i
FLASH FLOOD MODELING FOR UM QASIR RIYADH CITY, SAUDI ARABIA
Husam Abdullah Alshawi
A thesis submitted for the requirements of the degree of Master of Science in Geographical Information Systems (UNIGIS)
Supervised By Prof. Amer Althubaity
Faculty of Arts and Humanities KING ABDULAZIZ UNIVERSITY
JEDDAH – SAUDI ARABIA
Rabi 1 1440H – November 2018G
ii
iii
FLASH FLOOD MODELING FOR UM QASIR RIYADH CITY, SAUDI ARABIA
Husam Abdullah Alshawi
A thesis submitted for the requirements of the degree of Master of Science in Geographical Information Systems (UNIGIS)
Supervised By Prof. Amer Althubaity
Faculty of Arts and Humanities KING ABDULAZIZ UNIVERSITY
JEDDAH – SAUDI ARABIA
Rabi 1 1440H – November 2018G
iv
FLASH FLOOD MODELING FOR UM QASIR RIYADH CITY, SAUDI ARABIA
Husam Abdullah Alshawi
This thesis has been approved and accepted in partial fulfillment of the requirements for the degree of Master of Science in Geographical Information Systems (UNIGIS)
EXAMINATION COMMITTEE
Name Rank Field Signature
Internal Examiner External Examiner Advisor
KING ABDULAZIZ UNIVERSITY
Rabi 1 1440H – November 2018G
v
FLASH FLOOD MODELING FOR UM QASIR RIYADH CITY, SAUDI ARABIA
Husam Abdullah Alshawi
Abstract
Uncontrolled urban expansion and increased rainfall events, in certain regions of Saudi Arabia, have aggravated the impacts of frequent flash flooding on infrastructures and human life in recent years. With the availability of high-resolution datasets, detailed Terrain information resulting from the frequent ongoing aerial survey, it is promising for modeling and analyzing flash flood in urban areas more accurately using such higher precision data derived from LiDAR.
The purpose of this thesis was to evaluate and compare low and high-resolution elevation models and their applications for assessing and identifying flooding risk potential areas in Um Qaser Areas in Riyadh, Saudi Arabia. This study focused on generating water accumulation areas and direction of water flow using both the resolutions of DEMs to help ascertain the capabilities of LiDAR dataset.
It was found that LiDAR terrain data successfully identified 59 points of water accumulation , an increase of about 50% than processed Satellite ASTER 30m Elevation data which only identified 26 points. Processing LiDAR based terrain data also resulted in the generation of hydro network information in better detail than satellite DEM. It seems that LiDAR datasets producing high resolutions Terrain Models are ideal for dense focused urban areas for flood monitoring and analysis studies where lower resolutions DEM of 30m or high scale and wider coverage areas and/or semi urban rural cities. More future work needs to be carried out to integrate cadastral information to the DTM processing of LiDAR for higher accurate water flow network generation leading to better modeling
vi
ةجذمن نايرج
تاناضيفلا نمب ةئجافملا
قط ة رصق مأ ضايرلا ةنيدم يف
يواشلا اللهدبع نب ماسح
صلختسملا
تديازت جافملا تاناضيفلا ةئ
لويسلا يراجم يف ينارمعلا عسوتلا اهنم لماوع ةدعل ةجيتن ةريخلأا تاونسلا للاخ
ةيدولأاو
، عافتراو و ةريطخ تاديدهت يف ببست امم راطملأا لوطه تايوتسم ةيتحتلا ةينبلاو حاورلأا يف ةريبك رئاسخ
لثم ومنلا ةعيرس ةيرضح ةنيدم يف اميسلا ةنيدم
.ضايرلا
و ناكسلاب ةلوهأملا ءايحلأا دحأ ىلع راطملأا تاناضيف راطخأ ىوتسم مييقت يف ةساردلا هذه ةيمهأ زربت انه نم
و يتلا رد ةقطنمك( رصق مأ يح وهو ضايرلا ةنيدم يف ليسلا ىرجم يف عقت )ةسا
، ىلع ةساردلا تدمتعا دقو
ةقدلا ةيلاع تانايب نم تاعومجم
، ةفاضلإاب رمتسملا يوجلا حسملا نع ةجتانلا سيراضتلا نع ةيليصفت تامولعمو
ل ةجذمنل ةيثلاث ةاكاحملاو يحلا ميمصت اهيلع ماق يتلا جئاتنلل ةرياغمو ةقوثوم جئاتن ةيجهنملا هذهل ناكو ،داعبلأا
ارظن لإل بكلا فلاتخ ةعبتملا ةيجهنملا فلاتخاو تامولعملا رداصم ةقد يف ري
.
تانايب مادختسا نأ ىلإ ةساردلا جئاتن ريشتو LiDAR
ىلع تدعاس ديدحت
59 لويسلا هايم عمجتل ةطقن ةقطنم يف
كلذو ةساردلا يلاوح قرافب
50
% تانايب ىلع دامتعلاا نع ASTER
و تددح يتلا 26
نأ ىلإ ةفاضإ ؛طقف ةطقن
ع دامتعلاا تانايب ىل
LiDAR تامولعم ةكبش جاتنا يف دعاسي ةيجولورديه
أ لضف .ةيقوثومو ةقد رثكأو
vii
Table of Contents
Abstract ... v
صلختسملا ... vi
Table of Contents ... vii
Table of Figures ... ix
CHAPTER I INTRODUCTION ... 1
1.1. Objective and Justification ... 2
1.2. PROCEDURES ... 3
1.3. Study Area ... 4
CHAPTER II LITERATURE REVIEW ... 6
2.1. Flash Floods and Hazards ... 6
2.2. Flood Hazard Assessment... 7
2.3. Flood Modeling Approaches ... 8
2.4. Remote Sensing and GIS Technologies for Flood Mapping ... 9
2.4.1. Photogrammetry ... 10
2.4.2. Light Detection and Ranging (LiDAR) ... 14
2.4.3. GIS Hydrographical Tools: ArcHydro, HEC-RAS ... 16
2.5. Analytical Hierarchical Process ... 17
CHAPTER III Methodology ... 19
3.1. Data Collection and Preparation ... 22
3.2. Data Modeling and Hydrological Processing ... 25
3.3. Aerial LiDAR Processing and DEM/DTM Generation ... 25
3.4. Aerial LiDAR Issues and Fixes ... 28
viii
3.4.1. LiDAR Missing Ground Elevation Data ... 28
3.4.2. LiDAR Data Missing Projection Definition ... 29
3.4.3. Terrain Data Processing from LiDAR DEM ... 32
3.4.3.1. DEM Leveling and Pre-processing ... 32
3.4.3.2. Flow Direction, Accumulation and Stream Generation ... 33
3.4.3.3. Catchment Area Identification ... 35
3.4.3.4. Longest Water Flow Path Generation: ... 36
3.4.3.5. Hydro Network Generation: ... 37
3.4.3.6. Watershed Delineation: ... 39
3.4.4. Terrain Data Processing and Modeling using ASTER DEM ... 40
3.4.4.1. DEM Leveling and Pre-processing: ... 41
3.4.4.2. Flow Direction, Accumulation and Stream Generation ... 41
3.4.4.3. Catchment Area Identification ... 43
3.4.4.4. Hydro Network Generation ... 44
3.4.4.5. Watershed Delineation: ... 45
CHAPTER IV RESULTS AND DISCUSSION ... 47
4.1. Flood Risk Potential Maps ... 47
4.2. Summary of the results ... 53
4.3. Simulation Models of Flood Risks... 54
CHAPTER V CONCLUSION AND RECOMMENDATIONS ... 61
Bibliography ... 65
ix
Table of Figures
Figure 1 Study Area . ... 5
Figure 2 Photogrammetry ... 10
Figure 3 Aerial Film Scanner ... 11
Figure 4 Airborne digital camera ... 12
Figure 5 Configuration of multiple linear CCD arrays ... 12
Figure 6 Stereo data capture ... 13
Figure 7 Satellite Sensors ... 14
Figure 8 Typical Airborne Laser Scanning scenario ... 15
Figure 9 LiDAR DSM ... 16
Figure 10 Continuous Rating Scale for Pairwise Comparison ... 18
Figure 11 Workflow for pre-processing LiDAR derived DEM ... 20
Figure 12 Workflow for pre-processing ASTER satellite derived DEM ... 21
Figure 13 Integration of Satellite, Aerial Imagery and LiDAR/DEM ... 21
Figure 14 Aerial LiDAR data showing elevation ... 22
Figure 15 Derived Slopes from DEM ... 23
Figure 16 Study Area GIS Data - Source: ADA ... 23
Figure 17 Geological Formation for Riyadh City ... 24
Figure 18 Aerial LiDAR point cloud datasets ... 26
Figure 19: 3D view of the LiDAR Elevation Data using ERDAS ... 26
Figure 20 Ground cover LiDAR data converted to surface model ... 27
Figure 21: DTM Subset to Study area viewed as shaded relief model ... 28
Figure 22 Problem Areas ... 29
Figure 23 Study Area Geoeye-1 Satellite Image ... 30
x
Figure 24 LiDAR view in viewer due to missing projection definition ... 31
Figure 25 Defining Projection ... 32
Figure 26 Leveled and Filled DEM ... 33
Figure 27 Flow Direction generation using Terrain ... 33
Figure 28 Flow Accumulation generation from Terrain ... 34
Figure 29 Streamlines overlayed on Catchment Areas ... 35
Figure 30 Adjoint Catchment Areas and Drainage Points for the Process Output ... 36
Figure 31 Longest Flow path ... 36
Figure 32 HydroJunction and HydroEdge Generation ... 38
Figure 33 HydroNetwork and Longest WaterFlow Path Process Output ... 38
Figure 34: Watershed points and generated Watersheds of area ... 39
Figure 35: Water flow path per watersheds ... 39
Figure 36 Satellite: Leveled and Filled DEM ... 41
Figure 37 Flow Direction ... 41
Figure 38 Flow Accumulation ... 42
Figure 39 Water Streams ... 42
Figure 40 Catchment Areas Grid, Polygons and Drainage Lines ... 43
Figure 41 Drainage Points for after processing ... 44
Figure 42 HydroJunction and HydroEdge Generation ... 44
Figure 43 HydroJunction and HydroEdge ... 45
Figure 44 Watershed points and their associated watershed polygons ... 46
Figure 45 Water Stream Flow Direction Map (LiDAR) ... 48
Figure 46 Water Stream Flow Direction Focused Map ... 48
Figure 47 Water Stream Flow Direction Focused Map ... 49
xi
Figure 48: LiDAR: Catchment Areas Map ... 49
Figure 49: Satellite: Catchment Areas Map ... 50
Figure 50: LiDAR: Longest Water Flow Path Map ... 50
Figure 51: Satellite: Longest Flow Path Map... 51
Figure 52: LiDAR: Water Accumulation/Risk Areas Map ... 51
Figure 53: Satellite: Water Accumulation Risk Areas Map ... 52
Figure 51: Satellite: Water Drainage Streams and Flow Direction Map ... 53
Figure 52: Ground Elevations ... 55
Figure 53: Non-Ground Elevation ... 55
Figure 54: Water Layer ... 56
Figure 55: Water Level Simulated 3D Map ... 57
Figure 56 Satellite Imagery drape with Non Ground Elevation ... 58
Figure 57 Zoomed in 3D Map View 1 Map ... 58
Figure 58 Zoomed in 3D Map View 2 Map ... 59
Figure 59: 3D Visualization with simulated 30mm rainfall data. ... 60
Figure 60: 3D Visualization with simulated 110mm rainfall data. ... 60
1
CHAPTER I INTRODUCTION
Due to the frequently increased environmental hazards, urban flooding has become a major threat in the recent years. Flash flooding causes life losses, infrastructure damages, and spread of diseases. Assessing and managing the effects are significantly challenging especially in a rapidly developing urban city such as Riyadh, which is currently experiencing greater infrastructure activities and major Metro Rail, Financial economic districts and new residential district projects leading to increase in population density. The Riyadh city has been hit by more than 12 flash floods since 1985 and more than 9 flash floods since 2009 every year (Rahman, Aldossary, Kh Md & Reza, 2016).
Geographic Information System (GIS), Remote Sensing Imagery and data such as Light Detection and Ranging (LiDAR) and Digital Elevation Model (DEM) have been used for modeling urban flooding in several parts of the world (Ghazali and Kamsin, 2008). DEM has been used primarily to define the water surface flow networks in flooded areas including pathways and temporary ponds, and sinkholes.
This thesis comprises of 5 chapters. The first chapter describes the introduction to the study, the second chapter presents the literature review, and third chapter deals with the study area and procedures. The fourth chapter presents the results and discussion, and the fifth chapter provides information about the conclusion and recommendations.
2
1.1. Objective and Justification
This study seeks to show how the integration of Remote Sensing and GIS would help in modeling the flash flood impact in Riyadh region using datasets available from newer technologies such as LiDAR.
Flash Flooding in Riyadh occurs as a consequence of excessive rapid rainfall, which has increased recently over a few years (Rahman, Aldossary, Kh Md & Reza, 2016) possibly because of climate fluctuations, lack of proper urban development planning and / or implementation as well as construction activities taking place. These extreme natural events have become a source of great loss to properties and lives in the last 10 years or so, especially in Riyadh City and surrounding areas. Flood extents and depth are usually considered the most important flood parameters, especially when it comes to mapping flood hazards (Mugisha, 2015).
The main aim of this study is to augment the urban planning by determining the effectiveness of LiDAR based elevation data in comparison to satellite-derived DEM by visualizing and comparing the Hydro Network and accumulated risk areas. It focuses on:
1. Generating hydro drainage network and identifying areas that can lead to potential water collection.
2. Identifying flash flood risk using LiDAR and satellite-based elevation data in comparative manner.
3. Visualization of the study area with flood risk areas in 3D.
3
1.2. PROCEDURES
Idea is to study and analyze selected study areas in catchment characteristics, drainage network and flood simulation, which will assist in analyzing the impacts on the selected area. The methodology will be detailed in separate chapter.
The entire data will be processed using ERDAS Imagine for Remote Sensing processing and data preparation of DEM as well as 3D data from LiDAR, where ESRI ArcHydro is used for modeling processes. Rainfall data was obtained from The General Authority of Meteorology and Environmental Protection. The outcomes of this study will support and improve the daily activities of all relevant agencies, such as Municipality, ArRiyadh Development Authority, Civil Defense … etc.
The process involved High-resolution DEM generation and pre-processing from a previously conducted Riyadh Aerial Imagery, LiDAR Survey in 2011-2012 by ArRiyadh Development Authority ADA. This would be the key input for ArcHydro and other tools for flood hazard mapping, water flow generation, and 3D visualization along with Aerial and Satellite Imagery with flood inundation in the study area.
DEMs are characterized by different resolutions. On the one hand, the spatial resolution of low-cost DEMs from satellite imagery, such as ASTER and SRTM, is coarse around 30 to 90m. On the other hand, the LiDAR technique is able to produce high-resolution DEMs at around 1m (A. Md Ali et al, 2015).
As a comparative study, DEM was derived from ASTER sensor etc. Lower resolution previous/existing data sources were also processed side by side to access the accuracy and detail of the results in relation to the LiDAR-derived datasets output. This study is needed to be similar to A. Md Ali et al, 2015 in aim in order to determine whether the quality and accuracy of the DEM are more important than the resolution and precision of the DEM.
The final stage involved 3D visualization of the flooded areas using ArcGIS and ERDAS software as well as the production of flood map of the study area.
4
1.3. Study Area
The selected area for this study was Um Qaser, a residential district that is located in the Southwest region of Riyadh, the capital city of Saudi Arabia, as shown in Fig 1. The area is about 8.4 km2 and the population is more than 72 thousand. The area was selected due to its topography which is of interest due to varying elevations as well as wadi and low- lying areas. It is surrounded by a heavily built-up urban area that can be directly and greatly affected by any heavy flashing in the region.
According to ArRiyadh Development Authority Urban Indicators of 2015, the total developed urban area in Riyadh city was 2395 sq. km, and the area allocated for urban development is expected to be close to 3115 sq. km by the year 2029. The city consists of 211 districts and comprises of 13 municipalities including the Riyadh governorate.
Riyadh is a multi-cultural city with 65% Saudi and 35% non-Saudi population. It has been rapidly expanding. Climate of the city: hot dry summer and cool winter with humidity in the air. Rainfall has been typically ranging from 85mm to 116m for a period of about 20 years prior to 2005 (Nahiduzzaman, Kh Md & Aldosary, Adel & Rahman, Muhammad Tauhidur ,2015). However, in the recent past, weather patterns have changed locally, regionally as well as globally resulting in heavy downpours, lasting from hours to days more frequently year on year, causing high volume run-off (ADA, 2013).
5
Figure 1 Riyadh City, Saudi Arabia Capital, Red Highlighted is the Study Area.
6
CHAPTER II LITERATURE REVIEW
This chapter deals with the review and summary of various studies in relation to flood hazards and assessment, methods, modern tools, technologies and data requirements for using remote sensing and GIS to produce flood hazard maps and other visual representation of floods such as 3D views.
There have been numerous studies over the years in many countries to carry out mapping of flash floods, for example in United States (Mastin, 2009), China (Liang et al., 2011), Egypt (El Bastawesy et al., 2009; Ghoneim et al., 2002), Saudi Arabia (Saud, 2010; Dawod et al., 2011), India (Bhatt et al., 2010), and Ghana (Forkuo, 2011).
2.1. Flash Floods and Hazards
Flooding is the process of water overflowing onto land surface areas. It might happen during heavy rains, or when rivers overflow due to heavy waters, or when dams and/or levees break etc.
Flash flooding is a heavy onset of water on land area in a very short period of time. It occurs rapidly, generally within one hour of rainfall, and is sometimes accompanied by landslides, mudflows, ridge collapse, damage to buildings, and fatalities (Hapuarachchi et al., 2011). Flood extents and depth are usually considered the most important flood parameters, especially when it comes to mapping flood hazards (Mugisha, 2015).
7
Riyadh region is heavily arid in nature with dry atmosphere. In these regions, flash floods often cause significant damage to infrastructure, human lives, and properties. Moreover, people in these areas are not adequately aware of flooding consequences due to the occurrence of high dryness and low annual rainfall period (Subyani, 2012). Thus, land with a high risk of flash flooding is carelessly developed for settlement and infrastructure purposes (Subyani, 2012).
2.2. Flood Hazard Assessment
Flood hazard assessment involving creating and maintaining relevant flood information and flood maps of an area are important for policy formulation and its implementation for flood management initiatives.
One of the first steps to do so is to review the flooding history of the selected area.
However, this may not always be available or relevant since flash flooding is highly uncertain phenomena in terms of prediction of future location and strength as well as regular Landover changes in urban areas overtime (Bellos, 2012). Overcoming these problems can involve the use of statistical and modeling tools to calculate the hazards in a hypothetical scenario. There are various parameters that can be used to symbolize such flood hazards (Mugisha, 2015). These may include flood extents, water depth, flow velocity, duration, warning time, and the rate at which the water rises (Alkema, 2007).
While dealing with current and recent floods, the use of remote sensing technologies and satellite imagery is very helpful in determination of flood extents (Mugisha, 2015). The modeling approach helps understand the flood hazard and serves as a tool that is useful
8
in understanding the effects of mitigation measures (Bhattacharya, 2010). Taking flow direction into account, these models are nowadays available in many types, from simple 1D models to complex 2D models (Prachansri, 2007).
2.3. Flood Modeling Approaches
Models are a simplified representation of realities. They are the instruments that are used to mimic and provide insight into process or phenomenon, such as flow of water in channels, urban expansion etc. Although simplified, they are powerful tools to simulate and predict the implication of certain actions in the future (Couclelis, 2005).
Different types of flood models exist (Pender and Neelz, 2007); some of the examples include 1D and 2D flood models. These models have the principle of conservation of mass, momentum, and energy (Horritt and Bates, 2002).
If water is confined in a channel, it is best simulated as an unsteady 1D-flow model (Verwey, S.A. 2006). This 1D model defines flood only in terms of discharge and water level as a function of space and time (Mugisha, 2015). This approach has limitations of simulating floods in the overland. This situation causes a shift from 1D to 2D model that simulates flood in the overland (Horritt and Bates, 2002). 2D flood model predicts flood inundation based on 2D shallow water equation (Mignot et al., 2006).
Various approaches have been created as part of research and development activities around the world to integrate both 1D and 2D approaches for producing effective hydrodynamic models of flood plain (Mugisha, 2015), such as SOBEK (Alkema, 2007),
9
SWI2D (Finaud-Guyot et al., 2011), LISFLOOD and TELEMAC -2D (Horritt & Bates, 2002), and OpenLISEM (De Roo et al., 1996).
2.4. Remote Sensing and GIS Technologies for Flood Mapping
Due to rapid development of GIS technology and remote sensing techniques, availability and production of high resolution DEM using high resolutions stereo-paired imagery as newer technologies such as LiDAR as well as advancements in integrating GIS with hydrological modeling, flood prediction with distributed models tends to be more advantageous and competent by linking GIS with hydrological modeling (F. De Smedt et al., 2004).
Digital elevation model (DEM) is the most important input of the hydrological modeling to get flood maps. The precision of watershed calculation is directly dependent on the scale and precision of topographic maps (Elkhrachy, 2015). Availability of high-resolution stereo pair imagery and advancements in computer processing power and new algorithms and processes in software have resulted in high-resolution DEM generation from remotely sensed datasets. For example, using ERDAS IMAGINE Photogrammetric system capabilities, such as Automatic Digital Terrain Modeling and extraction to generate higher resolution Points cloud similar to imagery resolution yields high-resolution DEMs. In addition, greater availability of LiDAR data for DEM generation via Aerial Surveys in much higher resolutions than the ones derived from satellite stereo images has also produced much higher detailed elevation models, very beneficial for targeted study area flooding studies.
10
2.4.1. Photogrammetry
“Photogrammetry is the science of obtaining reliable information about the properties of surfaces and objects without physical contact with the objects and measuring as well as interpreting this information” (Schenk T., 2005).
According to American Society of Photogrammetry and Remote Sensing (ASPRS), photogrammetry is the art, science and technology of obtaining reliable information about physical objects and the environment through processes of recording, measuring, and interpreting photographic images and patterns of recorded radiant electromagnetic energy and other phenomena (National Research Council, 2007). Outputs created from the photogrammetric process may include:
Digital Elevation Models
Orthophoto imagery
2D maps and 3D Feature datasets
Figure 2 Photogrammetry uses multiple views of the same point on the ground from two perspectives to create a three-dimensional image. SOURCE: Image courtesy of David Maune, Dewberry and Davis.
11
Three types of sensors are used in the photogrammetry to produce imagery and mapping products; film cameras, digital cameras, and satellite imaging sensors.
Airborne film sensors have been used in the past to produce high precision mapping products having an additional chemical process to produce the films for scanning. The aerial film cameras are no longer in use. Due to aerial digital camera’s increasing maturity, higher accuracy, resolution, flexibility and cost-effectiveness as well as quicker capture processing and digital data production time, only scanners exist to scan older captured films; one example of the scanner is shown in Fig 3.
Figure 3 Aerial Film Scanner – Leica DWS700 Source: Leica Geosystems.
Digital cameras provide superior performance, increased spatial and spectral resolutions and greater product production from single capture, for example panchromatic, red, blue, green, and infrared as illustrated in Fig 4,5, and 6.
12
Figure 4 With an airborne digital camera, images can be captured simultaneously in true color (RGB), false-color infrared (CIR), and gray-scale (also called panchromatic) (PAN). SOURCE: Fugro Earth Data
International
Figure 5 Configuration of multiple linear CCD arrays for the Leica ADS40 airborne digital camera. SOURCE:
Leica Geosystems
13
Figure 6 Stereo data capture with the ADS40 push-broom sensor. SOURCE: Leica Geosystems.
High-resolution satellite imagery is available from a number of commercial sources from various sensors, such as Geoeye, Worldview, KOMPSAT, Quickbird, PLEADEAS, IKONOS etc.
as shown in Fig 7, having spatial resolutions ranging from 1m to 0.35m with High- Resolution RGB and multispectral imagery bands. Majority of these sensors provides stereo pair imagery products used in photogrammetry.
14
Figure 7 Satellite Sensors – Source: Introduction to EO Data at
http://www.charim.net/sites/default/files/handbook/flowchart/DMB/33/Intro_EO_ff/index.html [Accessed: 10th Aug 2016]
2.4.2. Light Detection and Ranging (LiDAR)
Airborne Laser Scanning (ALS) technology is one of the highly accurate methods of collecting large volume; geo-referenced 3D data (Weed, 2000; Shan, Toth, 2008).
LiDAR is an active remote sensing technology that uses a laser to measure distances for targeting points as shown in Fig 8. Since it generates its own energy, LiDAR survey can be conducted any time of day or night, even in slightly cloudy or hazy conditions (National Research Council, 2007). Using ALS, we collect “point cloud” data with three coordinates, X, Y, and Z representing latitude, longitude, and height above sea level.
These data contain additional information such as intensity and number of return (Korzeniowska and Łacka, 2011). To obtain a Digital Elevation Model (DEM), 3D buildings, power lines or various types of objects represented in the “point cloud”, high-quality classification of “point cloud” data is needed (Wang, Tseng, 2010). Most of this software tools offer scripts and algorithms for “point cloud” data classification to land cover classes,
15
bare earth extraction, generating Digital Surface Models (DSMs) and generating DEMs from points classified as bare earth (Fernandez et. al., 2007).
Generating DEM is one of the most important issues in various fields of science, especially the flood modeling using surfaces and morphological modeling as shown in Fig 9 (Weed, 2000; Rayburg et. al., 2009). During the last decade, several new methods for generating DEMs with high 3D positional accuracy have appeared.
Recent research conducted on ALS has shown that it is one of the best solutions to generate DEMs for greater swath sizes, up to 3/4km, at higher elevations in short time (Korzeniowska and Łacka, 2011).
Figure 8 Typical Airborne Laser Scanning scenario. Source: LiDAR Basics at
http://www.dot.state.oh.us/Divisions/Engineering/CaddMapping/RemoteSensingandMapping/Pages/LiDAR-Basics.aspx [Accessed: 15th Aug 2016]
16
Figure 9 LiDAR DSM (left) and gray-scale intensity image (right). Images show first return LiDAR data of Baltimore, Maryland. SOURCE: Fowler et al., 2007
2.4.3. GIS Hydrographical Tools: ArcHydro, HEC-RAS
Hydrological flood modeling and flood data management tools are now readily available to assist domain experts in analysis, modeling and management of flood modeling data and studies with greater decision support capabilities and accuracy. Tools available, such as ArcHydro integrated with ArcGIS platform, provide easy access to algorithms and workflows needed to process surface elevation data in order to produce necessary features for flow direction and accumulation of water in the relevant study area.
ArcHydro is a geospatial and temporal data model for water resources management and analysis, a tool jointly developed by ESRI and Center for Research in Water Resources (CRWR) of University of Texas, a collaboration called GIS and Water Resources Consortium (Maidment, D. R. 2002), involving representatives from industry, government, and
17
academia. The purpose of ArcHydro is enabling the analysis to define, process surface water hydrology and hydrography. It is not full-fledged analysis and simulation tool in itself, but it integrates other research and algorithm in flood modeling, simulation to exchange, process and manage the data in a way to achieve the desired results. ArcHydro is used in extracting characteristics of the watershed, such as stream network and catchment. Delineation is essential for hydrological analysis and water resource management in GIS (Zhang et al. 2013). The foundation of these hydrologic models lies in how to obtain hydrologic and topographic parameters, i.e. watershed characteristics, from Digital Elevation Models (DEMs) (Ames et al. 2009; Jenson 1991; Lacroix et al. 2002).
Another popular and widely used tool for flood extent and depth assessment is HEC-RAS from US Army Corps of Engineers. It is a Hydrologic Modeling System designed to describe the physical properties of streams and rivers as well as to route flows through them. Given the discharge computed by HEC-HMS or by other means, HEC-RAS computes the resulting water surface elevation. Using another program called HEC-GeoRAS, the elevations can then be mapped in ArcGIS to form a flood inundation map (Djokic and Maidment, 2012).
HEC-RAS is a 1-D model that can simulate both steady and unsteady flow conditions (A.
Md Ali et al, 2015)
2.5. Analytical Hierarchical Process
The Analytic Hierarchy Process (AHP), developed by Saaty (1977, 1980), is a widely used multi-criteria decision support method and tool that is considered most popular in many fields, including natural resource management (Mendoza and Sprouse, 1989, Murray and Gadow 1991, Kangas 1992, Rauscher et al., 2000, Reynolds, 2001, Vacik and Lexer, 2001
18
and Ananda and Herath, 2003). AHP is a mathematical theory of value, reason and judgment based on ratio scales for the analysis of multiple-criteria decision-making problems (Saaty, 2001). An example is indicated in Fig 10.
Figure 10 Continuous Rating Scale for Pairwise Comparison of Saaty’s Method
19
CHAPTER III Methodology
This chapter deals with methodology and processes, including data collection, pre- processing and data manipulation as well as modeling tools applied to flash flooding analysis and mapping of the study area. In order to accomplish the objective of this study, the integration of Remote Sensing and GIS in modeling flash flood impacts in the study area was established depending on datasets available from newer technologies such as LiDAR captured from Aerial sensor platform. Several stages of this study procedure were carried out as follows.
The initial stage involved the production of high-quality DEM, slope, and aspects related to LiDAR data of the target that has been surveyed recently. This stage also involved DEM processing and cleanup. Next, DEM manipulation processes performed DEM leveling using AGREE model in ArcHydro by using stream orders. GIS dataset available from a previous study was used to produce the water streams flow in the region of the rain.
In addition, tasks such as wall building and burning linear features in the DEM carried greater GIS processing to generate flow networks, catchment areas, and sink areas in order to support the modeling and assessment of the impacts on the urban area in higher accuracy due to the use of latest airborne LiDAR data. As a comparative study carried out using DEM derived from ASTER sensor etc., lower resolution data sources have also been used and processed side by side to assess the accuracy and details of the results in relation to the LiDAR-derived datasets output. This study is needed to be similar to (A. Md Ali et
20
al, 2015) to determine whether the quality and accuracy of the DEM is more important than the resolution of the DEM.
Final stage involved 3D visualization of the flooded areas using ArcGIS and ERDAS software as well as the production of flood map of the area. In 3D environment that closely mimics the real environment, users can view the impacts of storm surge flood on the landscape (Salman Yussof et al. 2015). 3D view does not only register a clear-to-nature scenario but also provides a more discerning outlook of the buildings and infrastructure during floods.
Finally, in this research, we have clearly shown that GIS and LiDAR technologies combined with hydrological modeling can significantly improve the decision-making and visualization of flood impacts needed for early emergency planning and flood rescue.
Following three Diagrams illustrate the higher-level approach and processing workflow tasks that were carried out for this study using both Aerial and Satellite elevation data:
Figure 11 Workflow for pre-processing LiDAR derived DEM and generating Floodwater flow direction and inundation and Final Flood Map generation.
21
Figure 12 Workflow for pre-processing ASTER satellite derived DEM and generating Floodwater flow direction and inundation and comparison to Aerial LiDAR
Figure 13 Integration of Satellite, Aerial Imagery, LiDAR/DEM, Flood Maps and Floodwater layers in a single 3-Dimensional view for better assessment.
22
3.1. Data Collection and Preparation
The Aerial Imagery, LiDAR, and Satellite data as well as study area vector GIS data were primarily acquired from ADA produced from Aerial Survey Project in 2011. ASTER data DEM was acquired from NASA Earth Explorer website. ERDAS software licenses were provided by Geosystems Company; regional ERDAS product distributors.
The Acquired Aerial LiDAR data was available in 2 separate multiple datasets of Ground and Non-Ground Data. The datasets were merged and mosaicked, and DEM was produced to retain the non-ground information in the resulting DEM, such as building footprints detailed elevations, trees, objects etc. as they all affect water flow and can result flood mapping with greater details and realistic visualizations as shown in Fig 14.
Figure 14 Aerial LiDAR data showing elevation in colored intensity.
Slope raster image was also produced via ERDAS software illustrating the slope of angle in percentage of change from 0 to 200 percent as shown in Fig 15.
23
Figure 15 Derived Slopes from DEM
Some GIS shape files provided by ADA having information as per their previous data for overall Riyadh city including the study area polygon and some relevant GIS information such as roads/streets, parcels, water drain manholes data shape files and geological Formation as shown in Fig 16 and 17
Figure 16 Study Area GIS Data - Source: ADA
24
Figure 17 Geological Formation for Riyadh City, Source: ADA
25
3.2. Data Modeling and Hydrological Processing
This section illustrates the work process that has been carried out in order to process the LiDAR DEM and Satellite DEM data to identify areas at risk of water accumulation during flooding time using both high resolution and low resolution Elevation Datasets output of the study area as per the research methods, which is later used as comparison and potentially affected area review, mapping of the risk areas in GIS maps as well as 3D visualization of the same.
3.3. Aerial LiDAR Processing and DEM/DTM Generation
The first process involved the generation of DEM/DTM from acquired LiDAR data, which consisted of around 200 datasets/tiles of LiDAR data files covering ground/non-ground classification LiDAR data point clouds of 1000m by 750m area per file. Overall, there were 44,718,859 points in source data and a final of 5,644,428 LiDAR data points from 8.12 Sq.km covered area in the study area. The DTM generation process from LiDAR data involves following steps:
- Mosaicking of LiDAR datasets files
- Selection and extraction of area of interest from Mosaicked DTM - Verification and correction of missing areas.
LiDAR datasets were selected for the study area in LAS version 1.0 format/extension. The LiDAR files for both ground and non-ground points were merged and mosaicked together.
26
Afterwards, ground elevation points of LiDAR output were extracted from the mosaic dataset for the study area of interest as shown in Fig 18 and 19.
Figure 18 Aerial LiDAR point cloud datasets merged and filtered with ground cover classification and export to new-mosaicked LiDAR dataset.
Figure 19: 3D view of the LiDAR Elevation Data using ERDAS
27
Once the ground cover LiDAR data was extracted from the mosaic, and then ERDAS Terrain preparation tool was used to generate the surface model DTM. Based on the source LiDAR data, ERDAS was able to achieve 1.3m DTM resolution from the aerial data as illustrated in Fig 20.
Figure 20 Ground cover LiDAR data converted to surface model.
The generated surface elevation model was then subset into the study area dimension using the available study area polygon from ADA. Fig 21 highlights the generated elevation model as shaded relief model.
28
Figure 21: DTM Subset to Study area viewed as shaded relief model.
3.4. Aerial LiDAR Issues and Fixes
In such study, many easy and complicated issues are expected with LiDAR during the preparation and processing. Below, the major issues and the fixes undertaken are I listed.
3.4.1. LiDAR Missing Ground Elevation Data
Up to 200 Aerial LiDAR (.las) files were processed and merged. The sub-setting was then run into flow accumulation and flow direction process and stream delineation process, which identified some areas with incorrect or missing information as shown in Fig 22. After further investigation, it was found that the area had missing ground elevation data for certain datasets or there was only low elevation. However, building classes were present in those areas. This was done by hiding all other classes and focusing on a ground class to identify the issue in the specific region. The troubling LiDAR datasets were replaced with 5m DTM acquired from an older source created via high-resolution stereo-paired Ortho-
29
satellite imagery from older project by another agency working with ADA to compensate the loss, since correct data could not be acquired due to ADA busy schedules.
Figure 22 Problem Areas visible as grey and light brown colored tiles showing some banding.
3.4.2. LiDAR Data Missing Projection Definition
Aerial LiDAR point cloud datasets were merged and filtered with ground cover classification and exported to new-mosaicked LiDAR dataset, but we could not perform further process due to missing Projection/Coordinate System. Since no projection definition was defined in individual LiDAR datasets, hence it could not be accurately and properly visualized along with reference satellite imagery for the same study area as shown in Fig 23.
30
Figure 23 Study Area Geoeye-1 Satellite Image: Source: ADA
No LiDAR data acquisition report was provided. Due to internal restriction, the projection definition was applied based on guesstimates since no projection definition was available in the LiDAR files headers. Firstly, the Geographic projection was defined for each individual LiDAR dataset and reprocessed as before, but it was completely off and did not overlay on top of Ortho-satellite imagery of year 2015, so could not be located on the map.
When zoomed in alongside the satellite, the footprints were corrupted, but if satellite image was zoomed out, only LiDAR was visible as shown in Fig 24. It means that the projection was not correct.
31
Figure 24 LiDAR view in viewer due to missing projection definition
The second projection selected was most common of the vector datasets for Riyadh city i.e. Ain Al Abd, but compared to the true Ortho Image, still the LiDAR had shift.
The third coordinate system used for the definition was UTM Zone 38 N, which covers the international UTM Zone projection containing Riyadh region area of KSA. After definition of individual 200+ datasets in batch, ERDAS process for definition of UTM Zone 38 N projection EPSG Code: 32638 (EPSG International projection code) is shown in Fig 25.
32
Figure 25 Defining Projection
After reprocessing and sub setting the LiDAR data over the Ortho Imagery, the correct projection was identified and defined.
3.4.3. Terrain Data Processing from LiDAR DEM
Once the DEM was generated from Aerial LiDAR survey data, the next step involved the pre-processing or reconditioning of DEM in order to be used for further processing.
The next processing has been performed using ArcHydro Toolset extension for ArcGIS in order to carry out Terrain Data processing to achieve DEM Leveling, water flow accumulation, path and catchment and drainage area identification for the study area.
3.4.3.1. DEM Leveling and Pre-processing
DEM Leveling Process, Filling DEM having holes and small uneven artificial areas were caused due to the nature of DEM generation process. The purpose of pre-processing of DEM was to reduce errors and fill null and gap areas so as to reduce errors that may propagate further and damage the further analysis, processing and results, as DEM is the key dataset flood modeling as shown in Fig. 26.
33
Figure 26 Leveled and Filled DEM
3.4.3.2. Flow Direction, Accumulation and Stream Generation
The next step in pre-processing involved the generation of flow direction grid dataset. The values in the cells of the flow direction grid indicate the direction of the steepest descent from that cell as shown in Fig 27.
Figure 27 Flow Direction generation using Terrain preparing Tools from ArcHydro using filled DEM
34
Once the flow direction grid is generated, the next step was to generate flow accumulation grid. This function computes the flow accumulation grid that contains the accumulated number of cells upstream for each cell in the input grid. Then using the flow accumulation drainage as stream, delineation was carried out to determine the water stream grid identifying the path as illustrated in Fig 28.
Figure 28 Flow Accumulation generation from Terrain Processing using Flow direction map data, stream and their linkages segmentation.
Process: Zoomed In View
35
3.4.3.3. Catchment Area Identification
Catchment area delineation and drainage line generation using stream flow and flow direction data processing were carried out next in order to generate the catchment area polygon for the study area, where the overall catchment area was about 8.12 km2. Next step was the creation of adjoint catchment areas by merging/aggregating some catchment areas to enable speedier point delineation processing later as shown in Fig 29.
Figure 29 Streamlines overlayed on Catchment Areas.
Using the flow accumulation and catchment areas enabled the generation of drainage area points associated with them as the below processing output indicates in Fig 30.
36
Figure 30 Adjoint Catchment Areas and Drainage Points for the Process Output
The catchment and drainage point processing extracted 59 areas and their drainage point was determined by the largest value from the flow accumulation grid produced earlier.
3.4.3.4. Longest Water Flow Path Generation:
Longest flow path generation for catchments and adjoint catchment areas will enable faster processing of longest flow path generation for the watershed process as the process output is highlighted in Fig. 31.
Figure 31 Longest Flow path
37
3D drainage area boundary line generation using the Terrain Morphology processing was carried out and visualized for the study area using ArcScene. As part of Terrain Morphology, various functions were carried out, such as drainage area characterization, boundary definition, and drainage connectivity. The purpose was to perform initial analysis on non- centric terrain to prepare it for further processing during the watershed process.
The process resulted in the generation of longest water flow path having 11.18 km of length with 28 branches connecting along the path. The average elevation of branches was calculated using the Surface Information tool of 3D Spatial Analyst in ArcGIS and was determined excluding the longest path to be a Z elevation mean of 671.92 m and the longest having a Z mean of 643.9m.
3.4.3.5. Hydro Network Generation:
Hydro network generation was one of the key aims of processing Aerial LiDAR datasets in order to generate hydrological water network, determine the water flow direction for the streams and identify areas where there is a risk for potential water accumulation in case of rains and flash floods. The Geometric Network dataset was created by carrying Drainage Connectivity Characterization in the Terrain Morphology toolset of ArcHydro.
This process results in the creation of HydroJunction points that identify the points or areas of water in the study area that are at risk of water storage.
In addition, a flow path of water was created from all directions or surface elevations moving towards the lower elevation areas as illustrated in process output in Fig 32.
38
Figure 32 HydroJunction and HydroEdge Generation
The next step is processing the Geometric Network dataset using the new HydroJunction and HydroEdge feature classes. It resulted in the generation of water flow network dataset generation, which has been used to visualize the direction of water flow in the study area as shown in Fig 33.
Figure 33 HydroNetwork and Longest WaterFlow Path Process Output
39
3.4.3.6. Watershed Delineation:
Additionally, in the end, watershed delineation was carried out based on the selected points of drainage area and hydro network junction marked as accumulating water, also using catchment area, flow direction and generated stream grid information from earlier processing resulting in watershed areas for the points and using this to process the longest water flow paths. Eventually, one main flow path was generated and determined from the highest to the lowest elevation of the DEM as shown in Fig 34 and 35.
Figure 34: Watershed points and generated Watersheds of area.
Figure 35: Water flow path per watersheds and Selected Water Flow across watershed for area.
40
The process resulted in the generation of longest water flow path having 11.18 km of length with 28 branches connecting along the path. The average elevation of branches was calculated using the Surface Information tool of 3D Spatial Analyst in ArcGIS and was determined excluding the longest path to be a Z elevation mean of 671.92 m and the longest having a Z mean of 643.9m.
3.4.4. Terrain Data Processing and Modeling using ASTER DEM
The second part involved downloading ASTER DEM from NASA’s Earth Explorer website for the study area in order to process the Digital Elevation Model and perform flood model processing to identify flood water accumulation risk areas so as to compare with LiDAR based High resolution DEM for determining the accuracy and quality gains achieved for the dense urban areas using both high and low resolution datasets. This is done in order to prove the perceived benefits and gains for using LiDAR derived DEM in urban areas for flood modeling analysis needed for Hydrologists and decision makers.
Similar pre-processing or reconditioning of DEM was carried out for the free available 30m ASTER DEM to be usable for further processing as was the case with LiDAR. After extraction, subsetting the DEM in ERDAS was done in ArcHydro Toolset extension for ArcGIS in order to carry out Terrain Data processing to achieve DEM Leveling, water flow accumulation, path and catchment and drainage area identification for the study area. Unlike initial LiDAR data corruption issues with some parts, no such issues were encountered with ASTER DEM.
41
3.4.4.1. DEM Leveling and Pre-processing:
DEM Leveling process with sink of 5m in elevation was performed to fill any potential sinks of 5m or less as shown in Fig 36, so that such a small amount does not cause unnecessarily processing delays, as that depth may be considered too insignificant to have effect; if not removed may yield unrealistic and impractical results.
Figure 36 Satellite: Leveled and Filled DEM
3.4.4.2. Flow Direction, Accumulation and Stream Generation
In the next step, similar to before where we used Lidar-based elevation model, this step is to involve flow direction, flow accumulation and stream grid generation performed as shown in Fig 37, 38, and 39.
Figure 37 Flow Direction
42
Figure 38 Flow Accumulation
Figure 39 Water Streams
43
3.4.4.3. Catchment Area Identification
Catchment area delineation and drainage line generation using stream flow and flow direction data processing was carried out next in order to generate the catchment area polygon for the study area as highlighted in Fig 40. The catchment areas extracted 56 areas with total drainage area of 6.92 Km2.
Figure 40 Catchment Areas Grid, Polygons and Drainage Lines
Using flow accumulation and catchment areas enabled the identification of drainage areas associated with them as shown in Fig 41. The catchment and drainage point processing extracted 56 areas and their drainage point is determined by the largest value from the flow accumulation grid produced earlier.
44
Figure 41 Drainage Points for after processing
3.4.4.4. Hydro Network Generation
The Hydro Network generation was also conducted for ASTER 30m Satellite DEM dataset in order to generate hydrological water network, determine the water flow direction for the streams and identify areas where there is a risk of potential water accumulation in case of rains and flash floods. The Geometric Network dataset was created by carrying Drainage Connectivity Characterization in the Terrain Morphology toolset of ArcHydro as shown in Fig 42.
Figure 42 HydroJunction and HydroEdge Generation
45
This process results in the creation of HydroJunction points that identify points/areas of water in the study area that are at risk of water storage. Also, a flow path of water was created from all directions or surface elevations moving towards the lower elevation areas as shown in Fig 43.
Figure 43 HydroJunction and HydroEdge (Arrows indicating water flow edge directions)
The next step for the creation of new Geometric Network dataset was using the HydroJunction and HydroEdge feature classes. It resulted in water flow network dataset generation which was used to visualize the direction of water flow in the study area.
3.4.4.5. Watershed Delineation:
Additionally, in the end, watershed delineation was conducted based on the selected points of drainage area, hydro network junction identified as accumulating water, using catchment area, flow direction, and generated stream grid information from earlier processing resulting in watershed areas for the points as shown in Fig 44
46
Figure 44 Watershed points and their associated watershed polygons
The Watershed delineation process resulted in the identification of 26 areas with associated watershed points towards them the individual water flow streams led. This led to the creation of identification of longest flow path of water with length of 7.74km with average mean Z of 654.29 meters. Overall, other areas’ average z was 679.55 meters. The combined watershed area using ASTER DEM was 6.96 Sq km, which was less than LIDAR DEM calculated area.
47
CHAPTER IV
RESULTS AND DISCUSSION
This chapter illustrates various outputs, mainly cartographic, generated through data manipulation and processing from both tools: LiDAR based data as well as Satellite information-based processing. Accordingly, one of the main outcomes of this study was the production of several maps that indicate some aspects of flash flood risk in the study area through visualized hydro networks and 3D visualization.
4.1.
Flood Risk Potential Maps
The study produced several maps regarding catchment area, longest path of water flow, and risk points of water accumulation; mainly derived using LiDAR and ASTER 30m DEM data. For the sake of comparison, this section presents the results of the maps developed from both LiDAR and ASTER DEM elevation data processing carried out as demonstrated in the previous section. These maps illustrate the modeled direction of water flow, identified catchments and longest path of water flow, and risk points of greater water accumulation in the study area, which are derived mainly from identified drainage points in the catchments of the study area.
Firstly, the water stream flow direction map illustrates the flow of water streams generated via DEM processing as well as modeled direction which extends from high elevation of 718 meters to low elevation areas of 580 meters. Color classification highlights the difference in ground DEM elevation in the Hydro DEM Layers as indicated in Fig 45, 46, and 47.
48
Figure 45 Water Stream Flow Direction Map (LiDAR)
Figure 46 Zoomed View 1: Water Stream Flow Direction Focused Map
49
Figure 47 Zoomed View 2: Water Stream Flow Direction Focused Map
Catchment areas identified by ArcHydro were 59 areas classified as critical to the further watershed, drainage, and hydro network processing as illustrated in Fig 48 and 49.
Figure 48: LiDAR: Catchment Areas Map
50
Figure 49: Satellite: Catchment Areas Map
Fig 50 and 51 illustrate the longest flow path of water across the study area generated from catchment area and water flow direction processing and identify the long single water stream flow from highest to lowest elevation in the area with a length of above 10 KM for the LiDAR and 7.7 KM for ASTER.
Figure 50: LiDAR: Longest Water Flow Path Map
51
Figure 51: Satellite: Longest Flow Path Map
The map data in Fig 52 illustrates the points overlaid on top of water manholes’
information along with longest water flow as well satellite Image of the study area.
Figure 52: LiDAR: Water Accumulation/Risk Areas Map
52
Figure 53 illustrates the water accumulation risk areas identified as 26 potential areas, where water may be collected due to lower depressed elevation in the points identified in the Hydro Junction processing.
Figure 53: Satellite: Water Accumulation Risk Areas Map
The map illustrates the modeled direction of water flow in the study area as well as identified catchments and longest path of water flow besides areas that are at risk of greater water accumulation mainly derived from identified drainage points in the catchments of the study area all similar to LiDAR based data processing, using the same tools and processing techniques in ERDAS, ArcGIS and ArcHydro etc. Fig 51 illustrates the generated water streams in the drainage area and directions of water flow.
53
Figure 51: Satellite: Water Drainage Streams and Flow Direction Map
Finally, main water accumulation risk areas map identifies the potential areas to focus on water accumulation due to flash floods, derived from drainage areas’ calculation during catchment area, water flow direction, water streams and elevation information crossing of different catchments. Total selected points were 59 out of 236 points in the generated Hydro Junctions focused on the potential risk sink area points. The map data in Fig 50 illustrates the points overlaid on top of water manholes’ information along with longest water flow as well as satellite image of the study area. Figure 53 illustrates the water accumulation risk areas identified as 26 potential areas where water may be collected due to lower depressed elevation in the points identified in the Hydro Junction processing.
4.2. Summary of the results
The below table summarizes and compares the results of the path length, number of risky points, and the average gradient
Path length km
Risk points
# Average Gradient
LiDAR 11.18 56 3.97
ASTER 7.7 26 5.77
54
4.3. Simulation Models of Flood Risks
Heavy rainfall events may generate flashfloods and natural disasters and cause severe damages to lives and properties. The average annual rainfall of about 111 mm in Riyadh city was estimated from the available rainfall data from 1982 to 2011. However, historical observations of extreme rainfall on a daily basis have been witnessed in central Arabia.
Moreover, extreme rainfall events have been documented in arid regions, where daily rainfall exceeded the total annual rainfall.
However, three scenarios of flood risks in the study area were established. First scenario was based on a rainfall of 30 mm as assumed figure. The second scenario was built on the highest monthly rainfall of 70 mm recorded in 2016 for the month of November in Riyadh (KAPSARC, 2016). The third scenario indicates the case of extreme daily rainfall that may exceed the annual rainfall of 110mm (assumed figure).
The below table summarizes and compares the three Scenarios in term of “What Happened” and the “Rise of Water” as well.
Scenario 1 Scenario 2 Scenario 3
What happened? More Ground is Visible Surface visibility reduces Higher amount of surface is covered with water
Rise of water? Low water level Rise in water Rise in Water more than 70mm
The 3D view was built via simulating water level in the study area along with LiDAR DEM dataset and overlaying along with satellite imagery as well as water streams and accumulation risk points and municipal drainage water manholes. The other 2 views are the views of the same area with zoomed inset views in greater imagery detail of the simulated effects if the water level rises in the area. The red points are potential water accumulation areas, and yellow points are water drain holes in the area.
In order to process data and visualize in 3D, both ERDAS Imagine and ESRI’s ArcScene products were used. The ERDAS was used to prepare Ground DEM as shown in Fig 52
55
Figure 52: Ground Elevations (Blue to RED – High to Low Elevation)
Non-Ground Elevation Model highlighted in Fig 53 visualizes features above ground, a water layer as raster with elevation from ground DEM and satellite imagery drape with Non-Ground elevation model.
Figure 53: Non-Ground Elevation (Blue to RED – High to Low Elevation)
56
Figure 54 highlights all the processed and used datasets. Draping Non-ground features on the Surface DEM can influence the flow of water in real life scenarios, highly affect the flow of water in urban areas by having obstruction to water flow direction and may even result in higher water accumulation in dense urban environments.
Figure 54: Water Layer (Draped on Top of Ground with 70mm extrusion)
The 3D visualizations illustrated in Fig 55, 56, 57, and 58 show the simulated view of 70mm water level rise in the study area and how it can potentially impact and affect the urban area infrastructure and the regions.
57
Figure 55: Water Level Simulated 3D Map (Red: Accumulation Points, Yellow: Manholes)
58
Figure 56 Satellite Imagery drape with Non Ground Elevation
Figure 57 Zoomed in 3D Map View 1 Map (Red: Accumulation Points, Yellow: Manholes)
59
Figure 58 Zoomed in 3D Map View 2 Map (Red: Accumulation Points, Yellow: Manholes)
In addition to the available average rainfall data of 70mm, 2 further scenarios sample rainfall averages, 30mm (Fig. 59) and 110mm (Fig. 60), were used and applied to the 3D process to visualize how affected area would look if there is much heavier rainfall or less than previous reported figure.
60
Figure 59: 3D Visualization with simulated 30mm rainfall data.
Figure 60: 3D Visualization with simulated 110mm rainfall data.
