Journal of Geomatics Science and Technology

fa تشخیص آنومالی در سری های زمانی کلروفیل در محدوده مکانی و زمانی زلزله های بزرگ ساحلی با روش جنگل تصادفی Anomaly Detection in Time Series of Chlorophyll Around the Time and Location of Large Coastal Earthquakes Using Random Forest Method فتوگرامتری و سنجش از دور Photo&RS پژوهشي Research <div style="text-align: justify;">زلزله یکی از مخرب&lrm;ترین بلایای طبیعی است که به طور جامع پیش بینی آن تاکنون محقق نگشته است. تحقیقات پیشین نشان داده&lrm;اند به کمک داده&lrm;های سنجش از دور می&lrm;توان به اطلاعاتی دسترسی پیدا کرد که نشان&lrm;دهنده رابطه&lrm;ای معنادار بین وقوع نابهنجاری در تغییرات برخی پارامترهای فیزیکی و شیمیایی با وقوع زلزله هستند. این پارامترهای فیزیکی و شیمیایی همان پیش نشانگرهای زلزله هستند. یکی از پیش نشانگرهایی که ارتباط آن با زلزله&lrm;های ساحلی در سال&lrm;های اخیر مورد مطالعه برخی از پژوهشگران واقع شده، میزان کلروفیل موجود در سطح آب است. پارامتر کلروفیل-&lrm;&lrm;آ از طریق روش&lrm;های مختلفی از جمله روش&lrm;های آزمایشگاهی طیف سنجی، اندازه&lrm;گیری فلورسانس کلروفیل و یا از طریق داده&lrm;های ماهواره&lrm;ای و با بکارگیری الگوریتم&lrm;های شاخص رنگ (CI) و نسبت باند&lrm;ریلی (OCX) و غیره بدست می&lrm;آید. با بررسی سری&lrm;های زمانی کلروفیل پنج زلزله بزرگ حاصل از محصولات سنجنده مودیس  (MODIS) بر روی سکوهای آکوا (Aqua) و ترا (Terra) و با استفاده از الگوریتم جنگل تصادفی مشاهده شد که  انتشار انرژی حرارتی و گازهای متصاعد زمین بر اثر فعالیت&lrm;های صفحات تکتونیکی و یا دیگر فعالیت&lrm;های فیزیکی و شیمیایی پوسته زمین قبل، هنگام و بعد از وقوع زلزله&lrm;های ساحلی و نزدیک به ساحل می&lrm;تواند به تغییر در میزان کلروفیل سطح آب منجر شود و این پارامتر می&lrm;تواند به عنوان یک پیش نشانگر زلزله در تحقیقات بعدی مورد بررسی قرار گیرد.</div> <div style="text-align: justify;">Earthquake is one of the most devastating natural hazards which efforts to predict the time, location and magnitude of it have not been yet completely successful. Remote Sensing data is proved to be an effective source of information about lithospheric and atmospheric activities around the impending earthquakes which are referred to as earthquake precursors. The issue of detecting anomalies in these precursors has been interesting to many researchers. One of the precursors that has been taken into consideration by the researchers, is the chlorophyll-a (chl-a) concentration on the sea surface. Since, over %70 of the Earth's surface is covered by water and many seismic active faults are located in coastal belts of the continents, the behavior of oceanic earthquake-related parameters such as Sea Surface Temperature (SST), surface latent heat flux, upwelling index and chl-a, is of particular importance. Elastic strain in rocks, formation of micro-cracks, gas release and other chemical or physical activities in the Earth's crust before and during earthquakes has been reported to cause changes in oceanic parameters. Chl-a parameter is obtained through various methods including laboratory methods of spectroscopy, chlorophyll fluorescence measurement or through satellite data using Color Index (CI) and Raily band ratio (OCX) algorithms etc. Changes from time to time in plankton population in ocean surface and chl-which is the indicator of the primary productivity of phytoplankton biomass in the ocean, can be continuously monitored from space by Ocean Color sensors. In this study, MODIS on Aqua and Terra products were used to examine the pattern of variations of chl-a. By examining the chlorophyll time series of five large earthquakes produced by MODIS sensor products on Aqua and Terra platforms and using a random forest algorithm, it was observed that the release of thermal energy and ground gases due to the activity of Tectonic plates or other physical and chemical activities of the earth's crust before, during and after coastal and near-coastal earthquakes can lead to changes in the amount of chlorophyll in the water surface and this parameter can be used and investigated as an earthquake precursor in future research. The results showed that chlorophyll-a levels exceeded the permissible limits 51, 48, 46 and 28 days before the Gujarat earthquake by 85, 45, 15 and 35%, respectively. In the 2004 Sumatra earthquake in the 20 days before and 18 days after the earthquake, the percentage of chlorophyll-a parameter crossing the upper limit was 110 and 190, respectively. In the 2006 Java earthquake, 42 days before, 15 and 16 days after the earthquake, the amount of chlorophyll-a suddenly changed to 136.84, 52.63 and 107.89% of the allowable threshold. In two other studies, this amount is equal to 199.87, 25, 150 and 190% more than allowable limit, respectively, on the 44th and 34th days before, on the day of the earthquake and 13 days after the Chile earthquake, and 321.42, 50 and 160.71% more than allowable limit 7 and 4 days before and 17 days after the earthquake in Mexico.  In addition, the clear superiority of the Random Forest (RF) algorithm in correct detection of anomalies showed that RF algorithm can be introduced as an effective tool in anomaly detection in time series.</div> زلزله, کلروفیل, جنگل تصادفی, مودیس, ناهنجاری Earthquakes, Chlorophyll-a, Random Forests, MODIS, Anomaly 165 174 http://jgst.issgeac.ir/browse.php?a_code=A-10-191-4&slc_lang=fa&sid=1 M. Nadri معصومه ندری Masoume.nadri@ut.ac.ir 10031947532846007062 10031947532846007062 No University of Tehran دانشگاه تهران M. Akhoondzadeh Hanzaei مهدی آخوندزاده هنزائی makhonz@ut.ac.ir 10031947532846007063 10031947532846007063 Yes University of Tehran دانشگاه تهران