Bhareh gharedaghy; amir ghasemzadeh
Abstract
Due to its high environmental diversity, Iran has a high rank in crises caused by natural disasters. Flood as one of the natural disasters, following the rapid growth of cities and climate changes in many regions, has caused severe social and economic, health and environmental damage. For this reason, ...
Read More
Due to its high environmental diversity, Iran has a high rank in crises caused by natural disasters. Flood as one of the natural disasters, following the rapid growth of cities and climate changes in many regions, has caused severe social and economic, health and environmental damage. For this reason, predict of flood susceptibility is so essential that failure to identify flood susceptibility may increase its destructive effects. Recently, with the advancement of remote sensing tools, geographic information, machine learning and statistical models, it is possible to create a more accurate flood susceptibility map. For this purpose, in this research, by using Sentinel satellite images and using the Ensemble approach with six machine learning models, flood susceptibility was predicted in the Karun watershed. Individual models include Generalized Linear Model (GLM), Boosted Regression Tree (BRT), Support Vector Machine (SVM), Random Forest (RF), Multivariate Adaptive Regression Splines (MARS), Maximum Entropy (MAXENT). The results of this study show that the northeast of Aligudarz city, parts of Durud and Azna in Lorestan province, Khademmirza, Shahrekord and Kiyar in Chaharmahal Bakhtiari province, Dana and Boyer Ahmad in Kohkiloye and Boyer Ahmad province, Semirom city in Isfahan province and the southern border areas of Karun River in Khuzestan province has the highest flood potential in this basin. The results of this research are effective for managers and planners and will prevent development in vulnerable areas and reduce financial and economic losses in the future.
Faraham Ahmadzadeh; Negar Amiri; Elham Ebrahimi
Volume 10, Issue 2 , September 2018, , Pages 95-108
Abstract
Today, it is well-known that predicting the distribution potential of endangered species by usingspatial modeling methods is highly beneficial and using these methods can greatly contribute toecological conservation and management. Rana pesudodalmatina is one of the Iranian endemicamphibian species of ...
Read More
Today, it is well-known that predicting the distribution potential of endangered species by usingspatial modeling methods is highly beneficial and using these methods can greatly contribute toecological conservation and management. Rana pesudodalmatina is one of the Iranian endemicamphibian species of Iran. In order to predict the potential geographic distribution of the species itsoccurrence points were collected through field work and 19 so-called bioclim climate variables asspatial environmental predictors were extracted from the Worldclim database. By applying Pearsoncorrelation test, the highly correlated variables with correlation coefficient of 0.75 were eliminated.Species distribution modeling was done using newly published R package which includes GLM,GAM, RF, MARS, CART, FDA, BRT and SVM models. All individual models were compound as anensemble to reduce the uncertainty which increase the accuracy and predictive power. The resultsrevealed that the long-legged wood frog has maximum distribution potential in Hyrcanian forest ofIran. Also, the results of the valuation of the models showed that the AUC and TSS had better statusand the SVM model was the most credible. In addition, the results of measuring the importance ofeach of the variables showed that BIO6 had the highest and BIO19 had the least importance for this.
