برآورد رواناب ذوب برف با سنجش از دور و مدل SRM در حوزه آبخيز سقز

Estimation of snowmelt runoff using remote sensing and SRM Model in Saqqez Watershed

اﻫﻤﯿﺖ ﺑﺮف و آب ﺣﺎﺻﻞ از ذوب آن در ﺗﺎﻣﯿﻦ ﻣﻨﺎﺑﻊ آب ﺑﺎﻋﺚ ﺷﺪه اﺳﺖ ﮐﻪ ﻣﻄﺎﻟﻌﺎت ﻓﺮاواﻧﯽ در ﻧﻘﺎط ﻣﺨﺘﻠﻒ در زﻣﯿﻨﻪ ﺑﺮفﺳﻨﺠﯽ و ﻣﺤﺎﺳﺒﻪ رواﻧﺎب ﺻﻮرت ﮔﯿﺮد. اﯾﻦ ﭘﮋوﻫﺶ در ﺣﻮزه آﺑﺨﯿﺰ ﺳﻘﺰ در اﺳﺘﺎن ﮐﺮدﺳﺘﺎن ﺻﻮرت ﮔﺮﻓﺘﻪ ﮐﻪ ﺑﺎ ﺗﻮﺟﻪ ﺑﻪ ﮐﻮﻫﺴﺘﺎﻧﯽ ﺑﻮدن ﻣﻨﻄﻘﻪ و اﻫﻤﯿﺖ ﺑﺎﻻي ﺑﺮف و آب ﺣﺎﺻﻞ از ذوب آن، ﺑﺮآورد رواﻧﺎب روزاﻧﻪ ﺣﻮﺿﻪ ﻣﻮرد ﺑﺮرﺳﯽ ﻗﺮار ﮔﺮﻓﺘﻪ و از ﻣﺪل رواﻧﺎب ذوب ﺑﺮف SRM ﺑﺮاي ﺑﺮآورد رواﻧﺎب روزاﻧﻪ اﺳﺘﻔﺎده ﺷﺪ. ﺑﺮ اﯾﻦ اﺳﺎس، دادهﻫﺎ و ﻣﺘﻐﯿﺮﻫﺎي ﻣﻮرد ﻧﯿﺎز در ﭼﻬﺎر ﺳﺎل ﻣﺘﻮاﻟﯽ از 2006 اﻟﯽ 2009 ﺟﻤﻊ آوري و رواﻧﺎب ذوب ﺑﺮف ﺑﺮاي ﭼﻬﺎر ﺳﺎل ﺑﺮآورد ﺷﺪ. ﺑﺮاي ﻣﺤﺎﺳﺒﻪ ﺳﻄﺢ ﺑﺮف از ﺗﺼﺎوﯾﺮ ﺳﻨﺠﻨﺪه MODIS اﺳﺘﻔﺎده ﺷﺪ. ﭘﺲ از ﺟﺪاﺳﺎزي ﭘﻮﺷﺶ ﺑﺮف اﯾﻦ ﺗﺼﺎوﯾﺮ، ﺑﺎ ﺑﻬﺮهﮔﯿﺮي از ﺳﺎﻣﺎﻧﻪ اﻃﻼﻋﺎت ﺟﻐﺮاﻓﯿﺎﯾﯽ ﻣﺴﺎﺣﺖ ﺑﺮف روزاﻧﻪ ﺑﺮآورد ﺷﺪ و ﻫﻤﺮاه ﺑﺎ ﻣﺘﻐﯿﺮﻫﺎي دﯾﮕﺮ وارد ﻣﺪل ﺷﺪ ﺗﺎ رواﻧﺎب روزاﻧﻪ ﺑﺮآورد ﺷﻮد. ﺑﺮاي ﮐﺴﺐ ﻧﺘﺎﯾﺞ دﻗﯿﻖﺗﺮ و ﻣﺪل ﮐﺮدن رﻓﺘﺎر رواﻧﺎب روزاﻧﻪ ﺣﻮﺿﻪ، واﺳﻨﺠﯽ و اﻋﺘﺒﺎرﺳﻨﺠﯽ ﻣﺪل اﻧﺠﺎم ﺷﺪ. ﺑﺎ اﻧﺠﺎم واﺳﻨﺠﯽ، ﻣﻨﺎﺳﺐﺗﺮﯾﻦ ﻣﻘﺪار ﺑﺮاي ﻫﺮ ﭘﺎراﻣﺘﺮ ﺑﻪدﺳﺖ آﻣﺪ. ﺑﺮاي ﺑﺮرﺳﯽ دﻗﺖ ﻣﺪل و ﻣﻘﺎﯾﺴﻪ ﻧﺘﺎﯾﺞ ﺑﺎ دادهﻫﺎي زﻣﯿﻨﯽ رواﻧﺎب از ﺿﺮﯾﺐ ﺗﺒﯿﯿﻦ و درﺻﺪ ﺗﻔﺎﺿﻞ ﺣﺠﻤﯽ ﻣﺪل اﺳﺘﻔﺎده ﺷﺪ. ﺑﺮ اﺳﺎس ﻧﺘﺎﯾﺞ، ﺿﺮﯾﺐ ﺗﺒﯿﯿﻦ ﺑﯿﻦ 0/90 ﺗﺎ 0/94 و ﺗﻔﺎﺿﻞ ﺣﺠﻤﯽ ﺑﯿﻦ 6/8 ﺗﺎ 7/2 درﺻﺪ ﺑﺪﺳﺖ آﻣﺪ ﮐﻪ ﮐﺎراﺋﯽ ﺑﺎﻻي ﻣﺪل را ﻧﺸﺎن ﻣﯽدﻫﺪ.

The importance of snow and its water equivalent in water resources supply has caused many studies and researches to measure snow characteristics and runoff. Conducted in the Saqqez Watershed, this research attempted to estimate snow–induced runoff in a mountainous area and the SRM Model was selected to simulate daily runoff from snow-melt. Based on the data and variables for four consecutive years of 2006 to 2009 collected and snowmelt runoff was estimated. MODIS satellite images were used to calculate the snow coverage area. After segregating the snow coverage from the images, the daily snow area was calculated using GIS, and along with the other variables, imported into the model. For better evaluation of efficiency of the model, the model was calibrated and validated. The process of calibration was led to the best estimate for each parameter. To evaluate the accuracy of model and comparing results with field data Nash-Sutcliffe coefficient and the percentage difference were used. The results of the Nash-Sutcliffe coefficient were between 0.90 to 0.94 and the differences in the volume were 6.8 to 7.2 percent, which indicates the high-performance of modeling.