Remote Sensing and GIS Applications in Environmental Sciences
Evaluation and Prediction of Flood Prone Areas Using Random Forest Algorithm
Document Type : Original Article
Authors
1 Hakim Sabzevari University - Faculty of Geography and Environmental Sciences
2 Hakim Sabzevari University - Faculty of Geography and Environmental Sciences - Department of Climate and Geomorphology
3 Department of Remote sensing and Geographic Information System, Faculty of Geography and Environmental sciences, Hakim sabzevari University, Sabzevar, Iran
Abstract
Floods, as one of the three primary natural hazards in Iran, cause significant damage annually to the environment, infrastructure, and livelihoods. This study aimed to map flood hazard zones in the Davarzan region and identify the key factors contributing to flood occurrence using the Random Forest algorithm. Sixteen environmental indicators, including elevation, slope, precipitation, geology, geomorphology, land use, plan curvature, profile curvature, topographic wetness index (TWI), proximity to stream, stream density, train ruggedness index (TRI), Normalized difference vegetation index (NDVI), Stream power index (SPI), geology and soil were included in the analysis. The Information Gain Ratio (IGR) was used to determine the importance of each factor, and less significant variables were excluded from the modeling process. The results showed that elevation, precipitation, and land use were the most influential factors in flood occurrence. According to the flood hazard mapping, approximately 67% of the study area is at high risk, 5% at very high risk, 6% at moderate risk, and 22% at low to very low risk. Additionally, it was found that about 50% of urban and rural areas are in high-risk zones, emphasizing the need for preventive measures. This research demonstrated that the Random Forest algorithm, in addition to providing precise hazard mapping, serves as an effective tool for flood management. Strategies such as enhancing vegetation cover, revising land-use practices, improving drainage infrastructure, and reevaluating the location of residential areas can significantly reduce flood-related damages. The study's findings offer a valuable basis for managerial decision-making in Davarzan and similar regions.
Keywords
Main Subjects
)