اثرات تغيير پوشش و كاربري زمين در منطقه تجريش بر رژيم آبدهي رودخانه دربند

Effect of land cover and land use change in north ofTajrish zone on Darband discharge river

رودخانه هاي شهري به دليل افزايش جمعيت و تغييرات پوشش و كاربري زمين در حوضه هاي آبخيز آنها درمعرضآسيب هاي گوناگون مي باشند. رودخانه دربند نيز از اين قاعده مستثني نبوده و در طول نيم قرن گذشته توسعه شهري تهران، رژيم هيدرولوژيكي آن را تحت تاثير قرار داده است. در اين پژوهش با استفاده از عكس هاي هوايي سال هاي 1345 ، 1358 و تصاوير?ETM?^+وIkONOSسال 1389،ميزان تغييرات پوشش و كاربري اراضي در منطقه تجريش شناساييشده و بر اساس آن ها نقشه پوشش و كاربري زمين براي سه دوره زماني تهيه شده است و با به كارگيري نقشه هاي مذكور،با روش SCSبه تخمين ميزان عدديمنحني و نگهداشت سطحي آباقدام شده است. همچنين با بهره گيري از مدل هاي شبكه عصبي مصنوعي و ماشين بردار پشتيبان، مدل سازي رواناب و بارش انجام گرفته و از آزمون من كندال به عنوان روشي براي بررسي روند داده هاي دبي و بارش استفاده شده است. با آشكار سازي تغييرات پوشش و كاربري زمين در سه دوره زماني ،مشخص گرديد كه سطوح توليد كننده رواناب زياد در طول اين سه دوره افزايش يافته است. در همين راستا شماره منحني و ضريب رواناب از سال 1345 تا 1389 افزايش يافته و ضريب نگهداشت سطحي از 66/122در سال 1345 به 26/107در سال 1358 و 76/104 در سال 1389 كاهش يافته است. در حاليكه بارش روند افزايشي خاصي نداشته، دبي داراي روند صعودي بوده كه علت اصلي آن تغييرات پوشش و كاربري زمين بوده است.

Extended Abstract Introduction Urban rivers due to population growth, land use and land cover change in upstream watersheds are vulnerable to various problems such as pollution and drastic changes in river discharge regime. Darband River in northern Tehran, due to the increase in population is no exception from such problems and during the past half century, the hydrological regime of Darband River has affected by Tehran urban development. Methodology In this study, rate of land cover and land use change in Tajrish zone has been studied by using aerial photographs and images of ETM^+ and IkONOS in the years of 1966, 1989 and 2010. The curve number and maximum potential retentionof the watershed were estimated from these images and the SCS method was used to estimate the annual watershed runoff. In addition, a comparative assessment was done for Rainfall-Runoff modeling by using artificial neural network (ANN) and support vector machine (SVM) models. Mann-Kendall statistics was used to detect the trend in precipitation and discharge data. Significant land cover changes were detected during three considered periods by overlapping of the images in the GIS environment. Significant land use changes were observed during the time periods. Results and discussion In the study urban watershed, low potential runoff production area (like low density, semi-intensive, high Density vegetation, and wilderness area) is decreased during the considered years but regions with medium levels runoff production are increased. Curve number is increased from 67 in the year of 1966 to 70 in the year of 1989 and 71 in years of 2010. But the maximum potential retention level is decreased from 122 in the year of 1966 to 107 in the year of 1989 and 104 in the year of 2010. Runoff coefficient computed based on the SCS method increases during the year of 1966 to 1989. Mann-Kendall statistics was used to detect the trend in precipitation and discharge data during considered time period. It shows that despite of precipitation data has no trend, discharge indicates upward trend. The SVM Model performance is strongly depend on the Kernel function which selected in the model structure. In this model, parameters of C و ?و ? are need to optimize during the training mode. In this study the grid search method is used to find the best fitted values of these parameters. In the ANN model, the number of neurons in the hidden layer affected the model efficiency and need to optimized during the training mode. According tothe results of this table, no significant difference are observed between the support vector machinewith theparameters of C= 50and?= 0.001and?= 50 and the artificial neural network model with considering six neurons in the hidden layer. The low coefficient of determination suggest that low correlation between precipitation and runoff comes back to land use and land cover change via decreasing of Maximum Potential Retention level. Conclusion The results of this study indicated that the most changes happens in the urban areas with low levels runoff to high level runoff during three periods. Curve number and runoff coefficient increased from years of 1966 to 2010. In which maximum potential retention level indicated decreasing behavior. The precipitation data did not indicate trend whereas upward trend was observed for discharge data which denotes to the land cover changed from 1966 to 2010 affected the increasing of discharge in this urban watershed.