Flood Impact Assessment on Agricultural and Municipal Areas Using Sentinel-1 and 2 Satellite Images(Case study: Kermanshah province)

doi:10.21203/rs.3.rs-2225762/v1

Download PDF

Research Article

Flood Impact Assessment on Agricultural and Municipal Areas Using Sentinel-1 and 2 Satellite Images(Case study: Kermanshah province)

https://doi.org/10.21203/rs.3.rs-2225762/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 16 Apr, 2024

Read the published version in Natural Hazards →

You are reading this latest preprint version

Flood is one of the most destructive natural events that can be investigated as soon as possible to reduce the extent of flood damage. Floods have caused adverse impacts on agricultural lands and settlements. RS is one of the most widely applicable and fast methods in this way. In this study, the S1A, SAR data with VH descending pass and S2 during 01/03/2019 to 20/03/2019 and 25/03/2019 to 20/04/2019 were used for the before and the post-flood period in Kermanshah province. MNDWI and NDWI were applied to detect the water zones of S2 and were compared and validate the S1 images. The quality of radar images is better than optical in flood conditions where the weather is cloudy and rainy. MODIS, Hydrosheds and SRTM satellite images were used as different filters to detect land use, permanent water and areas with slope of more than 5%. The results showed that totally 36,849 ha of lands are affected by the flood. It’s containing 7073 and 4224 ha of agricultural and urban lands that are exposed to destruction in this period. The NDWI and MNDWI indices were calculated flooded area as 30,179 and 32,540 ha, respectively which are lower values compared to the results of the S1.

Agricultural land

Flood

NDWI

MNDWI

Floods are very serious natural hazards that occur frequently, causing huge damage to infrastructure, life, and the economy. Management officials need timely information about flood conditions in sunken areas to effectively organize emergency responses. An estimated 21 million people worldwide are affected by river floods each year, which may increase to 54 million by 2030.Climate change and socio-economic growth are the main causes of these events )Gorelick et al. 2017). Various climatic factors affect floods mainly through rainfall, including its intensity, amount and duration. Rapid estimation of flood space range in large areas provides a key data source for disaster assessment. Satellite RS is increasingly important for flood mapping. Optical RS is essential for monitoring flood dynamics based on low water reflection in infrared and high reflection in blue/green strips Dumitru et al. 2015. RS technologies make it easier to map large flooded areas, provide early alerts quickly, and help provide support in critical situations and minimize damage. Space sensors are more effective at providing real-time data for flood extent monitoring than ground-based techniques. The satellite images are very valuable for emergency response because they provide constant observations about it (Tarpanelli et.al.2022). Choosing the right sensors to effectively produce flood outbreaks in large geographical areas is a major challenge. Satellite platforms and other advances in RS have provided a wide range of satellite data applications (Cohen et al. 2019). GEE is widely used for better and faster data processing and analysis. GEE is a cloud platform for geospatial analysis. It is an integrated platform designed to empower a wide audience, including RS scientists who lack the technical knowledge of supercomputers (Gorelick et al. 2017). SAR is of particular interest in observing flood situations, actively emitting electromagnetic waves that are not affected by weather and time of day and can be used to detect floods in vegetation or urban areas. Therefore, the use of SAR radar data has been investigated by many researchers Landuyt et al. 2019.

It is necessary to use RS techniques for zoning the recent floods of Iran in different regions and to inform the authorities about the amount of agricultural, social and economic damages. Ganji et al. 2021estimated the flood area and time series changes occurred in April 2019. Calculating the MNDWI and NDWI indices by pre-processing the images of S2 and Landsat8 depicted that the largest area of flood zones was related to the MNDWI index, which is equal to 262.453 and 51.991 km² for the city and town of AqQala, respectively. Bigham Sereshkeh et al. 2020 used two algorithms of SVR and RF for accuracy of classification of S2 imagery by pixel-based and object-based methods in zoning flood areas in Taleghan. To validate the extracted zones, hydraulic zoning method with return periods of 2, 5, 10, 25 and 100 years was used. The results showed that both RF and SVM methods in object-based approach often had the highest overlap with different return periods. The highest overlap with flooded areas was obtained in the two-year return period with 68% and 66% for RF and SVM. Mehrabi (2021) used S1 time series imagery, threshold and SBAS techniques to determine ground displacement and flood effects in Pol-e-Dokhtar, Iran. The results showed that the nature of the spatial extent of the flood was sinusoidal, so that the behavior of the flood corresponded with the rainfall periods. In addition, SBAS-In SAR findings showed how a severe flood can cause ground displacement and SAR data are effectively used for flood monitoring, water mapping, and flood-induced land displacement. Despite advances in RS, there is a clear gap between the natural disaster community and technology resources leverage in real time which hinders timely ground response efforts.

Other studies in the world using RS for flood detection are containing: Alexandre et al. 2020 introduced an automated processing chain for S1 (SAR) radar data. This processing chain is based on the S1-Tiling algorithm and the NDR, which can load and clip S1 images on S2 pixels multiple times. Tripathy and Malladi (2022) has introduced a new web application called GFM that generates flood maps quickly without getting into technical complexities. To extract flood extent from S1 satellite data, before and during flood is considered. Mountainous, urban and arid areas are highly vulnerable to floods due to severe storms and sudden rains, which cause flash floods to affect infrastructure and the ecological and biophysical environment. Elhag and Abdurahman (2020) used the S1TBX to support the display and analysis of the large archive of ESA SAR mission products. The results showed that the area of the flooded surface is approximately 9 km², which mainly covers agricultural lands and urban areas of Tabuk city in Saudi Arabia. Hutan et al. 2019 were compared flooded areas identified using HEC-RAS and NDWI index obtained from Landsat 7-ETM + satellite imagery processing. The study area, the upper part of the Jijia River on the Moldovan Plateau, was affected by the July 2010 floods. During that event, the level of the River reached 579 CM at the Dingeni hydrometric station. By performing different flood simulations by applying the NDWI index and HEC RAS model, the flooded area is 15.8 and 16.26 km² respectively. Reliable and rapid flood maps are vital parameters in the preparation of disaster management plans. Vanama et al. 2021 revealed an effective flood in the framework of mapping using multi-temporal images, including C-band S1A & 1B, SAR images and optical WorldView-3 images, to analyze the 2018 Kerala flood event, India. To identify floods, CD techniques, RI and NCI combined with semi-automatic thresholding were implemented on temporary descending pass SAR images. Mehmood et al. 2021 used Landsat 5, 7 and 8 to map flood outbreak areas. GEE was used to implement the FMA and Landsat image processing. The FMA relies on the development of a "data cube" that overlaps the pixels of Landsat images over a period of time. Temporary and permanent water zones were evaluated using MNDWI index, which estimated FMA accuracy in the range of 71–90% and overall accuracy in the range of 74–89%. In general, FMA is an accurate and efficient method for zoning flood points. Zhang et al. 2020 presents a new method for rapidly determining the extent of flooding in large, semi-arid regions with challenging environmental conditions, according to S1 data. First, a preprocessing scheme is applied to perform geometric correction and estimate the image intensity. Second, an automated threshold method is used to establish the initial classification of land and water by merging probability density distributions. Tarpanelli et al. 2022 evaluated a synthetic study of the S1 and 2 in the systematic assessment of floods in Europe. They collected ten years of river discharge data over almost 2000 sites in Europe and analyzed flood events over some established thresholds. Results show that assuming the configuration of a constellation of two satellites for each mission and considering the ascending and descending orbit, on average the 58% of flood events are potentially observable by S1 and only the 28% by S2 due to the cloud coverage. Konapala et al.2021 examined the various S2 bands by calculating water indices and images derived from S1 to assess their ability to create accurate flood maps. The performance of combinations of S-1 and S-2 bands was evaluated using 446 flood images. Pandey et al.2022 was monitored the large-scale flooding in 2020 in the Ganga-Brahmaputra basin using SAR data in cloud platform GEE, and estimated the impact of floods on the agriculture and population in the basin. The results showed that agricultural lands and settlements, affected by flood were 23.68–28.47% and 5.66–9.15%, respectively.

In short, the RS flood zoning method should be able to record floods in real time, meaning that instead of training and analyzing the parameters of the time-consuming model, it can work at high speeds and low computational costs, especially when using multiple images to observe flood evolution. In addition, flood detection method should be widely used for large, semi-arid areas with complex flood conditions and a variety of different land cover. Finally, flood process detection operations must be independent of reference image and specific satellite sensors. Accordingly, this paper uses an approach to automatically detect floods in semi-arid large scale region using S1 and 2 multi-temporal images of land range detection. Contrary to the methods presented above, the proposed approach has less calculations and higher detection speed. In this method, the flooded areas in different landuse areas are well separated according to different filters including slop, detection of permanent water areas, MODIS farm land maps and JRC global surface water and JRC golbal human settelement population density dataset, although CD method with thresholding 1.25 were applied and the results were validated and compared with two indices of NDWI and MNDWI in the Kermanshah province.

In this study, the effect of flood on agricultural, urban and endangered populations of kermanshah province was investigated. The province has a latitude of 33° 37ʹ 08ʹʹ to 35°17ʹ08ʹʹ North and longitude 45°20'39ʹʹto 48°01'58ʹʹEast. According to the 2017 census, the population of Kermanshah province was equal to 1,952,000 people. In terms of topography, is a mountainous region located between the Iranian plateau and the Mesopotamian plains and the high mountainous plains formed among it. Its average elevation from sea level is 1200 meters. In a general classification, Kermanshah province can be divided into two distinct tropical and temperate mountainous parts in terms of climatic conditions. The Iraqi border strip is one of the tropical parts of the province and the central and eastern parts of the province, which includes the high plains and mountainous parts, are among the temperate parts of the province. In March 2019 to April 2020, good rainfall occurred throughout the country. In this period, the amount of daily precipitation in Kermanshah synoptic station was investigated and the images were selected for the periods before and after rainfall.

The required data sets of S1GRD as well as S2 optical images from the GEE cloud computing platform (https://code.earthengine.google.com/) were considered to map the extent of flooding, respectively. In 2019, there was widespread flooding in parts of Iran, which caused water to rise in parts of the river in Kermanshah province and spread to the lands around the river. The images used in this S1 VH study were considered the scenes obtained for descending passage with a 12-day review cycle from March to April 2019. So that the images dated 01/03/2019 to 20/03/2019 and 25/03/2019 to 20/04/2019 are selected for the period before and after the flood. The conceptual framework of the flood mapping algorithm is shown in Fig. 2.

2.1 Sentinel-1 SAR imagery

The S1A satellite provided data in the C-band dual-polarization channels (VH and VV) with a repeat cycle of 12 days and was used to extract the flood inundated area. It’s the first satellite in the sentinel series and was launched by the ESA on April 3, 2014. The pixel spacing resolution of the S1A data is 10 m. This satellite is a song sun and includes a couple of polar orbit sensors (A and B) and is able to collect information from the earth's surface day and night. S1 GRD data taken from GEE in IW mode and with VH polarization. In the preprocessing stage, Range-Doppler, which includes ground correction for S1 images and radiometric calibration to zero sigma (dB) in GRD, was performed and modified to geographic coordinates (WGS-84) using global SRTM 1 images. The images were then calibrated radiometrically to reduce the back-scattering intensity values using the calibration parameters of the LEE filter sensor. The result of this preprocessing in GEE environment is a georeference, calibrated radiometric image. VH polarization mostly is the preferred polarization for flood mapping because of the overestimation of results obtained through ‘VV’ polarization. However, it always depends on the study area and it is more sensitive to the changes on the land surface (Goffi et al. 2020).

2.2 Change Detection approach

In this research, a direct CD approach was used, in which the mosaic of images after flood is divided into a mosaic of images before the flood, resulting in a checkered layer showing the degree of change per pixel. Large values, bright pixels, indicate a large change, and low values, dark pixels, refer to low change (Li et al., 2018). Then, the binary raster layer is created with the threshold value of 1.25, assigning 1 to all the values greater than 1.25 (flood pixels) and 0 to all values less than1.25 (non-flood pixels).The binary checkered layer created by this process shows the extent of potential flooding. The threshold of 1.25 is chosen through trial and error and can be adjusted if false positive or negative values are high.

2.3 Removal of permanent water zones from flood zone

Several datasets are used to remove false positives in the flood area layer, which includes JRC's global surface water dataset to remove all water-covered areas that have water for more than 8 months a year, which has a resolution of 30 m and was last updated in 2018. Digital elevation model (Hydrosheds) dataset for the removal of areas with a slope of more than 5%, which is based on SRTM data and has a spatial resolution of 3 arc seconds. In addition, the connection of flood pixels to remove pixels connected to eight neighbors or less was also considered, which reduces the noise operation of the product to reduce the extent of the flood (vanama et al. 2021).

2.4 Calculation of flood area

To calculate the area of flood, a new grid layer was created that calculates the area in square meters per pixel. By summing up all the pixels, the area information is extracted and converted to hectares. The result was displayed in the lower left corner of the map viewer in GEE.

2.5 Population at risk of flooding

To estimate the number of people exposed to flood, the JRC, which has a resolution of 250 meters and updated in 2015 is considered. It contains information about the number of people living in each cell. For the confluence of the flood layer with the population layer, the flooded area layer and the population data set were matched and displayed as a new layer in the map (De et al .2022).

2.6 Detection of farmland affected by floods

MODIS Land Cover Type was used to estimate the amount of damaged farmland. This dataset has a spatial resolution of 500 meters and is updated annually. It is the only global dataset on Land Cover that is currently available in GEE. The type 1 land cover band consists of 17 floors with two categories of farmland (class 12: at least 60% of the area is cultivated, and class 14: natural vegetation farmland/mosaic: small-scale cultivation of 40–60% with natural tree, shrub, or herbaceous vegetation). Both classes were extracted from the dataset and matched with the flood area layer, which was sampled to the MODIS layer scale (Vanama et al. 2021).The area of affected farmland was calculated by flood extent method and displayed in the "Results" panel.

2.7 Affected urban areas

The affected urban areas were calculated using MODIS land cover type dataset. "Urban Class 13" was extracted from the "Type 1 Land Cover" band to assess potentially affected urban areas. In this process, the affected urban areas are most likely to be underestimated due to the difficulty of detecting water in residential areas.

2.8 Sentinel-2 imagery

The S2A was launched by the ESA in 2015 and included 13 bands across the visible spectrum range, near infrared and short infrared wave of the electromagnetic spectrum. The resolution of these images in three categories is 10, 20 and 60 m.

2.9 Spectral Indicators

S2 images can observe floods only during the day and in good weather conditions, because sunlight cannot pass through clouds in the visible range (Drusch et al. 2012). Spectral indices improve the classification process by using band ratio. For optical data, water areas are usually determined using the NIR band because in this band, the reflectance of water surface is much lower than the reflectance given by other land cover types. However, optical images easily affect by atmospheric conditions, surface reflection of sun light, and water turbidity that cause difficult to estimate a regular threshold value for flood detection. To attenuate this problem, band calculations of NDVI and/or NDWI, MNDWI are calculated with a combination of visible, NIR and SWIR bands (Sakamoto et al. 2007; Seiler et al. 2009; Townsend et al. 1998). There are various methods such as PCA (Gianinetto et al. 2006), SVR and C-mean clustering (Longbotham et al. 2012), IHS (Yonghua et al. 2007; Goffi et al. 2020) to detect flooded areas from the optical imagery. Regularly, the flooded area mapping is extracted through thresholding such indexes.

2.9.1 NDWI and MNDWI Indicators

The NDWI index is the first water identification index presented in 1996 for flood zoning and open water detection using RS data. This index is obtained by combining NIR with wavelength of 842 nm and green band (GREEN) with wavelength of 560 nm, which is used in S2 for band images (Green) and (NIR) of bands 3 and 8 and Equation.1 (Goffi et al. 2020).

$$NDWI=\frac{Green-NIR}{Green+NIR}$$

In the NDWI index, by reducing the reflection of non-water characteristics such as plant and soil at the wavelength of NIR band and at the same time with increasing water reflection at green band wavelength, it detects flood zones from non-floodwater. The value of this index is between − 1 and + 1, which from − 1 to 0 indicates non-flood zones and 0 to 1 indicates flood zones (McFeeters, 1996).

MNDWI was presented by Xu (2006) to develop water insight even in existence of high turbidity and to decrease misunderstanding with built up areas. If the SWIR-1 band is replaced instead of the NIR band, the MNDWI index is calculated. The main advantage is to reduce the errors caused by the shadows of buildings, which makes the flood zones better identified. The value of this index is between − 1 and + 1, all values less than 0.2 are non-water and greater than 0.2 are considered as water (Ganji et al. 2021).

$$MNDWI=\frac{Green-SWIR1}{Green+SWIR1}$$

In this study, NDWI and MNDWI indices were calculated for images before and after the flood and normalize differentiation was performed to validate changes in flooding extent.

Since flood occurs as a result of heavy rainfall, the daily precipitation of Kermanshah synoptic stations in the 2019/03/01 to 2019/04/30 period are investigated in Fig. 3. The maximum amount of precipitation is in 01/04/2019, with about 70 mm in kermanshah station.

3.1 The results of the change detection approach

Considering that the change detection method has been applied in the GEE, differentiating between images before and after the flood is calculated by a threshold of 1.25. This value can be changed for different regions and is considered the best output. Figure 4 on the left shows the difference layer and the blue zones on the right map is associated with the flooded area. The land area affected by flood was estimated to be 36,849 ha. If the date of investigation of flood zone or polarization will change from VH to VV, the calculated numbers will be changed for the area or people affected by the flood.

3.2 Results of applying population density layer

The estimation of the population is necessary for the analysis of the severity of the flood hazard. There is a higher concentration of the population near the river banks. According to the population layer of JRC dataset used to identify residential areas of Kermanshah province, in the left form of Fig. 5, white areas show population density and in the right form using yellow to red color spectrum, denser areas with red color and less density with yellow are shown. The area of residential lands affected by the flood was estimated to be 4224 ha and the number of people at risk of flooding was calculated 129882 people.

3.3 Results of land use layer application

The land use layer was considered according to MODIS satellite data for the region. As is evident in Fig. 6, the light green color indicates the meadows and the gray color of urban areas and the red color of the fields. By matching this layer and the layer of flooded areas in the image on the right, the crops exposed to flooding are marked in dark green and the area of these lands is estimated to be 7073 ha.

3.4 Overall Flood Zone Results on the Agricultural, Residential and Population

The results of S1 radar images for flood land detection and its effects on agricultural, residential lands and its affected population are shown in Fig. 7. It shows flood zones with blue, green and gray color for flooded area, affected cropland and urban areas respectively. Population density that is affected by flood are shown with yellow to red for low to high density in Kermanshah province in the GEE environment.

The rising water of the Qarasu River due to flooding in the vicinity of Ghazanchi caused its spread in the agricultural lands of Chaghamaran, Nezamabad-Abad, Sarablah and the central area of Kermanshah city. The values of affected land and population for each sector are also shown separately on the left side of the map.

3.5 Results of NDWI and MNDWI Spectral Indices

The changes of NDWI in Fig. 8a and MNDWI in Fig. 8b and 8c have been shown. By applying the NDWI index, the values of pixels are in the range of -1 to 1. Negative values indicated arid areas and positive values indicated flood zones. This index takes into account exactly the flood zones that do not have mud and are shown in blue color in the images. In addition to identifying the water floods, the MNDWI index also detected all flood zones associated with mud. This index also considers the zones where the flood has subsided and created a high-humidity zone and is blackish brown in satellite imagery. For this reason, the obtained flood zones calculated by MNDWI are more than NDWI index. The total land area affected by flood with NDWI and MNDWI were estimated 30,179 and 32,540 ha respectively.

Every year, floods cause a lot of damage to people's lives and property. One of the biggest management problems in flooding is the lack of identification of areas affected by floods. Satellite-assisted RS technology is the cheapest and fastest way to detect flood zones. In recent years, researchers have investigated the use of satellites to determine the extent of floods due to their cost-effectiveness and suitable replacement for calibration of hydrodynamic flood modeling. Optical and radar satellite data are widely used for flood mapping. However, the spatial and temporal resolution of the atmospheric effects on satellite sensors as well as the characteristics of ground cover can prevent flood observation for calibration.

In the present study, flood vulnerability analysis was performed by identifying some endangered elements, which are defined by their exposure to floods. Endangered elements such as population, agricultural land and urban areas are identified with the help of population data set and LULC map. Future aspects of this research can in addition to the specifications considered in this research, including slope, land use, permanent water and buildings, other features of the study area such as height, aspect and infrastructure that help disaster management.

Therefore, S1 and optical images of S2 in the range of 01/03/2019 to 20/03/2019 and, 25/03/2019 to 20/04/2019 for before and after the flood are considered. For radar images, different filters including land slope, agricultural land detection, permanent waters, and populated areas, endangered urban areas were considered. NDWI and MNDWI indices were used for optical images. Flood zones, affected agricultural lands, population, urban areas as well as the number and the area were calculated and shown on the map along with the map guide (Fig. 7). The results of S1 showed that the total lands affected by floods were 36,849 ha and those at risk of flooding were 129,882 people. The population of Kermanshah city is at higher risk as compared to other zones and is a low-lying area with the highest density, which makes its population more prone to flood risk. Applying the zonal statistics showed the

agricultural and urban lands affected by floods were 19.2%, 11.5% where 25%, 7% comparing with Pandey et al. 2022.

In practice, however, most flood events occur in bad weather, characterized by heavy clouds and heavy rainfall, which prevents the use of optical imaging tools due to the lack of cloudless, high-quality images. In plant-covered areas, optical sensors are often unable to detect water under vegetation. Using radar data along with optical images can help to better understand flood areas (Amoah Addae, 2018). So the combination of the S1 and S2 satellite images was applied in the current study.

The study provides comprehension of the limitation of optical and SAR images for urban, agricultural and population that are in the risk of flood. MNDWI diminished built-up areas that often confused with water bodies. Water features greater positive values compared with NDWI so MNDWI has estimated more flooded areas compared to NDWI (Arshad et al. 2021). The total land area affected by flood with NDWI and MNDWI indices was calculated as 30,179 and 32,540 ha, respectively, which are lower values compared to the results of S1 that is consistent with (Konapala et al. 2021 and Tarpanelli et al. 2022).

Processing of SAR data obtained in GEE platform and flood extent map is very useful for large-scale flood zoning. Flood impact analysis approaches and information on farmland and affected populations can help disaster managers and policymakers take flood mitigation and flood risk assessment in sensitive and densely populated areas.

Data availability: The S1 and S2 data from the real case were
taken from the GEE platform (https://code.earthengine.google.com/) and are available in https://github.com/maryamhafezparast/Google-Earth-Engin.

Acknowledgments: Thanks to ESA's Copernicus program for making high-resolution (10m) satellite images freely available to the public.

Author Contributions: This manuscript is the result of the research of S.G under the supervision of M.H. and the Advising of R.Gh. All authors designed the study, developed the methodology, discussed the results and wrote the paper.

Funding Statement: The authors received no financial support for the research, authorship, and publication of this article.

Conflicts of Interest: The authors declare no conflict of interest.

Alexandre C, Johary R, Catry T, Mouquet P, Révillion C, Rakotondraompiana S, Pennober G. A (2020) Sentinel-1 based processing chain for detection of cyclonic flood impacts. Remote Sensing. 12(2):252.
Amoah Addae R (2018)nSatellite-based flood mapping for hydronamic flood moddel assessment: Accra, Ghana. MA thesis. University of Twente.
Arshad M, Ma X, Yin J, Ullah W, Liu M, Ullah I (2021) Performance evaluation of ERA-5, JRA-55, MERRA-2, and CFS-2 reanalysis datasets, over diverse climate regions of Pakistan. Weather Climate Extremes, 33: 100373.
Bigham Sereshkeh M, Kheyrkhah Zarkesh M, GhermezCheshmeh B (2020) Analysis Accruing of Sentinel-2 Image’s Classification Methods Based on Object Base and Pixel Base in Flood Area Zoning of Taleqan River. jwmseir. 14 (49):1-10 (In Persian).
Cohen S, Raney A, Munasinghe D, Loftis J.D, Molthan A, Bell J, Rogers L, Galantowicz J, Brakenridge G.R, Kettner A.J (2019) The Floodwater Depth Estimation Tool (FwDET v2.0) for improved remote sensing analysis of coastal flooding. Natural Hazards Earth Syst. Sci, 19: 2053–2065
De A, Upadhyaya D.B, Thiyaku S, Tomer S.K (2022) Use of Multi-sensor Satellite Remote Sensing Data for Flood and Drought Monitoring and Mapping in India. In: Kolathayar, S. Pal, I. Chian ,S.C. Mondal, A. (eds) Civil Engineering for Disaster Risk Reduction. Springer Tracts in Civil Engineering. Springer, Singapore.
Drusch M, Del Bello U, Carlier S, Colin O, Fernandez V, Gascon F, Hoersch B, Isola C, Laberinti P, Martimort P,Meygret A, Spoto F, Sy O, Marchese F and Bargellini P (2012) Sentinel-2: ESA's optical high-resolution mission for GMES operational services. Remote Sensing of Environment, 120: 25-36, doi:10.1016/j.rse.2011.11.026.
Dumitru C.O, Cui S, Faur D, Datcu M (2015) Data analytics for rapid mapping: Case study of a flooding event in Germany and the tsunami in Japan using very high resolution SAR images. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing. 8: 114–129.
Elhag M, Abdurahman S.G (2020) Advanced remote sensing techniques in flash flood delineation in Tabuk City, Saudi Arabia. Nat Hazards 103:3401–3413. https://doi.org/10.1007/s11069-020-04135-0
Ganji K, Gharechelou S and Ahmadi A (2021) Flooding Zone Extracting Using Spectrum Proportionality of Sentinel-2 and Landsat-8 Satellite Data (Case Study: Aq’Qala County Flooding), 8(31):42-53 (In Persian).
Gianinetto M, Villa P and Lechi G (2006) Postflood damage evaluation using Landsat TM and ETM+ data integrated with DEM," in IEEE Transactions on Geoscience and Remote Sensing.44(1): 236-243, doi: 10.1109/TGRS.2005.859952.
Goffi A, Stroppiana D, Brivio P. A, Bordogna G, and Boschetti M (2020) Towards an automated approach to map flooded areas from Sentinel-2 MSI data and soft integration of water spectral features. International Journal of Applied Earth Observation and Geoinformation 84: 101951,doi:10.1016/j.jag.2019.101951.
Gorelick N, Hancher M, Dixon M, Ilyushchenko S, Thau D, Moore R (2017) GEE: Planetary-scale geospatial analysis for everyone. Remote Sens. Environ. 202: 18–27.
Huțanu E, Urzică A, Paveluc L.E, Stoleriu C.C, Grozavu A (2019) Comparative analysis of flood areas using satellite images landsat 7-ETM+ and hydroluic model HEC-RAS. Case study: The Jihia River, Slobozia-dingeni section, International Scientific Conference Geobal canic: 619-625, Romania.
Konapala G, Kumar S.V, Khalique Ahmad SH (2021) Exploring Sentinel-1 and Sentinel-2 diversity for flood inundation mapping using deep learning, ISPRS Journal of Photogrammetry and Remote Sensing, 180: 163-173.
Landuyt L, Van Wesemael A, Schumann G.J.P, Hostache R, Verhoest N.E.C, Van Coillie F.M.B (2019) Flood mapping based on SAR: An assessment of established approaches. IEEE Transactions on Geoscience and Remote Sensing. 57: 722–739.
Li Y, Martinis S, Plank S, Ludwig R (2018) An automatic change detection approach for rapid flood mapping in Sentinel-1 SAR data, International Journal of Applied Earth Observation and Geoinformation, 73: 123-135.
Longbotham N, Pacifici F, Glenn T, Zare A, Volpi M, Tuia D, Christophe E, Michel J, Inglada J, Chanussot J, and Du Q (2011) Multi-modal change detection, application to the detection of flooded areas: Outcome of the 2009-2010 data fusion contest, IEEE J. Sel. Top. Appl. 5(1): 331 -342, doi:10.1109/JSTARS.2179638, 2012.
McFeeters SK (1996) The use of the Normalized Difference Water Index (NDWI) in the delineation of open water features. International Journal of Remote Sensing 17(7):1425–1432.
Mehmood H, Conway C, Perera D (2021) Mapping of Flood Areas Using Landsat with GEE Cloud Platform. Atmosphere. 12(7):866.
Mehrabi A(2021) Monitoring the Iran Pol-e-Dokhtar flood extent and detecting its induced ground displacement using sentinel-1 imagery techniques. Nat Hazards 105: 2603–2617. https://doi.org/10.1007/s11069-020-04414-w
Pandey A.C, Kaushik K, Parida B.R (2022) GEE for Large-Scale Flood Mapping Using SAR Data and Impact Assessment on Agriculture and Population of Ganga-Brahmaputra Basin. Sustainability. 14: 4210. https://doi.org/10.3390/su14074210.
Sakamoto T, Van Nguyen N, Kotera A, Ohno H, Ishitsuka N and Yokozawa M (2007) Detecting temporal changes in the extent of annual flooding within the Cambodia and the Vietnamese Mekong Delta from MODIS time-series imagery, Remote Sens. Environ., 109:295-313, doi:10.1016/j.rse.2007.01.011.
Seiler R, Schmidt J, Diallo O, and Csaplovics E (2007) Flood monitoring in a semi-arid environment using spatially high resolution radar and optical data, J. Environ. Manage, 90: 2121-2129, doi:10.1016/j.jenvman. 07.035, 2009.
Tarpanelli A, Mondini A. C and Camici S (2022) Effectiveness of Sentinel-1 and Sentinel-2 for Flood Detection Assessment in Europe, Natural Hazards Earth System Science. 22(8): 2473-2489. Discuss. https://doi.org/10.5194/nhess-2022-63.
Townsend P. A and Walsh S. J (1998) Modelling flood plain inundation using integrated GIS with radar and optical remote sensing, Geomorphology, 21(98):295–312, Doi: 10.1016/S0169-555X (97)00069-X.
Tripathy P, Malladi T (2022) Global Flood Mapper: a novel Google Earth Engine application for rapid flood mapping using Sentinel-1 SAR. Nat Hazards 114: 1341–1363 https://doi.org/10.1007/s11069-022-05428-2
Vanama V. S. K, Rao Y. S, Bhatt C. M (2021) Change detection based flood mapping using multi-temporal Earth Observation satellite images: 2018 flood event of Kerala. India, European Journal of Remote Sensing. 54(1): 42-58.
Xu H (2006) Modification of normalised difference water index (NDWI) to enhance open water features in remotely sensed imagery. Int. J. Remote Sens. 27: 3025-3033.
Yonghua S, Xiaojuan L, Huili G,Wenji Z and Zhaoning G (2007) A study on optical and SAR data fusion for extracting flooded area, Geoscience and Remote Sensing Symposium, IGARSS 2007, 3086–3089.
Zhang M, Chen F, Liang D, Tian B, Yang A (2020) Use of Sentinel-1 GRD SAR Images to Delineate Flood Extent in Pakistan. Sustainability. 12(14):5784.

Download PDF

Journal Publication

published 16 Apr, 2024

Read the published version in Natural Hazards →

Editorial decision: Major revisions
18 Dec, 2023
Reviewers agreed at journal
13 Dec, 2022
Reviewers invited by journal
11 Dec, 2022
Editor assigned by journal
05 Nov, 2022
First submitted to journal
05 Nov, 2022

You are reading this latest preprint version

Flood Impact Assessment on Agricultural and Municipal Areas Using Sentinel-1 and 2 Satellite Images(Case study: Kermanshah province)

Status:

Journal Publication

Version 1

Abstract

Figures

1. Introduction

2. Materials And Methods

2.1 Sentinel-1 SAR imagery

2.2 Change Detection approach

2.3 Removal of permanent water zones from flood zone

2.4 Calculation of flood area

2.5 Population at risk of flooding

2.6 Detection of farmland affected by floods

2.7 Affected urban areas

2.8 Sentinel-2 imagery

2.9 Spectral Indicators

2.9.1 NDWI and MNDWI Indicators

3. Results and Discussion

3.1 The results of the change detection approach

3.2 Results of applying population density layer

3.3 Results of land use layer application

3.4 Overall Flood Zone Results on the Agricultural, Residential and Population

3.5 Results of NDWI and MNDWI Spectral Indices

4. Conclusion and Discussion

Declarations

References

Status:

Journal Publication

Version 1