flood risk mapping in low coastal regions …

9
E-proceedings of the 38th IAHR World Congress September 1-6, 2019, Panama City, Panama doi:10.3850/38WC092019-1522 4005 FLOOD RISK MAPPING IN LOW COASTAL REGIONS CONSIDERING WATERSHED DELINEATION UNCERTAINTY EKKEHARD HOLZBECHER (1) & AHMED HADIDI (2) (1,2) German University of Technology in Oman (GUtech), Muscat, Oman e-mail [email protected] (2) e-mail [email protected] ABSTRACT Low-lying flat regions with a mountainous hinterland are prone to flash floods. Reports of flooding events are reported more frequently, coming from all parts of the world. Measures to mitigate hazardous effects from flash floods are discussed and put in operation at many places, among others the identification of flood prone zones in flood maps. These maps are drawn on basis of the knowledge of watersheds and the main watercourses therein. However, in lowlands on alluvial fans or on sedimentary terrain the delineation of catchments and streams is uncertain for several reasons. Streams may change their course as effect of a flood event. Due to small topographic gradients measurement inaccuracies amount to a higher relative error in flow calculations. We take the Batinah coast of Northern Oman as an example to demonstrate the deviations in delineations of downstream watersheds, resulting from different digital elevation maps. Finally we come up with recommendations concerning the construction of flood maps in low-lying coastal regions and the use of watershed delineation tools available in GIS software. Keywords: Flood maps, GIS, DEM, Watershed, Lowlands 1 INTRODUCTION The population in low elevation coastal zones is predicted to increase from 650 Mio. in 2000 to over 900 Mio. people in 2030 (Neumann et al., 2015), and may exceed 1.300 Mio in 2060. While in 2000 189 Mio. were living in 100-year flood plains this number will rise to approximately 275 Mio. in 2030, and will reach 315-411 Mio. in 2060, depending on different scenarios concerning global population and economic growth (Neumann et al., 2015). Examining 50 year trends Tanoue et al. (2016) find that the flood-exposed population is affected by historical changes in population distribution, with changes in flood vulnerability of up to 48.9%. Related to this development it takes no wonder that press reports on flash flood hazards become more frequent and can be found in the news almost every day. According to a World Bank study by 2050 there will be more than 143 Mio. internal climate refugees within the three regions of sub-Saharan Africa, South Asia and Latin America, if no action is taken until then (Rigaud et al., 2018). For Oman Al Qurashi (2012) has provided an overview on flood events and flood studies. Concerning the recent past most severe hazards were reported from cyclones making landfall in Northern Oman: Gonu in 2007 and Phet in 2010. In 2007 Gonu brought rainfall up to 714 mm in 24 hours and up to more than 900 mm in 36 hours. With a flood peak about 8160 m 3 /s Gonu caused severe damages and casualties in Muscat (Al Qurashi, 2012). In order to deal with flood events and to avoid or mitigate hazards various measures are discussed and put in operation. There is an increasing need to come up with relevant flood maps in order to cope with the requirements of active and pro-active flood management. Flood maps are important in different stages of disaster management (Webber et al., 2018). Flooding levels and especially hotspots (de Risi et al., 2018) can be identified in order to prepare for an approaching storm. They may indicate areas for which early warnings have to be issued (Wicht and Osinska-Skotak, 2016). During the impact they may give important clues concerning rescue and flood management decisions. Between events they are most important concerning the preparedness for prevention and mitigation. Flood hazard maps and vulnerability maps can be combined to flood risk maps that enable the users and urban planners to identify risk areas (Tingsanchali, 2012). Figure 1 shows examples of flood risk maps, prepared by the responsible administration in Oman in 1992. Flood prone areas are indicated by colours, red for hot-spots, and blue and yellow for adjacent risk zones of lower degree that are endangered in case of a more severe flood event.

Upload: others

Post on 08-Nov-2021

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: FLOOD RISK MAPPING IN LOW COASTAL REGIONS …

E-proceedings of the 38th IAHR World CongressSeptember 1-6, 2019, Panama City, Panama

doi:10.3850/38WC092019-1522

4005

FLOOD RISK MAPPING IN LOW COASTAL REGIONS CONSIDERING WATERSHED DELINEATION UNCERTAINTY

EKKEHARD HOLZBECHER(1) & AHMED HADIDI(2)

(1,2) German University of Technology in Oman (GUtech), Muscat, Oman e-mail [email protected]

(2) e-mail [email protected]

ABSTRACT

Low-lying flat regions with a mountainous hinterland are prone to flash floods. Reports of flooding events are reported more frequently, coming from all parts of the world. Measures to mitigate hazardous effects from flash floods are discussed and put in operation at many places, among others the identification of flood prone zones in flood maps. These maps are drawn on basis of the knowledge of watersheds and the main watercourses therein. However, in lowlands on alluvial fans or on sedimentary terrain the delineation of catchments and streams is uncertain for several reasons. Streams may change their course as effect of a flood event. Due to small topographic gradients measurement inaccuracies amount to a higher relative error in flow calculations. We take the Batinah coast of Northern Oman as an example to demonstrate the deviations in delineations of downstream watersheds, resulting from different digital elevation maps. Finally we come up with recommendations concerning the construction of flood maps in low-lying coastal regions and the use of watershed delineation tools available in GIS software.

Keywords: Flood maps, GIS, DEM, Watershed, Lowlands

1 INTRODUCTION The population in low elevation coastal zones is predicted to increase from 650 Mio. in 2000 to over 900

Mio. people in 2030 (Neumann et al., 2015), and may exceed 1.300 Mio in 2060. While in 2000 189 Mio. were living in 100-year flood plains this number will rise to approximately 275 Mio. in 2030, and will reach 315-411 Mio. in 2060, depending on different scenarios concerning global population and economic growth (Neumann et al., 2015). Examining 50 year trends Tanoue et al. (2016) find that the flood-exposed population is affected by historical changes in population distribution, with changes in flood vulnerability of up to 48.9%. Related to this development it takes no wonder that press reports on flash flood hazards become more frequent and can be found in the news almost every day. According to a World Bank study by 2050 there will be more than 143 Mio. internal climate refugees within the three regions of sub-Saharan Africa, South Asia and Latin America, if no action is taken until then (Rigaud et al., 2018).

For Oman Al Qurashi (2012) has provided an overview on flood events and flood studies. Concerning the recent past most severe hazards were reported from cyclones making landfall in Northern Oman: Gonu in 2007 and Phet in 2010. In 2007 Gonu brought rainfall up to 714 mm in 24 hours and up to more than 900 mm in 36 hours. With a flood peak about 8160 m3/s Gonu caused severe damages and casualties in Muscat (Al Qurashi, 2012).

In order to deal with flood events and to avoid or mitigate hazards various measures are discussed and put in operation. There is an increasing need to come up with relevant flood maps in order to cope with the requirements of active and pro-active flood management. Flood maps are important in different stages of disaster management (Webber et al., 2018). Flooding levels and especially hotspots (de Risi et al., 2018) can be identified in order to prepare for an approaching storm. They may indicate areas for which early warnings have to be issued (Wicht and Osinska-Skotak, 2016). During the impact they may give important clues concerning rescue and flood management decisions. Between events they are most important concerning the preparedness for prevention and mitigation. Flood hazard maps and vulnerability maps can be combined to flood risk maps that enable the users and urban planners to identify risk areas (Tingsanchali, 2012).

Figure 1 shows examples of flood risk maps, prepared by the responsible administration in Oman in 1992. Flood prone areas are indicated by colours, red for hot-spots, and blue and yellow for adjacent risk zones of lower degree that are endangered in case of a more severe flood event.

Page 2: FLOOD RISK MAPPING IN LOW COASTAL REGIONS …

E-proceedings of the 38th IAHR World CongressSeptember 1-6, 2019, Panama City, Panama

4006

Figure 1. Flood risk map examples, exerpts from MWR (1992)

A crucial task in the development of a flood risk map is the delineation of watersheds. Nowadays that step can be performed on the computer on basis of a DEM (Digital Elevation Model). DEMs have proven to be valuable tool for the topographic parameterization of hydrological models, which are the basis for any flood modeling process (Forkuo, 2013). We focus here on difficulties using an automated approach using GIS delineation tools based on DEMs, in particular for low coastal regions.

Watersheds are defined by the drainage divides, which separate one catchment from another. The advantage of the watershed approach for flood simulation is that it is not necessary to quantify fluxes across the boundaries, as these become zero by definition. For the utilization of the division into catchments and sub-catchments one has to be certain about the location of the boundaries, i.e. of the drainage divides. Here we focus on uncertainties resulting from inaccuracies of topography measurements.

GIS software allows the ‘automatic’ delineation of watersheds from a DEM. This task is usually performed in several steps (Jenson and J. O. Domingue, 1988). First a flow direction grid is created, which shows the steepest gradient at each grid location. In some cases an intermediate step is recommended to create a depression-less flow field, in which no sinks appear. Usually there are options to control the filling of the depressions. Depending on this option some (greater) depressions will remain. In the next step a flow accumulation grid is created, showing the number of upstream cells for each location. Downstream locations have a higher number of upstream cells than upstream positions. At raster points on drainage divides the number of upstream cells is zero. The next step is to create sub-watersheds for chosen outlet (pour) points. In an intermediate step chosen pour points have to be adjusted due to the flow accumulation maps and the raster. Finally GIS software converts watershed rasters to polygons, which represent the drainage divides. Figure 2, illustrating some steps, is taken from the manual of GIS software in which the handling of the watershed tool is described. Rahman et al. (2014) provide a detailed illustration of the steps involved.

Figure 2. Basic features of a GIS watershed toolhttp://resources.arcgis.com/en/help/main/10.2/index.html#/Understanding_drainage_systems/009z0000005m000000/

2 STUDY REGION The GIS watershed tool was applied to the Al Batinah region in Oman using different DEM maps. Al Batinah

is one of the nine Omani governates. From the country’s capital, Muscat, it extends along the coast to the NorthEast. The total area covering 12.500 km2 has a population of about 770.000 (2010), making it one of the highest populated parts of the country. Historically characterized by agriculture in the coastal strip and desert in

Page 3: FLOOD RISK MAPPING IN LOW COASTAL REGIONS …

E-proceedings of the 38th IAHR World CongressSeptember 1-6, 2019, Panama City,

Panama

4007

the hinterland, in the last decades the old urban centers and agricultural lands have dramatically shrunk under the pressure of urban sprawl (Benkari, 2017).

Geographically Al Batinah is located between the Hajar mountains and the sea. Geologically the dominating alluvium deposits originate from the erosion of the Hajar mountains that are composed of highly fractured siltstone, sandstone and limestone formations, as well as shallow marine carbonates. Clastics with grain size ranging from boulders to silt include Khabra deposits. These are the finest-grained materials accumulated in a vast khabra extending parallel to the coast on the northern side of the Batinah plain. Dune and sand sheets are the best outcrops supporting rainfall infiltration (Bechennec et al., 1986).

Only a small percentage of the rainfall events take place near the coast (80 mm/year), while most of the precipitation occurs at the Northern Oman mountains (330 mm/year) (Kwarteng et al., 2009; Ahmed and Askri, 2016). Due to rapid urban development associated with increased ground water consumption, seawater intrusion into the coastal aquifers affects wide areas along the Batinah coast (Ahmed and Askri, 2016; Al-Awadhi and Mansour, 2015). The development of seawater intrusion is reflected by the migration of agricultural activities away from the coast (Figure 3).

Figure 3. Landsat satellite imagery for South Al Batinah plain, 23.05.1986 and 10.04.2017; regions of agricultural activity are indicated by green color

3 WATERSHED DELINEATION

3.1 Maps Commonly watersheds have been delineated manually on basis of topographical maps.and of local

expertise. For Oman a satellite image atlas, covering the entire country, was edited by El-Baz (2002), prepared by the Center for Remote Sensing at Boston University. Aside from the satellite images, showing wadi streams, the watersheds are delineated. Figure 3 depicts two watersheds from the study region as examples.

Figure 4. South Batinah wadis Al-Rustaq (left) and Barka (right) according to (El Baz, 2002)

3.2 GIS tools

Page 4: FLOOD RISK MAPPING IN LOW COASTAL REGIONS …

4008

E-proceedings of the 38th IAHR World CongressSeptember 1-6, 2019, Panama City, Panama

Nowadays available DEMs allow the delineation of watersheds via GIS-software tools. A recent study evaluated nine DEMs with different origin and resolution, and tested those for a sub-watershed in a mountainous region in Tibet (Keys and Baade, 2019). Ray (2018) examined the performance of three DEMs for a regional watershed in India. In this study the special focus lies on the lowlands regions and the utilization for flood studies.

For the delineation tasks we used the Spatial Analyst toolkit of ARCGIS (ESRI 2018). The flow direction raster is created using default options. In an intermediate step (see above) we used the option to produce a completely sink-free flow field. Also for the final delineation task to obtain a raster of drainage basins, default options were chosen. Only one of the DEMs required special modifications of the original file in order to obtain reasonable results.

3.3 Results The Shuttle Radar Topography Mission (SRTM) is an international project spearheaded by NASA, in

partnership with the DLR and Italian Space Agency (Jarvis et al., 2008), aimed for the continental surface of the earth for all latitudes less than 60°, or 80%. It used interferometric synthetic aperture radar (SAR) technique. Absolute and relative height errors are below 10 m for most of the covered area. Datasets of 1 arcsecond and 3 arcseconds are freely available to the public. Files are distributed in both ArcInfo ASCII as well as in GeoTiff format.

Figure 5. Basins generated by ArcGIS tools using SRTM

The ASTER Global Digital Elevation Model (GDEM) was developed jointly by the U.S. National Aeronautics and Space Administration (NASA) and Japan’s Ministry of Economy, Trade, and Industry (METI) (NASA et al. 2009). It depends on producing single scene from stereo pairs token by near infrared cameras. It is available at intervals of 1 arcsecond for latitudes less than 83° and delivers data with a vertical accuracy within 20 m. The ASTER GDEM is distributed as Geographic Tagged Image File Format (GeoTIFF) files with geographic coordinates (latitude, longitude). The data are posted on a 1 arc second (approximately 30 meters at the equator) grid and referenced to the 1984 World Geodetic System (WGS84)/1996 Earth Gravitational Model (EGM96) geoid.

Page 5: FLOOD RISK MAPPING IN LOW COASTAL REGIONS …

E-proceedings of the 38th IAHR World CongressSeptember 1-6, 2019, Panama City,

Panama

4009

Figure 6. Basins generated by ArcGIS tools using GDEM

TerraSAR-X add-on for Digital Elevation Measurement (TanDEM-X) is the name of TerraSAR-X's twin satellite, a German Earth observation satellite using SAR (Rizzoli et al., 2012, 2017). The datasets are generally highly accurate: the absolute horizontal and vertical accuracy in the original dataset meets the goal of being within 10 m for 90% of all points and a relative accuracy of 2 m. The data are available in GeoTiff format.

Figure 7. Basins generated by ArcGIS tools using TanDEM, without specific modification of the data

Page 6: FLOOD RISK MAPPING IN LOW COASTAL REGIONS …

4010

E-proceedings of the 38th IAHR World CongressSeptember 1-6, 2019, Panama City, Panama

On order to obtain reasonable results with the TanDEM data some modifications of the original data set had to be performed. Major problem was the irregular distribution of values on the sea surface. Watersheds were not separated, as the algorithm connected them at some positions on the shore or slightly offshore. The dataset was corrected by the introduction of a separate layer for the sea surface and assigning a constant value to it.

4 DISCUSSION For comparison the results of the automated watershed delineation based the three chosen DEMs are put

together in Figure 8. As in the previous figures the boundaries of the basins are in red if based on the SRTM map, in blue for the GDEM map and in black for the TanDEM map.

Figure 8. Basins generated by ArcGIS Spatial Analyst tool using SRTM in red, GDEM in blue, TanDEM in black

In the upstream mountainous regions the GIS tools produce nearly identical results for all DEMs, confirming findings presented by Keys and Baade (2019). In the lowlands however, differences appear and they increase with the wadis approaching the coast. Near the coastline watershed delineation produces a plethora of small watersheds. It is in fact a well-known phenomenon in hydrological sciences that many local catchments accompany the regional, national or multi-national watersheds, the nearer the stream comes to its orifice. As this outcome itself is thus expected, the most important observation is that the deviations in the alignment of the local catchments apparently increase towards the coast, as clearly depicted in Figure 8. Along the shoreline there is hardly a coincidence in the output from the three examined DEMs. There are several reasons that explain this finding, which are noted below.

5 SUMMARY AND CONCLUSIONS In low coastal regions the use of watershed delineation tools has to cope with various severe issues.

Elevation values are near sea level, which becomes an issue, as data imperfections and measurement errors gain more relative weight. For the computation of flow accumulation it is even more important that the small topographic gradients are likely to show an even higher relative error. These factors clearly make the issue of data uncertainty much more important in the lowlands and explains, why different DEMs produce differences in watershed delineations and that such deviations increase when the coast line is approached. Differences in watershed delineation will lead to different flow maps and thus also to different flood risk estimations as a result.

Page 7: FLOOD RISK MAPPING IN LOW COASTAL REGIONS …

E-proceedings of the 38th IAHR World CongressSeptember 1-6, 2019, Panama City,

Panama

4011

In order to outweigh or soften the problem, high quality DEM are required, as also demonstrated by Alganci et al. (2018).

Another issue is the change of watersheds as result of a flooding event. Knox (1989) already stated, that the mobility and storage of sediments in watersheds is an episodic process occurring at nearly all time scales and that gradual accumulation of sediment often is followed by relatively abrupt erosion and transportation of that sediment. Rivers, gullies and wadis may change their course with every flash flood event. Recent studies on the importance of sediment transport during floods in semi-arid environments were presented by Berghout and Meddi (2016) and Gharbi et al. (2016). Concerning the utilization in flood studies, it can be concluded that there is a need for regular updating of DEMs in the coastal region!

Figure 9. Sketch to illustrate how a drainage divide in normal or low flood conditions (left) becomes irrelevant in case of a high flood event (right)

Another issue is of particular importance in low coastal regions. Drainage divides as identified on basis of a DEM may become irrelevant in case of a sufficiently strong flood event. In case that a drainage divide is inundated it looses is function as a water divide, as illustrated in the sketch of Figure 9. The flow direction grid, calculated in the first step by a GIS watershed tool, is not valid anymore, if the area is inundated. Water divides, as identified by the GIS delineation tools, are only relevant, if the water level is low. If the drainage divide is inundated in a flood event, the water division is not relevant anymore. The practical consequence is that flood modelling, based on the shallow water equations, has to be performed on a larger spatial scale.

Figure 10. Inundation map computed by flood modelling using 3DI (2019); example for wadi Majraf-Manumah in the study region (Hadidi et al., 2019); inundated area is shown in blue colour; the inserted graph

shows the rising water table at the position indicated by (i) in the map in dependence of time

Figure 10 depicts an inundation map as output of flood modelling as an example. The region of a flood model needs to be extended towards boundaries, which are not flooded. Thus the choice of the model region

Ñ

Ñ

drainage divide

no drainage divideÑ

Page 8: FLOOD RISK MAPPING IN LOW COASTAL REGIONS …

4012

E-proceedings of the 38th IAHR World CongressSeptember 1-6, 2019, Panama City, Panama

requires already an anticipation of the highest possible flooding level. GIS tools may assist in the process by the choice of convenient pour points, which are assumed to remain dry. The higher the expected flood level, the larger is the model region.

We conclude that automatic watershed delineation is not a suitable tool for flood risk mapping in regions with small topographic gradients, as in coastal lowlands. There are three major reasons for this. (1) Accuracy errors in the DEMs have a more relative weight for small elevation heights. (2) Watershed boundaries are shifting in response to flood events; a DEM may not reflect the current state relevant for a flood to come. (3) Water divides are relevant only for not inundated regions, and become irrelevant in case of inundation.

ACKNOWLEDGEMENTS The presented research is enabled as part of the project funding from The Research Council (TRC) of the

Sultanate of Oman under Research Agreement No. ORG/ GUTECH /EBR/13/026.

REFERENCES 3Di Water Management (2019). http://docs.3di.lizard.net/en/stable/, accessed 9.5.2019 Ahmed, A. T., and Askri, B. (2016). Seawater intrusion impacts on the water quality of the groundwater on the

northwest coast of Oman. Water Environment Research. 88, 732-740. Al-Awadhi, T., and Mansour, S. (2015). Spatial assessment of water quantity stress in Sultanate of Oman

provinces: A GIS based analysis of water resources variability. Scientific Research Publication, Journal of Geographic Information System, 7, 565-578.

Alganci, U., Besol, B., and Sertel, E. (2018). Accuracy assessment of different digital surface models. ISPRS International Journal of Geo-Information, 7(3), 114.

Al-Qurashi, A. (2013). An overview on flood studies in Oman. In: UNESCO, Int. Seminar on Natural Disaster Risk Reduction, Proc., 25-35.

Bechennec, F., Beurrier, M., Rabu, D., and Hutin, G. (1986). Geological map of Barka, Explanatory notes, Sheet NF 40-3B scale 1:100000, Ministry of Petroleum and Minerals, Oman.

Benkari, N. (2017). Urban development in Oman: an overview. In: WIT Transactions on Ecology and the Environment, 226, WIT Press, 143-156.

Berghout, A., and Meddi, M. (2016). Sediment transport modelling in wadi Chemora during flood flow events. J. of Water and Land Development, 31, 23-31.

El-Baz, F. (2002). Wadis of Oman: Satellite Image Atlas. Stacey International, London. ESRI Environmental Systems Research Institute (2018). ArcGIS Desktop, Redlands, CA,

http://www.arcgis.com. Forkuo, E.K., and Tsawo, V.A. (2013). The use of digital elevation models for watershed and flood hazard

mapping. Int. Journal of Remote Sensing & Geoscience, 2, 56-65. Gharbi, M., Soualmia, A., Dartus, D., and Masbernat, L. (2016). Floods effects on rivers morphological changes

application to the Medjerda River in Tunisia. J. Hydrol. Hydromech., 64, 56–66. Hadidi, A., Holzbecher, E., and Molenaar, R. (2019) Flood mapping in face of rapid urbanization: a case study

of wadi Majraf-Manumah, Muscat, Sultanate of Oman. Urban Water Journal (submitted) Jarvis, A., Reuter, H.I., Nelson, A., and E. Guevara (2008). Hole-filled seamless SRTM data V4,

International Centre for Tropical Agriculture (CIAT), http://srtm.csi.cgiar.org. Jenson S.K., and Domingue J.O. (1988) Extracting topographic structure from digital elevation data for

geographic information system analysis. Photogrammetric Engineering and Remote Sensing, 54(11), 1593-1600.

Keys, L, and Baade, J. (2019). Uncertainty in catchment delineations as a result of digital elevation model choice. Hydrology, 6(13), doi:10.3390/hydrology6010013.

Knox, J.C. (1989). Long- and short-term episodic storage and removal of sediment in watersheds of southwestern Wisconsin and northwestern Illinois. In: Sediment and the Environment, Proc. of the Baltimore Symp., IAHS Publ. No. 184, 157-164.

Kwarteng, A. Y., Dorvlo, A. S., Vijaya Kumar, G. T., 2009, Analysis of a 27-year rainfall data (1977-2003) in the Sultanate of Oman. International Journal of Climatology, 29, 605-617.

Moreira, A., Krieger, G., Hajnsek, I., Hounam, D., Werner, M., Riegger, S., and Settelmeyer, E. (2004). TanDEM-X: a TerraSAR-X add-on satellite for single-pass SAR interferometry. IEEE International Geoscience and Remote Sensing (IGARSS 2004).

MWR (Ministry of Water Res., Sultanate of Oman) (1992). Delineation of High, Medium, Low & Index Flood Risk Zones, Muscat.

NASA/METI/AIST/Japan Spacesystems, and U.S./Japan ASTER Science Team (2009). ASTER Global Digital Elevation Model [Data set], NASA EOSDIS Land Processes DAAC, doi: 10.5067/ASTER/ASTGTM.002.

Neumann, B., Vafeidis, A.T., Zimmermann, J., and Nicholis, R.J. (2015). Future coastal population growth and exposure to sea-level rise and coastal flooding - a global assessment. PLoS ONE, 10(3), e0118571.

Page 9: FLOOD RISK MAPPING IN LOW COASTAL REGIONS …

E-proceedings of the 38th IAHR World CongressSeptember 1-6, 2019, Panama City,

Panama

4013

Rahman, M., Rahman, M., and Khan, M.R. (2014). Small-Scale Catchment Delineation in Coastal Area of Bangladesh: A GIS Based Approach . International Journal of Engineering Research & Technology, 3(9), 1202-1208.

Rexer, M., and Hirt, C. (2014). Comparison of free high-resolution digital elevation data sets (ASTER GDEM2, SRTM v2.1/v4.1) and validation against accurate heights from the Australian National Gravity Database. Australian Journal of Earth Sciences, 61(2), 1-15.

Ray, L.K. (2018). Limitation of automatic watershed delineation tools in coastal region. Annals of GIS, 24(4), 261-274.

Risi, R. de, Jalayer, F., Paola, F. de, and Lindley, S. (2018). Delineation of flooding risk hotspots based on digital elevation model, calculated and historical flooding extents: the case of Ouagadougou. Stoch. Environ. Res. Risk Assess., 32, 1545-1559.

Rizzoli, P., Bräutigam, B., Kraus, T., Martone, M., and Krieger, G. (2012). Relative height error analysis of TanDEM-X elevation data. ISPRS Journal of Photogrammetry and Remote Sensing, 73, 30-38.

Rizzoli, P., Martone, M., Gonzalez, C., Wecklich, C., Borla Tridon, D., Bräutigam, B., Bachmann, M., Schulze, D., Fritz, T., Huber, M., Wessel, B., Krieger, G., Zink, M., and Moreira, A. (2017). Generation and performance assessment of the global TanDEM-X digital elevation model. ISPRS Journal of Photogrammetry and Remote Sensing, 132, 119-139.

Rigaud, K., Alex de Sherbinin, K., Jones, B.,Bergmann, J., Clement, V., Ober, K., Schewe, J., Adamo, S., McCusker, B., Heuser, S., and Midgley, A. (2018). Groundswell: Preparing for Internal Climate Migration, The World Bank, Washington DC.

Tanoue, M., Hirabayashi, Y., and Ikeuchi, H. (2016). Global-scale river flood vulnerability in the last 50 years. Nature Sci. Rep., 6:36021, doi: 10.1038/srep36021.

Tingsanchali, T. (2012). Urban flood disaster management. Procedia Eng., 32, 25-37. Webber, J.L., Gibson, M.J., Chen, A.S., Savic, D., Fu, G., and Butler, D. (2018). Rapid assessment of surface-

water flood-management options in urban catchments. Urban Water Journal, 15(3), 210-217. Wicht, M., and Osinska-Skotak, K. (2016). Identifying urban areas prone to flash floods using GIS - preliminary

results. Hydrol. Earth Syst. Sci. Diss.