Flood Mapping: Assessing the Uncertainty Associated with Flood Inundation Modelling. A Case Study of the Mora River, Sweden

(1)

Examensarbete vid Institutionen för geovetenskaper

Degree Project at the Department of Earth Sciences

ISSN 1650-6553 Nr 390

Flood Mapping: Assessing the

Uncertainty Associated with

Flood Inundation Modelling.

A Case Study of the Mora

River, Sweden

Översvämningskartering: Bedömning av

osäkerheter relaterat till modellering

av översvämningar. En fallstudie av

Moraån, Sverige

Isabelle Åberg

INSTITUTIONEN FÖR GEOVETENSKAPER D E P A R T M E N T O F E A R T H S C I E N C E S

(2)

(3)

Examensarbete vid Institutionen för geovetenskaper

Degree Project at the Department of Earth Sciences

ISSN 1650-6553 Nr 390

Flood Mapping: Assessing the

Uncertainty Associated with

Flood Inundation Modelling.

A Case Study of the Mora

River, Sweden

Översvämningskartering: Bedömning av

osäkerheter relaterat till modellering

av översvämningar. En fallstudie av

Moraån, Sverige

(4)

The work for this thesis was carried in cooperation with Tyréns.

ISSN 1650-6553

(5)

Abstract

Flood Mapping: Assessing the Uncertainty Associated to Flood Inundation Modelling.

A Case Study of the Mora River, Sweden

Isabelle Åberg

Expansion of cities and major infrastructure projects lead to changes in land use and river flows. The probability of flooding is expected to increase in the future as a result of these changes in combination with climate change. Hydraulic models can be used to obtain simulated water levels to investigate the risk of flooding and identify areas that might potentially be flooded due to climate change.

Since a model is a simplification of the reality it is important to be aware of a model’s uncertainties. A part of this study is therefore aimed to perform a sensitivity analysis to determine which parameter has the largest impact on the model result and has to be treated more careful and accurately. In this study the 1-dimensional flow model Hydrologic Engineering Center-River Analysis System (HEC-RAS) were assed to simulate predicted water levels within the studied river. Topographic data was used to draw cross sections in Geographic Information Systems (GIS) with additional tools of HEC-GeoRAS, in order to get information about the streams geometry. The purpose of doing a sensitivity analysis was attained by investigating changes of the model results when changing different input parameters. This work is based on a reach along Mora river, in Södertälje, Sweden, as a case study. The sensitivity analysis indicate that the number of cross sections has a significant effect when simulating water levels of low flows and that the absolute error of simulated water levels increases as the average spacing between cross sections increases. The second part of the study aims to examine the effects of climate change and how it will affect water levels for the studied river. The results of the study showed that simulated water levels with flows of 100, 200 and 500 years return periods stay within the river channel and do not indicate flooded areas. The results also showed that a backwater effect due to sea level rise would affect the water levels in the stream up to a specific critical point along the studied reach.

The lower reach indicated results to contain more uncertain region, where floodplain delineation changed easily as the number of cross section was changed. It is therefore important to identify the areas where uncertainties can be more critical for the results. Because of the uncertainties associated to the model used, it is important to notice that the results of this work correspond particularly to the case study in Mora River.

Keywords: Flood inundation modeling, hydraulic modeling, HEC-RAS, HEC-GeoRAS, sensitivity

analysis

Degree Project E1 in Earth Science, 1GV025, 30 credits Supervisor: Giuliano Di Baldassarre

Department of Earth Sciences, Uppsala University, Villavägen 16, SE-75236 Uppsala (www.geo.uu.se)

ISSN 1650-6553, Examensarbete vid Institutionen för geovetenskaper, No. 390, 2017 The whole document is available at www.diva-portal.org

(6)

Populärvetenskaplig sammanfattning

Översvämningskartering: Bedömning av osäkerheter relaterat till modellering av översväm-ningar. En fallstudie av Moraån, Sverige

Isabelle Åberg

Studien genomförs inom Ostlänken, som är den första delsträckan av en ny höghastighetsjärnväg mellan Järna och Linköping.

Bebyggelse och stora infrastrukturprojekt kommer förändra markanvändning och flöden i vattendrag. I följd av dessa förändringar, tillsammans med framtidens förändrade klimat, kommer risken för översvämningar kunna öka. För att undersöka riskerna för översvämningar och kartlägga områden som riskerar att översvämmas är en hydraulisk modell ett verktyg som kan användas.

Då en modell endast är en förenkling av verkligheten och påverkas av flera olika parametrar är det viktigt att vara medveten om modellers osäkerheter. För att få modellen att efterlikna verkligheten så bra som möjligt kan det vara bra att veta vilka parametrar som har störst inverkan på modellens resultat och som bör bearbetas mer noggrant. Därmed är en del av studiens syfte att göra en känslighetsanalys för att utreda vilka modellparametrar och indata som påverkar modellresultaten, med fokus på att analysera simulerade vattennivåer. Känslighetsanalysen utförs genom en fallstudie över Moraån, där den endimensionella flödesmodellen HEC-RAS används för att beräkna vattendragets vattennivåer.

Den andra delen av studiens syfte är att undersöka om hur framtidens klimatförändringar kommer kunna påverka det studerade området. En effekt av framtidens förändrade klimat är stigande havsnivåer som leder till ökad risk för översvämning vid kustnära områden. Till följd av dämningseffekter kommer de stigande havsnivåerna även ge ökade vattennivåer uppströms vattendragen, och beroende på vattendragens egenskaper och geometri kommer vattendrag längs med kusten att påverkas på olika sätt. För att undersöka riskerna för översvämningar i ett framtida klimat har modeller med olika klimatscenarios byggts upp där stigande havsnivåer kombinerats med flöden av varierande återkomsttider.

Nyckelord: Översvämningskartering, hydraulisk modellering, HEC-RAS, HEC-GeoRAS,

känslighetsanalys

Examensarbete E1 i geovetenskap, 1GV025, 30 hp Handledare: Giuliano Di Baldassarre

Institutionen för geovetenskaper, Uppsala universitetet, Villavägen 16, 752 36 Uppsala (www.geo.uu.se)

ISSN 1650-6553, Examensarbete vid Institutionen för geovetenskaper, Nr 390, 2017 Hela publikationen finns tillgänglig på www.diva-portal.org

(7)

Table of Contents (continued)

6.5. Change in channel elevation ... 31

6.5.1. How does a change in channel elevation affect the calibration of low flows? ... 31

6.6. Further studies ... 31

7. Conclusions ... 32

8. Acknowledgements ... 33

9. References ... 34

(9)

1

1. Introduction

Natural hazards such as floods can cause social and economic damage. For many countries, floods are a national threat and the results are difficult to prevent (Manfreda et al., 2015). In order to mitigate floods, it is necessary to understand how and where flood events might happen and how they can cause damage. Major infrastructure projects and urban expansion leads to changes in land use, which in turn lead to changes in river flows and a reduction of the infiltration capacity of the soil (Manfreda et al., 2015). As a consequence of these changes, along with expected climate change, it has been emphasized that the risk of flooding is likely to increase in the near future (Genovese, 2006).Studies toward integrated flood risk management are becoming a standard due to increased risk of flooding and impact on human activities.

Modelling can be used for several purposes, such as simulation of floodplain sediment dynamics, flood mitigation, evaluation of flood damages, construction of flood risk maps and analysis of the effects of urban development and new infrastructures (Mambretti, 2012). Given the increased risk of flooding, a number of recent studies have evaluated the effectiveness of models predicting river flood inundation by testing their ability to predict inundation extent for a given flood event.

Flood mapping is a tool that can be used to identify the hazard, risk and potential consequences of a flood event. To understand the behavior of a river system, hydraulic modelling of floods is widely used. Flood inundation maps can give predictions according to floods and future climate scenarios and can be used for commercial decisions and as basis for physical planning (Nordblom & Petzén, 2014). Simulations of flood events performed by hydraulic models make possible to explain the behavior of complex river systems and reproduce results of flood propagation and inundation extent for natural or altered river systems (Castellarin et al., 2009).

A hydraulic model is basically a mathematical representation of surface flow dynamics and can be used to calculate water levels and flood propagation. In order to calculate water levels along a stream, calculations from a hydraulic model are applied to an elevation model to produce a map of the zones that are likely to be flooded in a specific area. Fundamental input data when constructing a hydraulic model is topographic data, which provide information about the channel’s geometry and description of the floodplains from a digital elevation model (DEM), derived from laser scanned data (LIDAR). DEMs can be derived with various resolutions which in turn will affect the output of flood modelling (Md Ali et al., 2015). Other information such as physical structures (bridges, dams, culverts, etc.) along the rivers that may affect the natural flow can also be essential information for constructing an accurate flood map (Nordblom & Petzén, 2014). Several assumptions and simplifications have to be done when building a model; it is therefore important to do a sensitivity analysis to explore which parameters affect the model results the most.

(10)

2

The first aim of this study is to investigate how sensitive the model is to different input parameters in terms of simulated water levels. The sensitivity analysis is performed through a case study in Mora River.

Due to climate change and increasing risk of flood events, the second aim of the study is to examine how future climate change can influence the study area. Rise in sea level is one of the effects of climate change and becomes a relevant question because of impacts and damage from increased risk of flooding events in coastal areas (Walsh & Miskewitz, 2013). As a result of rise in sea level, backwater effects can provide increased water levels upstream, and depending on the river properties and geometry of the cross sections upstream, the stream can be affected in different ways.

The study was performed using the unidimensional (1D) model of the Hydrologic Engineering Center’s River Analysis System (HEC-RAS) together with Geographical Information Systems (GIS) software to calculate water levels in the river.

This study is carried out within the East Link project, a part of Sweden’s high-speed railway between Järna and Linköping together with the consulting firm Tyréns. The case study of this work, in the Mora River is one of the streams along the planned railway. The East Link project has an estimated life time over 100 years and it is therefore important that the railway is built for conditions that may exist throughout this period. Climate change is therefore an important aspect to take into account in this context.

2. Aim

The main objective of this thesis is to contribute to a better simulation of floods and assess the uncertainty associated to flood inundation modeling, by performing a sensitivity analysis of different model parameters. Parameters that are evaluated within this thesis are: sets of Manning’s coefficients (n-value), number of cross sections used to describe the geometry of the river system and boundary conditions. To investigate the risk of flooding in a future climate, models with different climate scenarios are also built by using scenarios of sea level rise as a downstream boundary condition combined with flows of varying return periods.

More specifically, the thesis aims will be attained by: 1) investigating changes of the model results when changing different parameters, 2) assessing how the studied area can be affected by climate change, by testing different climate scenarios. The results from the numerical simulations within the sensitivity analysis represent the differences in modeling approaches and were not intended to calibrate the model or validate the results. The results from the study can be used to produce flood maps that can be used for future planning.

(11)

3

3. Background

Climate change will affect Sweden in several ways, such as rising temperatures and increased precipitation. Expected consequences from this change are increased risk of flooding, landslides and erosion. A tendency is that Sweden might have a wetter climate, where we will have effects of rising sea levels and increased discharges in rivers and streams (Westlin et al., 2012).

3.1. Previous research

Several studies have focused on the topic of flood inundation modelling. Floodplain modelling could be improved by having a better understanding of the actual hydraulic behavior of the river system. These models are essential tools where development of flood risk maps can raise the awareness of decision-makers and people living in flood-prone areas. Even though flood risk maps are essential tools for several reasons, one needs to take consideration of the uncertainties associated to various parameters. There are several models available for predicting flood inundation. There are many aspects to take into consideration when choosing the most appropriate model for a particular study case, such as: available data, accuracy of topographic data, calibration data and reliability of boundary conditions (Castellarin et al., 2011). Topographic data has an influence on 1D modelling of floods and is one of the most fundamental input data for describing the geometry of the river and the characteristics of the floodplain (Md Ali et al., 2015). Nowadays accurate high resolution data from LIDAR is available to produce accurate topographic data, but limitations of uncertainties connected to flood mapping is still an issue that need to be investigated (Weichel et al., 2007). For 1D models, the stream’s geometry is represented by a series of cross sections. It is therefore important to identify the optimal number and locations of drawing cross sections to get an optimal topographic survey to use when developing the geometry for 1D modelling. To get the most accurate representation of the studied channel and flood plain, every effort should be done to get the most appropriate configuration of cross sections (Brunner, 2010a). A few guidelines are available for the selection of the most suitable distance between cross sections for 1D modelling, depending on the hydraulic river system in question (Castellarin et al., 2009). For example, the equation ∆𝑥𝑥 = 𝑘𝑘𝑘𝑘 is used to determine the cross section spacing, where ∆x is the recommended distance between cross sections, B is the bankfull width of the main channel and k is a constant that ranges between 10 and 20. Other equations can be used to represent for more complex unsteady conditions (Castellarin et al., 2009).

3.1.1. Hydraulic modeling and flood mapping

To perform a flood study for a specific catchment of a river system, both hydrologic and hydraulic investigations are needed and the reliability of the results depend on the quality of data (Queensland Floods Commission of Inquiry, 2012). A hydrological model can be defined as an estimation of runoff from a catchment area based on equations using several parameters that describe the watershed

(12)

4

characteristics. Fundamental information for a hydrological model is rainfall data and catchment area (Devia et al., 2015).

Flood risk is commonly represented using maps for events of different return periods (Di Baldassarre et al., 2010). Hydraulic models require an accurate description of the geometry of a stream based on topographic data to provide useful information about the behavior of a river system and to produce maps that show the potential inundated area. A hydraulic model will provide information about water depth, velocity and range of inundation along a watercourse for producing flood inundation maps (Queensland Floods Commission of Inquiry, 2012). Common 1D models, such as MIKE11 and HEC-RAS, solve St. Venant equations and use cross-sections based on topographic data to describe the river channel and floodplain geometries.

These kind of models have been widely used for simulation of river flood inundation due to reduced data requirements and costs compared to 2D models (Popescu, 2014). Additionally, 2D models are in overall more time consuming and more expensive. Previous studies show that 1D models are often able to be as accurate as 2D models (Horrit & Bates, 2002).

3.2. HEC-RAS

HEC-RAS is a 1D modelling program, developed by the U.S Army Corps of Engineers USA. The program is based on 1D hydraulic calculations with both steady and unsteady flow computed with a hydrograph for both natural and constructed channels (Brunner, 2010b). As described above, HEC-RAS solves the energy equations known as 1D St Venant equation (Equation 1) (Horrit & Bates, 2002).:

∂Q ∂x+b ∂h ∂t=q, (1) where Q= discharge

b= storage width at the water surface level (m) h= water depth (m)

q= lateral discharge x= distance (space step) t= time

(Popescu, 2014)

The energy losses for the model are evaluated by calculating friction losses for the channel with Manning’s equation over two cross sections. The program calculates the water level from a cross section

(13)

5

to another one by solving the energy equation (Equation 2) and is illustrated in figure 1. (Brunner, 2010a).

Z₂+Y₂+α2V22

2g =Z1+Y1+

α1V12

2g +he (2)

Z1,Z2= represents the elevation of the main channel inverts

Y1,Y2= depth of water at cross sections

V2

2g= represents the velocity head at a point

g= gravitational acceleration

h_e= represents the head loss over two cross sections

Figure 1. Figure representing the energy equation used in HEC-RAS (Brunner, 2010a)

Within this study a steady flow analysis was used, in which the model assumes a constant flow until another flow value is encountered. The number of calculated profiles can be determined by the modeler in order to do different flow scenarios that should be calculated.

A boundary condition is needed for the program to do calculations based on a starting water surface elevation from upstream to downstream. Known water surface, critical depth, normal depth and rating curve are four optional boundary conditions available in HEC-RAS. The different boundary conditions will be explained further on in the report.

HEC-RAS divides the flow for the main channel and overbanks based on Manning’s roughness coefficient to determine the conveyance and velocity coefficient for a cross section. To divide the flow

(14)

6

in a cross section different n-values represents the main channel and the overbanks. To calculate the conveyance of the flow, following equations are used in the model:

Q=KS_f 1 2 ₍₃₎ K=1.486 n AR 2/3 (4)

Where K is the conveyance (measure of the carrying capacity of the channel) for the subdivision of each cross section, n is Manning’s roughness coefficient for subdivision, A is the flow area for subdivision and R the hydraulic radius for subdivision which is determined with the area divided by the wetted perimeter. The total conveyance of a cross section is the sum of the subdivisions of each cross section, which are main channel, and left and right over bank (Brunner, 2010a). The main stages to develop a model geometry as an import file to HEC-RAS is represented in the methodology (section 4).

3.3. Climate scenarios

A return period estimates the probability of a flow to occur and is usually based on the statistical analysis of historical data (Nordblom & Petzén, 2014). A return period of a given flow is the average time between two different flows of the specific size or larger. For example, a flow with a return period of 100 years has a probability of occurrence of 1/100 for any given year and a 63% probability for the event to occur during a 100 years period (Sverige & Klimat- och sårbarhetsutredningen, 2006). When doing risk assessments for larger infrastructure projects that will stand for a long time, it is therefore important to take into account for the accumulated probability for a flow to occur (Återkomsttider | SMHI, 2015). A rise in sea level is an expected result from climate change and could be a threat to coastal areas and cause increasing risk of floods (Walsh & Miskewitz, 2013). HEC-RAS is used to predict the change in streams behavior due to changed discharge and water elevation to investigate potential flooding.

3.4. Area of Study, Mora River

3.4.1. Case study- East Link project

The East Link project is part of Sweden’s first high-speed railway. The East Link will make it possible to travel with speeds up to 320 km/h and planning to be operational by 2028. The new double track railway will go from Järna in Södertälje municipality to Linköping and is approximately 150 km long (Trafikverket, 2015). The railway will cross several streams along the planned railway; the focus for this study will be the Mora River. Mora River is Södertälje municipality’s largest river and has its drainage in the Baltic Sea. The catchment area is 93 km2_{and consists of approximately 67,85% forest,}

19,32% agricultural land, 6,86% urban and 6% lakes, (SMHI, 2016b). Besides Mora River, the catchment includes two other rivers, Oga river and Kalfors river that flows from the lakes Ogan and Vällingen. Mora river begins where Kalfors river and Oga river flows together (Sjöar och Vattendrag,

(15)

7

2004). The results will be based on a reach of approximately 2 kilometers, where observed data is available. The studied area is delineated after the position of planned railway, figure 2.

Figure 2. Overview map of study area over Mora River (©Lantmäteriet CC BY)

3.4.2. Field observations

The project includes field work in terms of flow measurements and observations of the watercourse structure and geometry to get an overview and understanding of the area. Field observations used for

(16)

8

this study have been done at a point in the middle section along the reach. The photo in figure 3 shows the middle of the studied reach. Flow measurements, water depth and measured cross sections have been used to calibrate the model within the study.

Figure 3. Mid-section of the studied reach where field measurements have been made. The digital elevation

(17)

9

4. Methodology

To investigate the uncertainty in flood mapping, a case study over Mora River has been performed. The main research questions were: How do uncertain data and parameters affect simulated water levels? And how does this in turn affect flood mapping? Information about existing structures along the watercourse was based on information available on Swedish transport administration (Trafikverket) webpage to be aware of existing structures along the studied reach (BaTMan). The focus of this work was to assess how different input parameters can affect the calculated water levels along the studied reach, the model has been simplified and no existing structures was included along the river.

Three main steps to develop a HEC-RAS (HEC-RAS, 2008) model are 1) preprocessing the terrain model over the studied area in ArcGIS (ArcGIS 10.3.1 for Desktop, 2015), 2) development of river features and cross sections in HEC-GeoRAS, 3) simulating water levels in HECR-RAS based on exported data from HEC-GeoRAS of the rivers geometry. The sequence of programs used within the study is described below (Figure 4).

4.1. Data requirements

Required input data to develop a HEC-RAS model are discharge, geometry of the channel based on topographic data (DEM), boundary conditions. The channel discharge is given data produced for Mora River by SMHI and is presented in next section 4.1.1. The used terrain model for the study is a grid with a resolution of 2,5*2,5 m. This grid has been produced by Lantmäteriet based on laser points collected using aerial laser scanning of the terrain and delivered in the regional projection zones. The quality of the elevation data has a vertical standard error of 0,05 m and a horizontal standard error of 0,25 m (Lantmäteriet, 2015).

4.1.1. Flow

The study investigated a set of flows with different return periods produced for Mora River by SMHI, table 1. The calculated return periods were based on a frequency analysis of neighboring stations with similar characteristics due to missing discharge station in Mora River. The location for the calculated flows are presented in figure 5. Simulated water surface profiles for different flows can be found in appendix 1.

ArcGIS

HEC-GeoRAS

HEC-RAS

(18)

10

Figure 5. DEM of the river system Mora River, showing location of measured flow and calculated flow from

SMHI.

Return period Flow m3_/s

MQ 0,6

HQ100 9

HQ200 10

HQ500 11

(19)

11

4.2. Hydraulic model over Mora River

The first step to develop the geometry file for the hydraulic model was creating a Triangular Irregular Network (TIN) surface from the digital elevation model (DEM) of the river system in ArcGIS. The TIN will represent the topography of the floodplain areas and the channel.

4.2.1. Configuration of cross sections

River features such as main channel, banks, flow paths and cross-sections were located along the studied reach by using HEC-GeoRAS to represent the stream’s geometry and its floodplain. The final step was to export the HEC-GeoRAS file to create a geometry model in HEC-RAS. Adjustment of the geometry file could be done in HEC-RAS. A summary of input parameters in HEC-RAS are presented in table 2. The main steps when developing the geometry model are presented in figure 6 and example of how a cross section is defined in HEC-RAS is illustrated in figure 7. Each cross section has a station number which corresponds to the distance from each cross section to the downstream end of the river. The reference model with drawn cross sections can also be found in appendix 2 and 3.

Figure 6. Illustration of three main steps when developing a geometry file. A) ArcGIS TIN surface B) TIN

surface with HEC-GeoRAS features added C) Exported HEC-RAS geometry

Table 2. HEC-RAS model parameters, reference model

Reach length (m) 2000

Number of cross sections 39

Manning’s n-Value (ROB, LOB) 0,1

(20)

12

Figure 7. Example of definition of a cross section in HEC-RAS. The blue lines are simulated water levels for

different flow scenarios.

The cross sections are drawn from left side to right side looking downstream perpendicular to the stream. Two cross sections should not intersect or crossing the main channel multiple times.

4.2.2. Boundary conditions

For the calculations the model needs a boundary condition to define the starting water level at the end of the river. There are several options available for specifying a boundary condition in HEC-RAS. To perform a steady flow simulation, the following boundary conditions are available for HEC-RAS: known water surface, critical depth, normal depth and rating curve. Within this study a boundary condition of known water surface and normal depth has been used to compare the different results and the effect of different boundary conditions. Known water surface specifies the known water level for a specific discharge for a profile. Normal depth is the most commonly used boundary condition for the program and it uses the energy slope and Manning’s equation to approximate the channel slope, where HEC-RAS calculate the water level for each simulated flow in cross section based on the slope of the energy line of the river. The boundary condition used for normal depth was set to an approximation of the channel slope with the equation 𝑑𝑑ℎ

𝑑𝑑𝑑𝑑= (

(4,66−0,59)

1906 ), where h is the difference in height from upstream to

downstream and L is the reach length. The boundary conditions used for known water surface is summarized in table 3. 0 50 100 150 200 250 300 350 4.0 4.5 5.0 5.5 6.0 6.5 7.0

Referensmodell20140413 Plan: Plan 01 2016-05-04 River = Morariver Reach = Centerline

Station (m) E lev at ion ( m ) Legend WS HQ500 WS HQ200 WS HQ100 WS MQ Ground Levee Bank Sta .1 .045 .1

(21)

13

Table 3. Boundary conditions used for known water surface. Different high-water level of three return periods

(100 years, 200 years and 500 years) and the mean water level. Estimated sea water levels for different climate scenarios for 2018, 2100 and 2150 (Rapport klimattest, working material, 2015).

Sea level rise scenarios

Water level

Sea level rise (2018) (cm)

Sea level rise (2100) (cm)

Sea level rise 2150 (cm) MW (mean water) 0 66 146 HW100 (high water) 119 185 265 HW200 123 189 269 HW500 127 193 273

4.2.3. Calibration and validation, low flows

Calibration of the model is based on measured flows and water levels from October and November 2015. The calibration was done through adjustment of Manning’s roughness coefficient (n-value) to minimize the errors between observed and simulated water levels. Calibration within this study also includes a sensitivity analysis according to number of cross sections. The model was calibrated within an accuracy of millimeters at cross section 1140, compared with observed water levels for that section. The accuracy of the calibration was tested through validation to determine the errors of simulated data against measured data. The validation is performed by running the model with a higher flow measured from another occasion with the calibrated n-values.

4.3. HEC-RAS mapping

The results from HEC-RAS were exported to perform a flood inundation map in GIS, showing flood depths and extent, using the Spatial Analysis and 3D Analyst extension tools in GIS. The exported file from HEC-RAS was an SDF-file converted to XML format so the results could be processed in GIS. The feature class to cut lines held the location for the cross sections with attribute data of water surface elevation at each specific cross section for each profile. The result of the inundation mapping comprised a grid with inundation depths and a vector file that represents the floodplain boundary. These results allowed to determine if the cross sections extend far enough to cover the floodplain area (Cameron & Ackerman, 2009). If the result shows deviating areas in GIS, the areas can be studied in HEC-RAS to identify the areas that should probably not be wet by adding a levee (Figure 8).

(22)

14

Figure 8. Illustration of a cross section where a levee should be added to avoid areas that should not be wet.

4.4. Sensitivity analysis

A sensitivity analysis enables to reduce the probability of errors and to get the best approach of a model. A sensitivity analysis was carried out for various parameters and scenarios to investigate the change of simulated water levels. The sensitivity analysis was based on a number of tests in which parameters and model settings in HEC-RAS were varied within reasonable intervals.

(23)

15

5. Results

5.1. Sensitivity analysis

The adjusted parameters within the sensitivity analysis were: boundary conditions, sea level, and number of cross sections. The results were compared to identify which factors have a more significant impact on the model results in terms of simulated water levels. The results of the sensitivity analysis correspond to the case study area in particular. Similar analyses in other areas and scenarios might result in different results.

5.2. Sensitivity analysis of boundary conditions

The aim in this section is to present results of how the average bed slope used as a downstream boundary condition influences water levels in the river. The assumption for the analysis using Normal Depth as a boundary condition was to use the energy slope of the channel. The diagram in figure 9 represents the results when using Normal Depth with three different (but realistic) energy slopes to see how the results in water levels divergealong the profile.

Figure 9. Sensitivity study of Normal Depth as a boundary condition with three different energy slopes.

The presented result in figure 9 showed that a steeper slope of the channel resulted, as expected, in lower water levels in the downstream reach of the river. A flatter slope, on the other hand, resulted in increased water levels downstream. The results from simulations with different energy slopes gave different water levels from station 57 up to 642 which is below the area of interest.

0 1 2 3 4 5 6 7 1963 1941 1883 1869 1814 1789 1762 1723 1704 1685 1662 1628 15767 1533 1430 1358 1285 1232 1195 1170 1140 1102 1082 1059 966 929 888 843 814 766 724 685 642 554 381 308 57 El ev at io n ( m) Channel distance (m) Slope 0,001 Slope 0,002 Slope 0,003 Bottom level (m) Area of interest

(24)

16

The uncertainty interval of simulated water levels when using different energy slopes varied from 1,73 m up to 2,03 m, where the spread of the results are greatest (furthest downstream). The range of uncertainty for calculated water levels for this sensitivity analysis is 0,3 m.

A model with Normal Depth and Known water surface as a boundary condition were constructed to compare the results of two different options of boundary condition in HEC-RAS (Figure 10). The range of uncertainty from this result had a spread of 0,42 m. The variation in the two selected boundary conditions showed different results only in the downstream section of the study area, from 57 to 308 m.

Figure 10. Sensitivity analysis with Normal Depth versus Known water surface as downstream boundary

condition.

5.3. Sensitivity analysis of sea level as downstream boundary condition

The aim here is to get results of how increasing sea water levels can influence the drainage for the studied reach, as backwater effects can increase the risk of flooding. Several simulations with increased sea water levels for different return periods were assumed as downstream boundary conditions combined with a flows, which creates different climate scenarios.

Since the studied area is located near the outlet at the Baltic Sea, downstream boundary conditions with known water levels become important to analyze for different profiles. Figure 11 represents simulations with flows of different return period combined with the same sea water level as a downstream boundary condition. This was done to investigate the differences in water levels upstream for different climate scenarios.

0 1 2 3 4 5 6 7 1963 1941 1883 1869 1814 1789 1762 1723 1704 1685 1662 1628 15767 1533 1430 1358 1285 1232 1195 1170 1140 1102 1082 1059 966 929 888 843 814 766 724 685 642 554 381 308 57 El ev at io n ( m) Channel distance (m) HQ500, Slope 0,002 HQ500, Sea water level 0 m Bottom level (m)

(25)

17

Figure 11. Diagram of flows with different return periods combined with sea water level of 2,73 m.

The results showed that from station 57 to 814 the rise in sea level could lead to a damming effect for the stream’s discharge (Figure 11). The area that could be subject of flooding due to rise in sea level does not include the area of the projected railway. The probable flooding area due to rise in the sea level corresponded to the areas closer to the coast.

Figure 12 shows the outcome of a simulation with a 500-years flow combined with sea water levels at four different times.

0 1 2 3 4 5 6 7 1963 1941 1883 1869 1814 1789 1762 1723 1704 1685 1662 1628 ₁₅₇₆₇ 1533 1430 1358 1285 1232 1195 1170 1140 1102 1082 1059 966 929 888 843 814 766 724 685 642 554 381 308 57 El ev at io n ( m) Channel distance (m)

MQ, sea water level 2,73 HQ100, sea water level 2,73 HQ200, sea water level 2,73 HQ500, sea water level 2,73 Bottom level (m)

(26)

18

Figure 12. Diagram of flow with 500 years return period combined with different sea water levels for 2018,

2100 and 2150.

The water levels between the cross sections 814 and 1963 do not show to be influenced by changes in the sea water level. Figure 12 shows that only the lower reach of the river would be affected by rise in sea water level. The results showed that the different flow scenarios are the main driving parameters that control water levels in the upstream reach. Figure 13 shows an overview over the studied area. The cross sections within the marked area will not be influenced by rise in sea level. The results indicated that there is a critical point in station 814 up to where the rise in sea level affects the water levels for the studied reach. Table 4 summarizes how much the water levels will increase in each cross section up to section 814 due to a rise in sea level.

0 1 2 3 4 5 6 7 1963 1941 1883 1869 1814 1789 1762 1723 1704 1685 1662 1628 ₁₅₇₆₇ 1533 1430 1358 1285 1232 1195 1170 1140 1102 1082 1059 966 929 888 843 814 766 724 685 642 554 381 308 57 El ev at io n ( m) Channel distance (m)

HQ500, sea water level 0 m HQ500, sea water level 1,27 m HQ500, sea water level 1,93 m HQ500, sea water level 2,73 m Bottom level (m)

(27)

19

Table 4. Summary of stations affected of a rise in sea level compared to assumption with no rise in sea level.

Station Increased water level due to rise in sea level of 1,93 m

Increased water level due to rise in sea level of 2,73 m 814 +/- 0 + 0,18 766 +/- 0 + 0,16 724 +/- 0 + 0,18 685 +/- 0 + 0,23 642 +/- 0 + 0,56 554 + 0,05 + 0,61 381 + 0,06 + 0,64 308 + 0,06 + 0,67 57 + 0,52 + 1,32

Figure 13. Map of study area Mora River, marked area shows the reach for where sea level rise not affects the

(28)

20

5.4. Sensitivity analysis, configuration of cross sections

The results of this analysis were used to determine the effect of different number of cross sections account for the simulated water levels and representation of the rivers geometry. The results showed that different sets of cross sections do have an effect in resulting water levels and floodplain delineation. The simulations with different sets of cross sections also showed that the downstream reach has a bigger effect in floodplain delineation than that of the upstream reach.

Two different models were constructed with 4 and 28 cross sections. The first model corresponds to the assumed minimum number of cross sections. The second model corresponds to the number of cross sections determined according to equation in the background section. The resulting flood maps are based on a high flow of 95m3_{/s (in order to see floodplains) and with a boundary condition of the channel}

slope. No further consideration was taken into account in the model for the studied area, to evaluate the effect of the distance between cross sections in floodplain delineations in particular.

The results of this analysis showed that the 4-cross sections model (Figure 14) compared with the 28-cross sections model (Figure 15) produce different simulated water levels and floodplain delineation (Figure 16).

Figure 14. a) TIN-model with positioning of cross sections along the channel. b) Calculated water levels

imported from HEC-RAS based on 4 cross sections.

(29)

21

Figure 15. a) TIN-model with positioning of cross sections b) Calculated water levels imported from HEC-RAS

based on 28 cross sections.

Figure 16. Floodplain delineation with 28 respectively 4 cross sections.

(30)

22

The results from the constructed models using 4 and 28 cross sections demonstrated that different number of cross sections produce different results. The results showed that a small number of cross sections shows increasing extend of flood plain delineations with decreasing number of cross sections. The results (Figure 16) are different between the upper and lower part of the studied reach. The downstream reach showed a larger spread in floodplain delineations compared to the upstream reach.

5.5. Calibration and validation

5.5.1. Calibration and validation of low flows

The model setup throughout the calibration corresponded to a steady flow, the downstream boundary condition of normal depth with a slope of 0,002 and initial roughness coefficient (n) of 0,035. The n-values that were used to find the best representation of simulated water levels range from 0,005 to 0,3. The Manning’s n-value for the main channel was tested to calibrate the model. The calibration was carried out for low flows due to lack of information during high flow conditions. The observed data used for the calibration and validation is presented in table 5. In order to determine the sensitivity of the model to changes in Manning’s roughness coefficient, simulated water levels versus different Manning’s roughness coefficient is plotted in figure 17.

Table 5. Observed data used for calibration and validation

Date Flow (m3_/s) _{Width (m)} _{Water elevation (m)}

2015-10-26 0,16 5 3,53

(31)

23

Figure 17. Water profile from simulations with different Manning’s coefficient.

Figure 18. Simulated mean water levels versus Manning’s roughness coefficient.

The simulated mean water level (Figure 18) ranges from 3,59 m to 3,76 m. The results show a deviation of 0,17 m in the simulated water levels when using different n-values for the model.

0 1 2 3 4 5 6 1963 1941 1883 1869 1814 1789 1762 1723 1704 1685 1662 1628 ₁₅₇₆₇ 1533 1430 1358 1285 1232 1195 1170 1140 1102 1082 1059 966 929 888 843 814 766 724 685 642 554 381 308 57 El ev at io n ( m) Channel distance (m) n=0.01 n=0,035 n=0,045 n=0,055 n=0,1 Bottom level (m) 3,5 3,55 3,6 3,65 3,7 3,75 3,8 0,01 0,035 0,045 0,055 0,1 El ev at io n ( m) Mannings n-value

(32)

24

5.5.2. Sensitivity to number of cross sections

A specific cross section has been studied to investigate the sensitivity of the simulated water levels when changing the amount of cross sections. Different simulated water levels at cross section 1140 were produced by the model using different number of cross sections along the reach (Figure 19).

Figure 19. Comparison of water levels from simulations with 3, 5, 10, 15, 20, 28 and 38 cross section at section

1140.

Model results with different number of cross sections showed a difference in water elevation in a 0,25 m range from 3,51 to 3,26 m. These results indicate that reducing the number of cross sections affect the water level in the studied cross section. The accuracy of the model is determined by comparison with water levels observed in field of an elevation of 3,53 m.

To justify the results of how the amount of cross sections affects the model, mean water level for the entire reach included in the study was evaluated (Figure 20). The results show a difference in mean water elevation for the whole reach in a range of 1,015 m (3,65-2,635).

3,1 3,15 3,2 3,25 3,3 3,35 3,4 3,45 3,5 3,55 3,6 38 28 20 15 10 5 3 El ev at io n ( m)

Number of cross sections

Elevation (m)

(33)

25

Figure 20. Mean water level for the whole reach, for simulations with different amount of cross sections.

The model was calibrated for simulations with different number of cross sections to explain the sensitivity to the number of cross sections when simulating water levels. The results from the calibration of the simulations showed that different number of cross sections corresponded to different n-values to produce the minimum error between simulated and observed data (Table 6).

Table 6. Summary of calibration results to find the best manning’s n-value.

NUMBER OF CROSS SECTIONS CALIBRATED N-VALUE

38 0,045 28 0,055 20 0,05 15 0,06 10 0,04 5 2,25 3 0,15

The calibrated n-values were then tested through validation against a different observed discharge and a corresponding water stage to see the performance of the calibrated model (Figure 21, Table 7). The absolute error was calculated by taking the difference between simulated and observed water levels (Equation 5). 0 0,5 1 1,5 2 2,5 3 3,5 4 38 28 20 15 10 5 3 El ev at io n ( m)

(34)

26

Figure 21. Validation of calibrated n-values, graph showing absolute error in terms of water levels for

simulations with different amount of cross sections, control section 1140. Absolute error=measured water level-simulated water level (5)

Table 7.Summary of results from validation of the model when testing the calibrated n-value for the different models against measurements from another occasion

Number of cross sections

Observed water level m.a.s.l.

Simulated Water levels m.a.s.l. Absolute error hs-hm Used n-value 38 3,56 3,6 0,04 0,045 28 3,56 3,61 0,05 0,055 20 3,56 3,61 0,05 0,05 15 3,56 3,62 0,06 0,06 10 3,56 3,61 0,05 0,04 3 3,56 3,63 0,07 0,15

The calibration and validation showed that a value of n=0,045 with a model of 38 cross sections gave the best representation of simulated water levels for the model with an absolute error of 0,04 m based on the given data for this analysis. The magnitude of errors are between 0,04-0,07 m, the results showed that higher water levels result from a smaller number of cross sections when using the calibrated n-values. 0 0,01 0,02 0,03 0,04 0,05 0,06 0,07 0,08 38 28 20 15 10 3 Ab so lu te er ro r ( m)

(35)

27

5.6. Comparison with observed data

5.6.1. Changes in channel elevation

The DEM is produced from laser scanned data in which only the surface of the rivers can be measured. The difference between bottom profiles observed in field and profiles based on topographic data was evaluated to determine how the model responded to a changed bottom profile. A corrected bottom profile was created based on surveyed data in which the difference in elevation between measured data in field and the elevation based on DEM was around 1 m. The change in elevation for the bottom was assumed to be uniform for the whole river due to lack of data. Figure 22 represents the cross section geometry based on topographic data in GIS and surveyed data with GPS. The figure illustrates that the cross sections shapes are similar, although there are some general differences between the profiles. The thalweg (deepest part of the river) based on surveyed data was deeper and the geometry of the channel showed steeper slopes along the streams banks than the represented geometry profile based on the topographic data. The location of surveyed data with GPS is shown in appendix 4.

Figure 22. Comparison between cross section constructed from surveyed data and cross section based on DEM in GIS.

The results showed that a change in bottom elevation produced decreased simulated water levels for the studied reach (Figure 23). The result illustrates the water surface profile calculated for 500–years flow for Mora River. The boundary condition were established as normal depth at the downstream section.

0 1 2 3 4 5 6 0 2 4 6 8 10 12 14 16 18 20 El ev at io n ( m) Distance (m)

Surveyed cross section data

Edited cross section in HEC-RAS

Cross section based on topographic data

(36)

28

Figure 23. Results from simulated water elevation with change in bottom elevation versus original data based on

topographic data.

The result showed a minor influence of a corrected bottom elevation of the channel in simulated water levels. The mean water elevation for the whole reach showed a deviation of 0,1 m between the results of the edited bottom profile and the profile based on the DEM.

5.6.2. Calibration of low flows with corrected bottom profile

Although the cross sections geometry under the water surface showed to play a minor role in the determination of the water levels at extreme flows, the corrected bottom profile may affect the calibration of low flows. The calibration of the model after correcting the bottom geometry resulted in a n-value of 0,35. After the validation using the larger flow and the calibrated n-value of 0,35 an error of 0,2 m was produced. These results showed an unrealistic n-value related to reference tables for Manning’s n values for channels (Chow, 1973).

0 1 2 3 4 5 6 7 1963 1941 1883 1869 1814 1789 1762 1723 1704 1685 1662 1628 ₁₅₇₆₇ 1533 1430 1358 1285 1232 1195 1170 1140 1102 1082 1059 966 929 888 843 814 766 724 685 642 554 381 308 57 El ev at io n ( m) Station (m)

Original cross section based on topographic data

Edited cross section in HEC-RAS

(37)

29

6. Discussion

6.1. Sensitivity analysis of boundary conditions

The use of different downstream boundary conditions of normal depth versus known water surface showed a minor impact on simulated water level for the whole reach. On the other hand, the simulations showed a variation of +/- 0,42 m in the downstream area . The range of uncertainty when using different energy slopes of normal depth as a boundary condition also showed a change in water levels at the cross sections located downstream, with a difference of +/- 0,3 m. The outcome shows that simulations results are modified only in downstream locations. A general guideline based on these results could be therefore to set the downstream boundary condition further downstream from the area of interest to make the model more stable.

6.2. Climate scenarios and effects of rise in sea level

Based on simulations of flow scenarios with return periods of 100-, 200- and 500-years, it has been determined that under those conditions the water levels would stay within the channel and with no indication of the occurrence of flooding. The result shows that a rise in sea level would cause a backwater effect that would result in increased water levels along the river up to a certain point. The resulting water levels in the river due to sea level rise showed to depend on the rivers properties, such as flow rates and the elevation of the river upstream.

At station 814, assuming no rise in sea level and a 500-years flow, the model simulates a water elevation of 2,69 m. At the same station together with the same flow conditions, a rise in sea level of 2,73 m resulted in a water elevation of 2,87 m. These results indicate that the rise in sea level would lead to an increased water level of + 0,18 m due to backwater effect at the station furthest upstream that is affected by the rising sea water level. The locations further downstream are more affected by the backwater effect due to rise in sea level, since the bottom level is located at a lower elevation (table 4). The case study and all the results corresponds in particular to Mora River. However, the increased sea level will probably affect all the rivers along the projected railway that has it drainage in the Baltic Sea. Although the results do not show backwater effect affecting the area of interest for the railway project, the consequences further down may in turn cause other effects, such as erosion and change in sediment transport. These could lead to modified water levels in the river, which in turn can cause problems.

6.3. Sensitivity analysis, configuration of cross sections

The simulation with 3 cross sections resulted in larger floodplains in comparison with the result of the simulation with 28 cross sections. The results showed a larger difference in the floodplain delineation further downstream the studied reach. This could be the results of larger changes in the geometry of the channel in the downstream reach and its meandering characteristics. The larger floodplain areas could

(38)

30

be explained by the way in which HEC-RAS interpolates the topography between cross sections. It is important to include all the areas that are of major importance for the flow to provide an accurate model when developing cross sections. When the areas between the cross sections are similar to the areas corresponding to the cross sections the in-between areas are not of a major importance. The results of this case study confirmed this affirmation, showing almost no variation of the delineated floodplains in the upstream reach, which has a more homogeneous geometry and surrounding area than the downstream reach. Thus, the floodplains delineated by the model using both a small number as well as a larger number of cross sections were relatively similar. A few cross sections might have been enough in the upstream reach while more and closer cross sections would be needed for the downstream reach. Although the result showed larger floodplain when using less cross section, lower water elevations also resulted when using less cross sections. These results mean that the model using less cross sections underestimates the water levels and overestimates the floodplain at the downstream reach. An important conclusion from this is that if a cross section would be located at a location were the river is wide, this may result in higher water levels than expected in the reality. The same flow in a narrower cross section will result in higher water levels than what is calculated in a wider cross section.

6.4. Calibration and validation

The Manning’s n-value obtained from the calibration was 0,045, which is within the range of reasonable values according to predefined values for natural channels (Chow, 1973). The validation of the model showed no significant differences compared with measured values. The validation of the model using 38 cross sections based on measured data used for calibration and validation from two occasions showed an absolute error in water levels of no more than +/- 0,04 m.

Low flows and shallow depth can cause significant model instability (Brunner, 2010b). It was therefore important to keep in mind such potential instability when evaluating the calibration of the model within this study because it is based on low flows. Small changes or irregularities in the channel geometry of the model can lead to a significant change in water depth.

Since the aim of this study was to use the model to simulate water levels for higher flows, it was important to take into account the uncertainties, even if the calibrated model is adjusted for low flows (as data for high flow conditions were not available). Calibration for higher flows is preferable to minimize the uncertainties to obtain an accurate simulation of higher flows.

The model was calibrated for simulations with a different number of cross sections to explain the sensitivity of the amount of cross sections when simulating water levels. The results from the validation showed larger absolute errors for simulations with smaller number of cross section which explains the accuracy of the model using higher number of cross sections.

(39)

31

6.5. Change in channel elevation

The result shows a small changes in simulated water elevations after correcting the elevation of the bottom profile. A difference of 0,1 m in the mean water elevations resulted from a model based on a 500-years flow. The standard error for the topographic data is 0,25 m (Lantmäteriet, 2015), which is important to keep in mind due to several uncertainties. The difference in simulated water levels may therefore be negligible in such way that even if the bottom of the channel is corrected the results would be more or less the same. The bottom of the channel showed to have small influence on the results of simulated water levels.

6.5.1. How does a change in channel elevation affect the calibration of low flows?

This analysis showed that the model is very sensitive to an elevation change when modelling low flows. Based on the data used for calibration of low flows, the results indicate that the calibration without bottom profile changes gave more reasonable results. This may be due to accuracy in field data, few and inadequate measurements, structure of the model and unawareness of the bottom profile of the river. The low flow values used for calibration of the model could be a source of instability. If the bottom profile should be corrected, accurate bottom measurements are necessary to reduce the uncertainties in the results.

Since the model will be used for simulations of higher flows, calibration data collected during high flows conditions with the corresponding water levels would probably give more reliable data. In other words, using low flow data for modelling and calibration intended for high flows introduces many uncertainties and makes the processes highly sensitive to changes in the channel geometry. This cause a need of more cross sections to enable the analysis of the flow. Since the study purpose is to modelling higher flows, it may not require as many cross sections along the reach.

6.6. Further studies

The results in this analysis should be further studied with additional investigations. Assumptions and generalizations done for this case study should be evaluated by analyzing several streams and comparing how the results may differ. It would also be interesting to consider structures along the studied reach to calculate water levels due to back water effect for different climate scenarios. Lack of data resulted in calibration based on low flows for the model; it would therefore be of interest to calibrate the model using higher flow data and compare the calibration results of low flows versus high flows.

(40)

32

7. Conclusions

The main objective of this thesis was to simulate floods in the Mora River and assess the uncertainty associated to the simulations of high water levels using sensitivity analyses.

- The results indicate that the spacing between cross sections has a significant impact in the simulation of water levels under low flow conditions even if allowing for calibrated n-value for the model.

- The results show that the lower reach is more sensitive to the configuration of cross sections. It is therefore important to identify the areas where uncertainties can be more critical for the results.

- The validation exercise shows that the absolute error of simulated water levels increases as the average spacing between cross sections increases.

- Even if this study shows that there was no improvement of model results by correcting the bottom profile, it is important to note that the uncertainties according to the bottom profile can give different results with more accurate data according to Manning’s roughness coefficient. Uncertainties of simulated water levels for higher flows can thus be of a larger size.

The second goal of this study was to investigate the risk of flooding in a future climate scenario. For this purpose, models with different climate scenarios were built using the rise in sea level as a downstream boundary condition combined with high flows of varying return periods.

- Based on the simulations of flow scenarios with return periods of 100-, 200- and 500-years, the simulated water levels are likely to stay within the main channel limits and no indication of flooding inundation was evidenced.

- A backwater effect due to sea level rise can impact the water levels along the river, in particular in the area downstream from the area of interest. It is necessary to keep in mind other possible effects of this outcome, such as erosion and sediment deposition which can lead to different water levels.

Since the model is unavoidably associated to a number of uncertainties, the method presented here should be used only as a tool to support a more comprehensive assessment of flood risk. It is also important to notice that the results from this analysis correspond to the case study in Mora River in particular.

(41)

33

8. Acknowledgements

I will give my greatest appreciation to Tyréns, with all employees who made this project possible and contributed with help and advice. Special thanks to my supervisor at the company Jannike Sondal that has given me valuable assistance and guidance throughout the project.

I also would like to thank my supervisor at the University of Uppsala, Giuliano Di Baldassare for valuable comments, support and discussions of work.

(42)

34

9. References

Brunner, G. W. (2010a). HEC-RAS River Analysis System Hydraulic Reference Manual. Version 4.1. Davis, CA: US army corps of engineering, hydrologic engineering center.

Brunner, G. W. (2010b). HEC-RAS River Analysis System Users Manual. Version 4.1. Davis, CA: US army corps of engineering, hydrologic engineering center.

Cameron, T & Ackerman, P. (2009). HEC-GeoRAS. Version: 4.0. Davis, CA: US Army Corps of Engineers Institute for Water Resources Hydrologic Engineering Center.

Castellarin, A., Di Baldassarre, G., Bates, P. D. & Brath, A. (2009). Optimal Cross-Sectional Spacing in Preissmann Scheme 1D Hydrodynamic Models. Journal of Hydraulic Engineering, 135(2), pp. 96– 105.

Castellarin, A., Domeneghetti, A. & Brath, A. (2011). Identifying robust large-scale flood risk mitigation strategies: A quasi-2D hydraulic model as a tool for the Po river. Physics and Chemistry of the Earth, Parts A/B/C, 36(7-8), pp. 299–308.

Chow, V. T. (1973). Open-channel hydraulics. New York: McGraw-Hill book company.

Devia, G. K., Ganasri, B. P. & Dwarakish, G. S. (2015). A Review on Hydrological Models. Aquatic Procedia, 4, pp. 1001–1007.

Di Baldassarre, G., Schumann, G., Bates, P. D., Freer, J. E. & Beven, K. J. (2010). Flood-plain mapping: a critical discussion of deterministic and probabilistic approaches. Hydrological Sciences Journal, 55(3), pp. 364–376.

Genovese, E. (2006). A methodological approach to land use-based flood damage assessment in urban areas: Prague case study, EUR 22497 EN. European Commission. Joint Research Centre. Institute for Environment and Sustainability.

Horrit, M. & Bates, P. D. (2002). Evaluation of 1D and 2D numerical models for predicting river flood inundation. Journal of Hydrology, Volume 268, pp. 87–99.

Klimat- och sårbarhetsutredningen (2006). Översvämningshot- Risker och åtgärder för Mälaren, Hjälmaren och Vänern. (SOU 2006:94). Stockholm: Miljö- och samhällsbyggnadsdepartementet. Mambretti, S. (Ed) (2012). Flood risk assessment and management. Southampton and Boston: WIT Press. (Safety & security engineering series).

Manfreda, S., Samela, C., Gioia, A., Consoli, G. G., Iacobellis, V., Giuzio, L., Cantisani, A. & Sole, A. (2015). Flood-prone areas assessment using linear binary classifiers based on flood maps obtained from 1D and 2D hydraulic models. Natural Hazards, 79(2), pp. 735–754.

Md Ali, A., Solomatine, D. P. & Di Baldassarre, G. (2015). Assessing the impact of different sources of topographic data on 1-D hydraulic modelling of floods. Hydrology and Earth System Sciences, 19(1), pp. 631–643.

Nordblom, O. & Petzén, M. (2014). Vägledning för översvämningskartering av vattendrag. Myndigheten för samhällsskydd och beredskap (MSB).

(43)

35

Queensland (2012). Queensland Floods Commission of Inquiry. Available from:

http://www.floodcommission.qld.gov.au/__data/assets/pdf_file/0007/11698/QFCI-Final-Report-March-2012.pdf. [2016-03-20].

SMHI (2016a). Flödesberäkning för Moraån, working material.

Södertälje kommun (2004). Sjöar och Vattendrag i Södertälje. Södertälje kommun: Miljökontoret. Tyréns och ÅF (2015). Rapport klimattest, working material.

Walsh, S. & Miskewitz, R. (2013). Impact of sea level rise on tide gate function. Journal of Environmental Science and Health, Part A, 48(4), pp. 453–463.

Weichel, T., Pappenberger, F. & Schulz, K. (2007). Sensitivity and uncertainty in flood inundation modelling - concept of an analysis framework. Advances in Geosciences, 11, pp. 31-36.

Westlin, S. Modigh, A. Valen, C. Frost, C. Gauffin, J. Sydow, K. Fröberg, L. (2012). Klimatanpassning i fysisk planering: vägledning från länsstyrelserna. [Malmö: Länsstyrelsen i Skåne län]

Internet resources

Lantmäteriet (2015) GSD-Höjddata, grid 2+ Available from: https://www.lantmateriet.se/sv/Kartor-och-geografisk-information/Hojddata/GSD-Hojddata-grid-2/. [2016-04-25].

SMHI (2015) Återkomsttider. Available from: http://www.smhi.se/kunskapsbanken/aterkomsttider-1.89085. [2016-02-15].

SMHI (2016b) Vattenwebb. Available from: http://vattenwebb.smhi.se/. [2016-02-05]. Trafikverket (2015). About The East Link Project. Available from:

http://www.trafikverket.se/en/startpage/projects/Railway-construction-projects/Ostlanken---East-Link-project/about-the-east-link-project/. [2016-02-02].

Trafikverket (2016) BaTMan. Available from: https://batman.trafikverket.se/externportal. [2016-02-04].

Software

Esri (2015). ArcGIS 10.3.1 for Desktop. Version: 10.3.1. Redlands, California: Environmental Systems Research Institute, Inc.

U.S. Army Corps of Engineers (2008). HEC-RAS. Available from: www.hec.usace.army.mil. [2016-02-16]

(44)

36

10. Appendix

Appendix 1. Simulated water surface profiles in HEC-RAS, with flows with different return periods.

Appendix 2. Reference model constructed in HEC-GeoRAS

0 500 1000 1500 2000 0 2 4 6 8 10

Referensmodell20140413 Plan: Plan 01 2016-05-13

Main Channel Distance (m)

E lev at ion ( m ) Legend WS HQ500 WS HQ200 WS HQ100 WS MQ Ground LOB ROB Morariver Centerline

(45)

37

Appendix 3. Reference model in HEC-RAS.

Appendix 4: Location for measured cross sections with GPS.

Centerline 1963.878 1941.405 1883.616 1814.685 1789.004 1723.271 1704.722 1685.66 1662.159 1628.648 1567.672 1533.186 1358.705 1285.287 1195.112 1140.577 1102.839 1082.847 1059.702 966.2252929.1038 888.7863 724.3755 685.1987 642.1485 381.5645 308.5692 57.76294 M o r_{a r} i v_{e r}

(46)

(47)

(48)