@article { author = {zakerinejad, reza and Amoshahi, Nahid}, title = {Assessment of Landslide Hazard Using Remote sensing data and the Maximum Entropy Model (Case Study: Kome watershed, in south of Isfahan Province)}, journal = {Quantitative Geomorphological Research}, volume = {11}, number = {2}, pages = {128-149}, year = {2022}, publisher = {Iranian Association of Geomorphology}, issn = {22519424}, eissn = {}, doi = {10.22034/gmpj.2022.340900.1349}, abstract = {Extended AbstractIntroductionInstability of slope is one of the geomorphological and geological phenomena that plays an effective role in deforming the earth's surface. This process can affects human activities, because it can be dangerous phenomena. One of the important tasks of applied geomorphological knowledge is to study the location of hazardous places against various geomorphological hazards. Landslides is one of the most dangerous gemorpgholocal event, due to the recurrence of this phenomenon and its harmful damages, it is very important. The aim of this study was to identify landslide risk zoning in Kameh basin in the south of Semirom city, in Isfahan Province. Hu et al. (2020) used altitude, slope direction, curvature, lithology, distance from fault, slope, precipitation, land use, and NDVI to predict landslides in China's Jinsha Basin. The evaluation results show that the negative samples of "non-slip regions" produced according to the FT model are more reasonable and the combined method supported by the FT and ML models has the highest advance efficiency of about 94% of the total production accuracy. By Scenario-FT and then Scenario-SS (87%) and Scenario-RS (65%)Semirom is one of the landslide sensitive areas due to being located on the seismic zone of Zagros Mountain. Occurrence of many landslides is evidence of the sensitivity of these areas, which highlights the need for this studyMethodologyIn this study in first step we have collected the landslides locations with using aerial photos, filed survery, and google earth images (GE). The second step of this research was to preparing 13 the most important variable including: Elevation, slope, aspect, geology, vegetation density, land use, soil texture, distance from fault, distance from road, distance from stream, topographic moisture index, waterway strength index and average annual rainfall . In the last step we applied the Maxent model for zoing of the landslide map. The Maxent model is one of the most common machine learning algorithms. The principle of this method goes back to the maximum entropy or close to reality. Maximum entropy or maximum turbulence is the name of the second law of thermodynamics, also known as turbulence law. Results and Discussion In this study, the Area-Under-Curvae index (AUC) for the ROC curve was used to evaluate the validity of the final model. The area under the curve was calculated for traing and testingdata. If the area under the curve is more than 0.9, the detection power of the model is considered very high. This value was obtained for educational data equal to 0.931 and for testing data 0.887 and shows that the detection power of the model is considered very high and the model can well cover different areas of landslide risk. To determine the effect of each variable on the landslide risk in the region, the percentage of participation of each variable was used using Maxent output. Percentage of participation in the model indicates that the most influential parameters on the landslide risk model are distance from waterway (17.1%), geology (14.1%) and distance from road (13%), respectively. And the least effective parameters are aspect (0.5%), stram power index (SPI)(0.7%) and topographic moisture index (0.9%), respectively. Based on the results, of the total area of the study area, 139760 hectares (84.69%) very low risk, 15449.5 hectares (9.36%) low risk, 6903.39 hectares (4.18%) relatively low The risk is 2,456.85 hectares (1.48%) high risk and 445.91 hectares (0.27%) are in the high risk category.ConclusionIn this study, in result of poorly vegetation and agricultural, the amount of landslides is high. The results of this study showed that from 0 slope to slope of approximately 35%, the probability of landslide has increased exponentially and then with a gentler slope, the amount of landslide has increased.According to the classification of images, the area of very low-risk to very high-risk areas were 84.69%, 9.36%, 4.18%, 1.48% and 0.27%, respectively. To evaluate the validity of the final model, the sub-curve surface index (AUC) for the ROC curve was used. If the area under the curve is more than 0.9, the detection power of the model is considered very high, which in this study was equal to 0.931, indicating that the detection power of the model is considered very high and the model It can well differentiate between different areas of landslide risk In the slopes of zero to 35%, where man-made areas and agricultural use are the predominant uses in the region, it shows the great impact of man on the occurrence of landslides. Slope direction factor in the region shows that the most effective slope in the northern direction of the region due to rainfall and humidity, play an effective role in creating slippery movements in the region.}, keywords = {Landslide,MaxEnt model,komeh basin}, title_fa = {ارزیابی خطر زمین لغزش با استفاده از داده های سنجش از دور و مدل حداکثر آنتروپی (منطقه مورد مطالعه: حوضه آبخیز کمه، جنوب استان اصفهان)}, abstract_fa = {ناپایداری دامنه‌های طبیعی یکی از پدیده‌های ژئومورفولوژیکی و زمین شناسی است که در تغییر شکل سطح زمین نقش موثری دارد. در این پژوهش با استفاده از مدل مکسنت (حداکثر آنتروپی) به پهنه‌بندی زمین لغزش در حوضه آبخیز کمه در جنوب استان اصفهان پرداخته شد. در پژوهش مورد نظر از 13 متغیر (ارتفاع، شیب، جهت شیب، زمین شناسی، تراکم پوشش گیاهی، کاربری اراضی، بافت خاک، فاصله از گسل، فاصله از جاده، فاصله از آبراهه، شاخص رطوبت توپوگرافی، شاخص قدرت آبراهه و میانگین بارش سالانه) استفاده شده است. نتایج پژوهش نشان داد که مهمترین عوامل تاثیرگذار بر مدل احتمال خطر زمین لغزش، فاصله از آبراهه (1/17 درصد)، زمین شناسی (1/14 درصد) و فاصله از جاده (13 درصد) می‌باشند. همچنین نتایج حاصل از سطح زیر منحنی AUC برای داده‌های آموزشی برابر با 931/ 0 و برای داده-های تعلیمی 887/0 به‌دست آمد و نشان دهنده این است که قدرت تشخیص مدل بسیار بالا بوده است و مدل به خوبی می‌تواند مناطق مختلف خطر زمین لغزش را از یکدیگر تفکیک کند. بر اساس نتایج حاصل شده، از کل مساحت منطقه مورد مطالعه، 139760 هکتار (69/84 درصد) خیلی کم خطر، 5/15449 هکتار (36/9 درصد) کم خطر، 39/6903 هکتار (18/4درصد) نسبتا کم خطر، 85/2456هکتار (48/1 درصد) پر خطر و 91/445هکتار (27/0 درصد) در طبقه خیلی پر خطر قرار دارد.}, keywords_fa = {زمین لغزش,مدل مکسنت,حوضه آبخیز کمه}, url = {https://www.geomorphologyjournal.ir/article_151095.html}, eprint = {https://www.geomorphologyjournal.ir/article_151095_8911e6b4e703e689ac896b6e3ad8290b.pdf} }