ارزيابي عملكرد مدل هاي AWBM، Sacramento و SimHyd در شبيه سازي رواناب حوضه امامه با استفاده از بهينه ساز واسنجي خودكار الگوريتم ژنتيك

Performance Evaluation of AWBM, Sacramento and SimHyd models in Runoff Simulation of the Amameh Watershed using Automatic Calibration Optimization Method of Genetic Algorithm

ﯾﮑﯽ از روشﻫﺎي ﻣﻌﺘﺒﺮ در ﺷﺒﯿﻪﺳﺎزي ﻓﺮآﯾﻨﺪ ﭘﯿﭽﯿﺪه و ﻏﯿﺮﺧﻄﯽ ﺑﺎرش–رواﻧﺎب اﺳﺘﻔﺎده از ﻣﺪلﻫﺎي ﻫﯿﺪروﻟﻮژﯾﮑﯽ اﺳﺖ. ﻫﺪف ﻣﻘﺎﻟﻪ ﺣﺎﺿﺮ ﺑﺮرﺳﯽ ﮐﺎراﯾﯽ ﺳﻪ ﻣﺪل ﺑﺎرش-رواﻧﺎب ﯾﮑﭙﺎرﭼﻪ و ﻣﻔﻬﻮﻣﯽ Sacramento ،AWBM و SimHyd در ﺷﺒﯿﻪ-ﺳﺎزي رواﻧﺎب روزاﻧﻪ در ﺧﺮوﺟﯽ ﺣﻮﺿﻪ آﺑﺮﯾﺰ ﻣﻌﺮف اﻣﺎﻣﻪ ﺑﺎ ﮐﺎرﺑﺮد ﺑﻬﯿﻨﻪﺳﺎز واﺳﻨﺠﯽ ﺧﻮدﮐﺎر اﻟﮕﻮرﯾﺘﻢ ژﻧﺘﯿﮏ اﺳﺖ. ﻣﺸﺎﺑﻪ ﺳﺎﯾﺮ ﻣﺪلﻫﺎي ﻫﯿﺪروﻟﻮژﯾﮑﯽ، در ﻫﺮ ﺳﻪ ﻣﺪل ﻣﺬﮐﻮر ﻣﺤﺪوده ﺗﻐﯿﯿﺮ ﭘﺎراﻣﺘﺮﻫﺎ زﯾﺎد اﺳﺖ و ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ دﺷﻮاري واﺳﻨﺠﯽ ﺑﺎ روشﻫﺎي ﻣﺘﺪاول ﻣﺒﺘﻨﯽ ﺑﺮ ﺳﻌﯽ و ﺧﻄﺎ، در اﯾﻦ ﻣﻘﺎﻟﻪ ﮐﺎرﺑﺮد روشﻫﺎي ﺑﻬﯿﻨﻪﺳﺎزي ﺑﻪ ﻣﻨﻈﻮر واﺳﻨﺠﯽ ﺧﻮدﮐﺎر ﻣﺪلﻫﺎي ﻫﯿﺪروﻟﻮژﯾﮑﯽ ﺑﺮرﺳﯽ ﺷﺪ. ﺗﻬﯿﻪ ﻧﻘﺸﻪﻫﺎي ﻣﻮرد ﻧﯿﺎز در ﻣﺤﯿﻂ ﻧﺮماﻓﺰار ﺳﺎﻣﺎﻧﻪ اﻃﻼﻋﺎت ﺟﻐﺮاﻓﯿﺎﯾﯽ ﻧﺴﺨﻪ 10,4,1 اﻧﺠﺎم ﮔﺮﻓﺖ و دادهﻫﺎي ﺑﺎرش، ﺗﺒﺨﯿﺮ و ﺗﻌﺮق ﭘﺘﺎﻧﺴﯿﻞ و رواﻧﺎب ﻣﺸﺎﻫﺪاﺗﯽ روزاﻧﻪ ﺳﺎلﻫﺎي 2001-2005 ﺑﺮاي واﺳﻨﺠﯽ و ﺳﺎلﻫﺎي 2007-2006 ﺑﺮاي ﺻﺤﺖﺳﻨﺠﯽ در اﻣﺮ ﺷﺒﯿﻪﺳﺎزيﻫﺎ اﺳﺘﻔﺎده ﺷﺪ. ﻣﻌﯿﺎرﻫﺎي ارزﯾﺎﺑﯽ ﺷﺎﻣﻞ ﺿﺮﯾﺐ ﻧﺶ-ﺳﺎﺗﮑﻠﯿﻒ )NSE(، ﺿﺮﯾﺐ ﺗﺒﯿﯿﻦ )R2( و ﺿﺮﯾﺐ ﻣﺠﺬور ﻣﯿﺎﻧﮕﯿﻦ ﻣﺮﺑﻌﺎت ﺧﻄﺎ )RMSE( ﺑﺮاي ارزﯾﺎﺑﯽ ﻣﺪلﻫﺎ اﺳﺘﻔﺎده ﺷﺪ. ﻧﺘﺎﯾﺞ آﻣﺎري و ﮔﺮاﻓﯿﮑﯽ واﺳﻨﺠﯽ و ﺻﺤﺖﺳﻨﺠﯽ ﺣﺎﮐﯽ از ﻋﻤﻠﮑﺮد ﺑﻬﺘﺮ ﻣﺪل SimHyd ﻧﺴﺒﺖ ﺑﻪ دو ﻣﺪل دﯾﮕﺮ ﺑﺎ داﺷﺘﻦ ﺿﺮﯾﺐ ﻧﺶﺳﺎﺗﮑﻠﯿﻒ 0/575 و 0/731، ﺗﺒﯿﯿﻦ 0/61 و 0/80 و ﻣﺠﺬور ﻣﯿﺎﻧﮕﯿﻦ ﻣﺮﺑﻌﺎت ﺧﻄﺎ 1/033 و 0/829 ﺑﻪ ﺗﺮﺗﯿﺐ در دوره واﺳﻨﺠﯽ و ﺻﺤﺖﺳﻨﺠﯽ ﺑﺎ ﮐﺎرﺑﺮد ﺑﻬﯿﻨﻪﺳﺎز واﺳﻨﺠﯽ ﺧﻮدﮐﺎر اﻟﮕﻮرﯾﺘﻢ ژﻧﺘﯿﮏ و اﻧﻄﺒﺎق ﮔﺮاﻓﯿﮑﯽ ﺧﻮب ﺑﺎ ﻣﻘﺎدﯾﺮ ﻣﺸﺎﻫﺪاﺗﯽ اﺳﺖ. ﻫﻤﭽﻨﯿﻦ ﻣﺪلﻫﺎي AWBM و Sacramento ﻧﯿﺰ ﻧﺘﺎﯾﺞ آﻣﺎري و ﮔﺮاﻓﯿﮑﯽ رﺿﺎﯾﺖﺑﺨﺶ و ﻣﻄﻠﻮب در ﺣﻮﺿﻪ ﻣﻨﺘﺨﺐ داﺷﺘﻪ و ﮐﺎراﯾﯽ ﺧﻮب ﺑﻬﯿﻨﻪﺳﺎز واﺳﻨﺠﯽ ﺧﻮدﮐﺎر اﻟﮕﻮرﯾﺘﻢ ژﻧﺘﯿﮏ را ﻣﻮرد ﺗﺄﮐﯿﺪ ﻗﺮار ﻣﯽدﻫﻨﺪ.

One of the valid methods for simulation of the complex and non-linear process of rainfall-runoff is to use hydrological models. The purpose of this paper is to investigate the performance of three conceptual and lumped rainfall-runoff models; AWBM, Sacramento and SimHyd for simulating daily runoff at the outlet of the Amameh watershed using automatic calibration optimization genetic algorithm. Similar to other hydrological models, the range of parameters' variations is high in all three models and due to the difficulty of calibrating with trial and error-based methods, in this paper, the use of automatic calibration optimization methods for the hydrological models investigated. Preparation of required maps carried out by the GIS software version 10.4.1. Daily rainfall, potential evapotranspiration and observation runoff data of 2001-2005 used for calibration and 2006- 2007 data for simulations verificasion. The evaluation criteria including Nash-Sutcliff coefficient (NSE), coefficient of determination (R2) and root mean square error (RMSE) used to evaluate the proposed models. The statistical and graphical results of calibration and verification steps showed that SimHyd model performed better than the other two models with Nash-Sutcliff coefficient of 0.575 and 0.731, determination coefficient of 0.61 and 0.80 and the root mean square error of 1.033 and 0.829 respectively in the calibration and verification periods, using the automatic calibration optimization of the genetic algorithm and good graphical matching with the observational values. Also, AWBM and Sacramento models have satisfactory and desirable graphical and statistical results in the selected watershed and emphasize the good performance of automatic calibration optimization of the genetic algorithm.