ac.ir/id/eprint/916

Actions (login required)

View Item View Item