Page 508 - NGTU_paper_withoutVideo

Modern Geomatics Technologies and Applications

Proposing a combined method of optic and SAR images to estimate flood damage:
Case study - Sahand Mountain and its surroundings

3*
2
1
Arash rahmani.z , Mohammadreza Emami goyjesoltan , Misagh Sepehriamin
1 Assistant Professor, University of Tabriz, arahmanizadeh@tabrizu.ac.ir, Tabriz, Iran
2 Student, University of Tabriz, emamiarticle@gmail.com, Tabriz, Iran
*3 Student, University of Tabriz, misagh.sepehriamin@gmail.com, Tabriz, Iran

Abstract: Floods are a natural hazard of the Earth's ecosystem, and they are frequently caused by heavy rains, melting
snow and large chunks of ice, or river uprising. This study examined the extent of damage to flooded areas in April 2112
by using sentinel-1 and landsat-8 pictures from the Sahand mountain and its surroundings. The flooded areas were
separated by a threshold after applying the filters necessary to eliminate errors in the SAR image. In addition, areas with
steep slopes as well as permanent rivers and pre-flooded waters were eliminated based on the slope and NDWI
parameteres (obtained from the SRTM data and the optical image, respectively). The final result demonstrates that
18492.88 hectares was covered by the floods.

keywords: flood, SAR, NDWI, change detection, thresholding

1 . Introduction

Flooding is one of the unpredictable natural disasters that occur as a result of flooding and flooding of a large portion of the
country. [1, 2 ] When a flood occurs, the river runoff rises, causing water from the river to overflow. One of the greatest threats
to human life and the ecosystem is that the bed itself overflows and flows on the ground. To efficiently organize an emergency
response, crisis management authorities need timely information on flood conditions in flooded areas, which must be given
immediately and from flood-prone areas by a source that provides critical data for disaster risk assessment and environmental
planning [3,9]. Flood change detection methods widely use synthetic aperture radar (SAR) images to detect flood zone [5].
Given that there will be no obstacles in imaging, both day and night and weather conditions, ensuring a continuous view of the
earth, it is one of the powerful tools for drawing flooded areas from space [6]. Areas of still water, on the other hand, can be
easily identified in radar data. Still water acts as a mirror in this type of data. As a result of this phenomenon, radar data has a lot
of dark pixels. The Digital Earth Model (DEM) can also be used to locate elevation data that is missing additional information
about surface phenomena [2]. The algorithm for identifying flood-damaged areas is based on a multi-step process that includes
image preprocessing, image thresholding, refinement and validation, and finally post-processing to ensure accuracy. Using the
available SENTINEL-1 images, the proposed method for determining the amount of flood damage in the study area during the
floods in April 2112 is proposed. On this basis, a preprocessing scheme is used to perform geometric correction and image
intensity estimation before using a threshold to separate the water areas. Finally, the data obtained from SAR data processing is
subjected to post-processing, which includes the removal of single pixels and areas that are mistakenly considered flooded areas,
yielding the final flood map. The NDWI parameters [8] for detecting water areas in the pre-flood area, SRTM data [4] for slope
mapping, and the majority filter [11] for improved display and detectable results were used in this post-processing. This method
can detect the amount of flood damage with low computational cost.

Then, in the second part, the case study area is discussed, in the third part, the working method is explained, in the fourth part,
the results and final data obtained from GOOGLE EARTH ENGINE and MATLAB software are expressed, and in the fifth part,
the results are analyzed.

1