The town of Morpeth, which lies in the Wansbeck catchment, experienced a catastrophic flood on the 6th September, 2008 which has resulted in this research. The main aim of this study is the development of a continuous simulation approach for flood risk estimation applied to the Wansbeck catchment during the 2008 flood.
The objectives include: Analysis spatial rainfall forecast from the Met Office with 1 km high resolution and use 24 ensemble rainfall models for every 5 minutes across the area of a subcatchment as input to rainfall-runoff model. Run calibrated rainfall- runoff model (SHETRAN) using inputs of catchment rainfall (24 ensemble rainfall models), catchment maps including maps of catchment area, DEM, soil type and land use to simulate flood flows in each tributary during the event. Choose peak flows simulated by maximum, medium and minimum ensemble rainfall through the flows model simulated by all 24 ensemble rainfall models, as the input to NOAH 1D model. Run calibrated NOAH 1D (with existing river cross sections and structure dimensions) using inputs of SHETRAN flows at Font and Upper Wansbeck (upstream flows) to get flood flows and levels at Mitford (downstream flow) without storage. Build a conceptual model of the upstream storage areas (in Upper Wansbeck) to compare water levels and flows with and without storage during the event. Calculate water volume that has been stored with the storage in different elevations.
Contents
Acknowledgements
Abstract
Contents
List of Figures
List of Tables
1 INTRODUCTION
1.1 Project Background
1.2 Aim and Objectives
2 LITERATURE REVIEW
2.1 Introduction
2.2 The FEH statistical method
2.3 The FEH Rainfall-Runoff method
2.4 Continuous Simulation Approach
2.4.1 Case studies
2.4.2 Advantages and Disadvantages of Continuous Simulation
2.5 Overview of the Models
2.6 Forecast ensemble rainfall data
2.6.1 Analysis of space-time rainfall characteristics
2.6.2 Advantages and Disadvantages of Ensemble Rainfall Data
2.7 SHETRAN
2.7.1 Limitations of Rainfall Runoff Models
2.8 NOAH 1D
2.8.1 Manning’s n
2.8.2 Limitations of 1D River model
2.9 Previous study: Reconstruction of the Dynamics of a Flood Event Based on publicly sourced information
2.10 Previous study: Continuous Simulation Approach for Flood Risk Estimation
2.11 Study Area Description
2.12 Flood risk characteristics – Wansbeck catchment
2.12.1 Rainfall
2.12.2 Physical Factors
2.12.3 Impact of Flooding
2.13 Existing Defences
2.14 Future Management Proposed Flood Defences for Morpeth
3 Materials and Methods
3.1 Data Requirements
3.2 Model preparation and processing
3.3 Model Linking and Application
3.3.1 Wansbeck Rainfall During the Event
3.3.2 SHETRAN Model Development
3.3.2.1 Elevation, Land use and Soil type maps
3.3.3 Calibration of the SHETRAN model
3.3.4 Continuous Simulation for 2008 flood
3.3.5 NOAH 1D model development
3.3.6 Calibration and Validation of NOAH 1D Model
4 MODEL LINKING and APPLICATION
4.1 Design 2008 Flood without Storage
4.2 Storage near Mitford Hall and Font Ford Lodge
4.3 Design Event with Storage
4.3.1 NOAH 1D Flows with Storage
4.3.2 NOAH 1D Levels with Storage
5 DISCUSSION
6 CONCLUSION
7 RECOMMENDATIONS FOR FUTURE STUDY
8 REFERENCES
9 APPENDICES
The 2008 Morpeth Flood: Continuous Simulation Model for the Wansbeck
Catchment
Safieh Javadinejad
Submitted in partial fulfilment of the requirements for the degree of Master of Science in Hydrology and Climate Change
Newcastle University School of Civil Engineering & Geosciences
August 2011
Acknowledgements
First of all, I am heartily thankful to my supervisors, Dr Vedrana Kutija and Dr Geoff Parkin for all the help, encouragement and support during this study. I would also like to thank Dr Steve Birkinshaw for guidance and help in relation to the SHETRAN rainfall-runoff model.
I also wish to acknowledge my mother, Manijeh Mohammadi for standing beside me during this project. She has been my motivation for continuing to improve my knowledge and without her I could not have undertaken this study. I would also like to thank my sister and my family for all the help and support through this project.
Last but not least, I would like to thank all my teachers, who participate in the Water Resource Engineering programme (in Newcastle University) for their generous help and the knowledge they shared during the year as I have experienced a very pleasant and valuable academic life in the university.
Abstract
The town of Morpeth, which lies in the Wansbeck catchment, experienced a catastrophic flood on the 6th September, 2008 which has resulted in this research. Occurrence of similar catastrophic floods around the world has increased the interest of the scientific community in using an alternative method i.e. continuous simulation to estimate flood risk. This method is completely different from current standard methodology [e.g. Flood Estimation Handbook (FEH)] in the UK for estimating flood risk.
So, this study has examined the methodology used to develop a continuous simulation approach for the Wansbeck catchment and then made the model for upstream storage (near Mitford Hall and Font Ford Lodge) to reduce flood risk in the town of Morpeth. A previous study of the catchment has investigated the possibility of utilizing Lightwater storage to reduce flood risks to the town. The approach was to use data from 100 years of rainfall and to simulate the rainfall-runoff model to obtain the most significant flood. Then unsteady 1D hydraulic river model with and without Lightwater storage was modelled. Therefore, in this project, in order to show a fuller picture of a flood hydrograph, 24 ensemble rainfall model (during the event) with high resolution for every 5 minutes has been used to input rainfall-runoff and 1D river model during continuous simulation. During this project, an analysis of rainfall around the Wansbeck catchment has been made by using the 24 forecast ensemble rainfall model. Furthermore, a calibrated SHETRAN model has used the rainfall to simulate a rainfall-runoff model for every 15 minutes. The SHETRAN model was applied because it is a physically based, spatially- distributed modelling system for water flow, sediment and solute transport in the catchment. Hence the peak-flows hydrograph simulated by maximum, medium and minimum rainfall has been used as inputs for the NOAH 1D model with and without storage. Detailed results are shown during the project.
List of Figures
Figure 1.1- Map of Wansbeck catchment
Figure 2.1 - Areas of Morpeth severely flooded on 6th September 2008
Figure 2.2 - 100 year flows at Mitford, with and without Lightwater storage
Figure 2.3 - Morpeth, Nurthumberland and location in UK
Figure 2.4 - Morpeth before and after the 2008 flooding
Figure 2.5- Rainfall map during 04-06 September 2008- Wansbeck catchment
Figure 2.6 Exceed of surface runoff during 6 September 2008, Morpeth
Figure 2.7 - Impacts of 2008 flooding- Morpeth
Figure 2.8 - Flood cells in Morpeth
Figure 2.9 - Location of potential storage areas in the Wansbeck catchment
Figure 3.1- Wansbeck catchment and the tributaries
Figure 3.2 - Overview of project methodology
Figure 3.3- Compare observed data and average forecast ensembles rainfall data during the event (30 hours) in Wansbeck catchment
Figure 3.4- 48 hour total rainfall over Northumberland for 0:00 5 September to 24:00 6 September 2008, showing the Wansbeck catchment to the Mitford gauging station (station 22007)
Figure 3.5-Location of rain gauges around the Wansbeck catchment
Figure 3.6-24 hour rainfall totals for A) 5 Sept 2008 (9am 5th to 9am 6th) and B) 6 Sept 2008 (9am 6th to 9am 7th) interpolated from measured daily rain gauge totals
Figure 3.7- 1km DEM map for the catchment area
Figure 3.8- 1km land use map for the catchment boundary at Bothal Mill
Figure 3.9- 625km superficial geology map for the catchment
Figure 3.10- SHETRAN calibration with time series between 1999 and 2000
Figure 4.7- Schematise location of the cross sections for the culverts at the Font
Figure 4.8- Peak flow simulated by maximum ensemble rainfall at Mitford
Figure 4.9- Peak flow simulated by medium ensemble rainfall at Mitford
Figure 4.10- Peak flow simulated by minimum ensemble rainfall at Mitford
Figure 4.11- Peak level simulated by maximum ensemble rainfall at Mitford
Figure 4.12- Peak level simulated by medium ensemble rainfall at Mitford
Figure 4.13- Peak level simulated by minimum ensemble rainfall at Mitford
List of Tables
Table 2.1- Main Processes Represented in SHETRAN
Table 2.2 - Peak flows at Mitford for various outflow rates from Lightwater
Table 2.3 - Existing flood defences in Morpeth
Table 3.1- Data required to undertake the project
Table 3.2- Overview of Model linking & Application
Table 3.3- Rainfall statistics 4-6th September 2008
Table 3.4- The maps that were used as inputs for SHETRAN
Table 3.5- Categories of land use types for use in SHETRAN
Table 3.6- Identification of soil types with using the Soil Site Reporter for Wansbeck
Table 3.7- Final parameter values for calibration of SHETRAN model
Table 3.8- SHETRAN peak flow simulated by maximum, medium and minimum ensemble rainfall for Wansbeck tributaries
Table 3.9- Description of cross section reaches and chainages
Table 3.10- Procedure to approach NOAH 1D model for Wansbeck
Table 4.1- Peak flow observed and simulated at Mitford for the 2008 flood
Table 4.2- Total length of each Edge
Table 4.3- Heights and areas for the Upper Wansbeck storage and cross sections before
Table 4.4- Heights and areas for the Font storage and cross section before
Table 4.5- Peak flows at Mitford with and without the storage
Table 4.6- Peak levels at Mitford with and without the storage
Table 4.7- Water volume with and without storage
Table 5.1- Positive and negative impacts of upstream storage
1 INTRODUCTION
1.1 Project Background
Morpeth town lies in the Wansbeck catchment area in Northumberland, a relatively small river catchment which covers 331 km2 . The main reach of the Wansbeck has an active flood plain that is between 100m to 300m wide and the town itself is situated within this floodplain. The river has two main tributaries, the River Font and Upper Wansbeck that flows into the Mitford (Figure 1.1).
This image was removed by the editorial team due to copyright issues.
Figure 1.1- Map of Wansbeck catchment
On the 6th and 7th September 2008, Morpeth, experienced severe flooding which resulted in approximately 1000 properties being flooded or affected by the prolonged, heavy rainfall combined with an already very wet catchment which caused problems in the town of Morpeth. With no soil capacity to absorb additional moisture, the rain was rapidly conveyed into the rivers causing flows and water levels in the River Wansbeck to rise very quickly. The existing defenses that had protected the town for almost 50 years were overtopped and, in places, were damaged (JBA, 2009).
It is therefore important to have a flexible approach to flood risk estimation. The current standard methodology for estimating flood risk in the UK is the Flood Estimation Handbook (FEH), which suggests a robust approach based on observed data and catchment characteristics. The hypothesis of the project is that the FEH is unable to predict changes in the future and does not offer a method to estimate the possible impact of changes in climate and land usage. As a result the FEH focuses on the storm hydrograph for a given return and does not give the full effect of antecedent catchment conditions. Therefore, it is unable to show base flows on a seasonal basis or to consider wider flood generating mechanisms such as events caused by widespread snowmelt. In order to combat the limitations of the FEH, this project indicates a new methodology of continuous simulation for flood risk estimation with the Wansbeck catchment during the 2008 flood as a case study. The continuous simulation includes three successive models; at first 24 forecast ensembles rainfall model will generate and then put the rainfall model in a calibrated rainfall- runoff model (SHETRAN) to produce flow for every 15 minutes. Hence 1D river model will be used to simulate river levels with and without conceptual upstream storage for the event (NOAH 1D).
1.2 Aim and Objectives
The main aim of this study is the development of a continuous simulation approach for flood risk estimation applied to the Wansbeck catchment during the 2008 flood.
The objectives include:
- Analysis spatial rainfall forecast from the Met Office with 1 km high resolution and use 24 ensemble rainfall models for every 5 minutes across the area of a subcatchment as input to rainfall-runoff model.
- Run calibrated rainfall- runoff model (SHETRAN) using inputs of catchment rainfall (24 ensemble rainfall models), catchment maps including maps of catchment area, DEM, soil type and land use to simulate flood flows in each tributary during the event.
- Choose peak flows simulated by maximum, medium and minimum ensemble rainfall through the flows model simulated by all 24 ensemble rainfall models, as the input to NOAH 1D model.
- Run calibrated NOAH 1D (with existing river cross sections and structure dimensions) using inputs of SHETRAN flows at Font and Upper Wansbeck (upstream flows) to get flood flows and levels at Mitford (downstream flow) without storage.
- Build a conceptual model of the upstream storage areas (in Upper Wansbeck) to compare water levels and flows with and without storage during the event.
- Calculate water volume that has been stored with the storage in different elevations.
To approach the objectives, 1km high resolution of 24 ensemble forecast systems for generation of rainfall data were used between the 7th of July and 8th of September 2008. Hence, a SHETRAN model was made and the ensemble rainfall data was used to approach flows in each tributary. In this research just the hydrographs of maximum, medium and minimum ensemble rainfall are used. Then the peak flows from SHETRAN are used as input to the NOAH 1D model to get water levels in each cross section during the event. Two simulations of the NOAH model are run. In one of them, the situation of the catchment during the event without storage is run and in another the situation with storages to compare peak flow and peak levels simulated by different ensemble rainfall.
2 LITERATURE REVIEW
2.1 Introduction
The literature review for this study will start by introducing the Flood Estimation Handbook (FEH) which is divided into two methods; the Statistical Method and Rainfall-Runoff Method. These methods are current methodologies for flood risk estimation in the UK. The limitations of the methods are defined during the literature review. Then, because the Continuous Simulation method is used in this project, the method and limitations for flood risk estimation are surveyed. Following this, there is an overview of the models that have been applied in this study. This includes an explanation of the Rainfall Forecasting model, the Rainfall-runoff model (SHETRAN) and the 1D river model (NOAH 1D). Following this, the Wansbeck catchment and Morpeth flood 2008 are explained as a case study. In addition existing flood defenses and future flood defenses, e.g. upstream storage in Morpeth, are introduced. Finally, a summary of previous studies for flood risk in Morpeth are explained.
2.2 The FEH statistical method
The FEH (1999) makes up the primary flood estimation methods in the UK (JBA, 2008). Whilst the FEH shows both statistical and rainfall-runoff methods, the statistical method is the main method (Spencer et al., 2004). A rainfall-runoff based approach was developed in the early 1970’s from the original method and published in the flood studies report (FSR) (Kjeldsen et al., 2008). A statistical method based on an index flood that is the median annual flood, the FEH statistical method, is the preferred method for measuring QMED at ungauged sites. This is by data transfer from a similar gauged catchment peak flow of the required probabilities which are distinguished by applying growth factors to the index flood. Overall, the statistical method uses two approaches; the single site analysis and the pooling group analysis. The single site analysis only uses the flood data from the specific site. However, a pooling group contains similar catchments which can be anywhere in the particular area. The FEH and software that implements the statistical method, WINFAP-FEH, generates annual maxima (Amax) and peaks -over-threshold flows and catchment descriptors for approximately 1000 gauging stations on CD. The new WINFAP-FEH 3 software has been improved under license by Wallingford HydroSolutions Ltd, a spin-out company from the Natural Environment Research Council (NERC) and cooperates with different significant scientific advances from the flood research team at NERC’s centre for Ecology & Hydrology. The new science methods cooperate with the software include:
- Increase single site analysis
- Improved estimation of QMED from catchment descriptors
- Catchment centroid and a new catchment descriptor, flood plain extent
- Increase pooling procedure and flood frequency curve wizard
- New urban adjustment method
- Developed calculator tools for measuring flood characteristics
The new version of WINFAP-FEH links to the EA’s HIFLOWS-UK website allowing fast assessment of gauging station quality and relevance (Centre for Ecology & Hydrology, 2009). However, the FEH statistical method (using the pooling group)has some limitations which include the many gauging stations, the number of different offices and their different data archiving systems, and the several different sources of data that are used in the statistical method (especially in the pooling group) cause the complexities of some rating histories.
Also there is a limitation in the use of the single site method for the period of T/2 length (Kilsby, 2011). In addition, there is often limited information about past events, especially before the start of digital records. A frequent source of uncertainty has been how some stations really behave at the highest flows, particularly where they have been designed and operated primarily for the measurement of low flows (Spancer et al., 2004). In addition the FEH statistical method cannot measure changes in the future. For example the effect of climate change or land use change cannot be analyzed (Calver et al., 2000 and environmental statement, 2009). As a result it is based solely on past flood flow data (Wilby et al., 2007).
The last limitation of the FEH statistical method is that the impact of convective storms, snowmelt and frontal systems are not considered in the method (Adamowski, 2000 and Kokkonen et al., 2004) as this method is based on peak flow data.
2.3 The FEH Rainfall-Runoff method
Rainfall runoff models for design flood estimation have been applied by engineers and hydrologists for more than a century. Throughout this time, methodologists have improved this by increasing computing power, the development of analytical techniques and by steadily increasing data records. The rainfall runoff method was published in 1975 as part of the FSR. It has been central to design flood estimation in the UK (Rodding Kjeldsen, 2007).
Rainfall-runoff (RR) models are commonly applied tools to extrapolate stream flow time-series in time and space for operational and scientific surveys (Beven, 2000). Rainfall runoff models allow available extending stream flow records during the time to anticipate the behaviour of catchments for different climate scenarios. Through extrapolation in space, they can simulate the response of catchments for which little or no time series of stream flow measurements are available (Wagener and Wheater, 2006). Also flood frequency can be measured by using the FEH rainfall-runoff method under standardized design conditions (Ashfaq and Webster, 2007).
The FEH rainfall-runoff method has three main parameters including; unit hydrograph time to peak, percentage runoff and base flow (Kilsby, 2011). The model parameters were measured for 143 gauged catchments in the UK. To allow estimation of T year events using the rainfall-runoff model, a depth-duration-frequency (DDF) model was used. A simulation study was considered specifying combinations of antecedent soil moisture conditions and the return period of the design rainfall needed to make flood hydrographs with a given return period (Rodding Kjeldsen, 2007).
The disadvantages of the FEH rainfall-runoff include:
- There are many parameters with considerable interaction and redundancy, resulting in the model identification
- No satisfaction state- correction scheme if used with flow routing component
- The model is complex with fairly insensitive to initial soil moisture conditions(Wagener and Wheater, 2006)
- It needs more designing for use at a daily time step. Also probably the model requires further development for practical use at a 15 minute time-step
- The rainfall-runoff model needs more experimentation with different structures
- Prediction of effect of parameter changes can be difficult. Only antecedent conditions accommodated by updating
- It provides poor simulation- mode performance for catchments with complex responses
- The model needs minimum understanding of hydrology to be used
- When the rainfall-runoff model was used on complex catchments with a significant ground water component, the model was ineffective
- Run RR model works slowly making automatic calibration impractical (Bell et al., 2001)
- The rain gauges record the point measurements of rainfall, so there are subsequent uncertainties in then using the point data to measure the catchment rainfall (Asquith and Famiglietti, 2000)
Therefore, to conclude, the RR model provides storm hydrographs. This means that it has been used widely for a range of situations and catchments with success. On the other hand, there is no framework via which land use or climate change impacts may be assessed easily. In addition to this the mode takes an event based approach which makes it impossible to observe the benefits for flood alleviation over the entire water cycle or in the context of integrate river basin management.
2.4 The Continuous Simulation Approach
With the increase of available computing power and sub-daily rainfall and flow data in digital form, it is becoming easy to move towards standard techniques for a flood estimation hydrograph based upon the continuous simulation of flow (Cameron et al., 2000). By using a continuous rainfall series running a continuous model and going on to analyze the flood peaks of the simulated flow series, it is no longer compulsory to provide the assumptions behind the interest event approach or the simplifying conception of the event based derived-distribution approach (British Hydrological Society,2011). In addition, as soil moisture accounting is dealt with automatically, assigning a precedence wetness condition is no longer a problem. Although this method has not yet been fully confirmed, initial consequence results are encouraging. Also, the use of continuous simulation opens up different areas of research tendencies. These include: the potential impact of climate change upon flood frequency, the selection stochastic rainfall model, and consistency between continuous flow simulation and flood frequency estimation rainfall-runoff model parameterisations (Lan Littlewood, 2002).
Continuous simulation has the potential to be used for multipurposes. Also, the performance of reservoir regulation schemes and combined sewer overflows are probably estimated by this methodology. As a result, all the flow range is simulated that encapsulates winter refilling of reservoirs and wet and dry sequencing of river flows (Calver et al., 2005). In what follows, a continuous simulation methodology is appropriate for flood frequency estimation (Lan Littlewood, 2002).
2.4.1 Case studies
A recent study applied a continuous simulation approach to the Wansbeck catchment in Morpeth. The study used a stochastic rainfall model to generate 100 years of hourly rainfall data. The rainfall data was then modelled in SHETRAN to simulate flows in Wansbeck. A series of high flow hydrographs were used in the river model software NOAH to assess the effect of the proposed lightwater storage on the water level and to prevent flood risk in the catchment. The study found that the 100 year flow at Wansbeck without storage was 139 m3 /s from the continuous simulation and 134m3 /s with storage (with maximum outflow rate of 80 m3 /s by modelling in NOAH (Ward, 2008).
Another study considered the South Tyne catchment in Northumberland and used a continuous simulation for flood forecasting for an upland catchment. The study applied a real time forecasting of rainfall. Then it used to model in a rainfall-runoff model (RFFS model) to simulate flows. Hence, peak flows from the hydrographs were modelled in the river model (Neural network model) to alleviate flood risk. Finally, the neural networks produced reasonable forecasts of the flood events within the validation period (Cameron et al., 2002).
2.4.2 Advantages and Disadvantages of Continuous Simulation
The continuous simulation modelling (CSM) rainfall-runoff approach to design flood estimation has many benefits and has the potential to compensate for the limitations of the Design Event Approach. The aim of continuous simulation models is to show the major processes responsible for converting the input catchment rainfall into stream flow hydrographs (that are made over long periods of time from the input of historical rainfall series, potential evaporation and other climatological and catchment attributes). An important feature of these models is the use of the continuous water budget model for the catchment, therefore precedence conditions for each storm is known. A relatively sophisticated hydrological method is able to simulate the whole hydrological cycle The view of different authors is that CSM has the potential to overcome the limitations of the current single event approach to design flood estimation. Firstly, the request for the use of synthetic storms is eliminated, by applying actual storm records. The second is that, subjectivity in choosing antecedent conditions for the land surface is not needed as a water budget accounts for, in each time step of the simulation, antecedent moisture conditions (AMC). Also the duration of the storm is not an issue as AMC and the water budget are modelled precisely for each time step and the assumption for the return period of the input rainfall no longer should be made (Rahman et al., 1998).
Some of the disadvantages of continuous simulation are: • If the modelling time scale is too large, it loses sharp events.
- The extensive data requirements, which require a lot of time and try to obtain and organize the input data.
- The expertise needed to estimate parameter values so that historical hydrographs are simulated adequately.
- The continuous simulation approach at present is largely untested for practical use; for more than practical use of the FEH (DEFRA and EA, 2005).
- CSM requires radar data to capture the space/time structure of rainfall
- Physical parameters based on rainfall-runoff model need to simulate impacts of land use/management practices
- Regionalized rainfall and rainfall-runoff models for implementation on ungauged catchments is compulsory (Kilsby, 2011).
Taking into account the limitations, continuous simulation modelling may appear to be the most powerful means of measuring flood frequency from rainfall. This type of analysis is getting increasing interest and applies in the USA (ASCE, 1997). Also continuous simulation may form the basis for the next generation of flood frequency estimation in the UK (Culver and Lamb, 2000).
2.5 Overview of the Models
The models that have been used to apply the continuous simulation approach to the Wansbeck are shown below.
Abbildung in dieser Leseprobe nicht enthalten
2.6 Forecast ensemble rainfall data
The town of Morpeth, Northumberland, was affected by a hard flood on the 6th September 2008. Moderate intensity but persistent rainfall from a slow-moving occluded frontal system over the Wansbeck catchment created a severe flood. The evidence shows that some regions of the town were affected by localized pluvial flooding from two main tributaries of the Wansbeck (the Font and Upper Wansbeck) and from poor urban drainage as well as by the main source of flooding from the river. This study has analyzed the use of the Meteorological Office’s 1km high resolution 24 ensemble forecast system for generation of rainfall data in the context of localized pluvial flood prediction. The results show that the ensemble forecast rainfall data generally indicated spatial and temporal rainfall characteristics across different scales reasonably well. On the other hand it generally under-predicted longer duration totals. The ensemble rainfall data is used with a rainfall-runoff model and a 1D river model to generate peak flows and peak levels for the Wansbeck catchment (Parkin et. al., 2011).Actually, operational and research flood forecasting systems around the world are moving toward using ensembles of numerical weather prediction (NWP), known as ensemble prediction systems (EPS), more than single deterministic forecasts, to obtain flood forecasting systems (Cloke and Pappenberger, 2009). For medium term forecasts, NWP models should be used, especially when upstream river discharge data is not available or when the equipment or transmission of data fails, as is often the case in extreme floods (such as 2008 Morpeth flood).
2.6.1 Analysis of space-time rainfall characteristics
Comparison of the modeled ensemble data against observed rainfall was done using an estimation of maximum rainfall amounts from the observed data for the rain gauges and from the 1 km grid ensemble data at the corresponding locations, for durations from 15 minutes up to 2 days for the observed data, and 15 minute up to 1 day for each of the 24 ensemble members (as only 30 hours of data were available). The general spatial patterns of rainfall during the event were surveyed by interpolation of the rainfall totals across the area for each of the two days. A direct comparison against the ensemble data was impossible, as these were only available for the 30 hour period from 12:00 am on the 5th September to 6:00 pm on the 6th September, so the rainfall totals before and after 9:00 am on the 6th September were estimated to provide an approximate indication of the modeled change in the spatial rainfall pattern over the period of the storm. Finally, the ensemble rainfall data is used for SHETRAN as input to generate a rainfall-runoff.
2.6.2 Advantages and Disadvantages of Ensemble Rainfall Data
The atmosphere is a non- linear and complex system and it is not possible to predict its exact state. Weather forecasts are limited by not only the numerical representation of the physical process, but also by the resolution of the simulated atmospheric dynamics and the sensitivity of the solutions to the pattern of initial conditions and sub- grid parameterization. So ensemble forecasting techniques have been applied to take account of these uncertainties and result in multiple weather predictions for the same location and time ( Xuan et. al., 2009). This causes EPS to forecast an attractive product for flood forecasting systems with the potential to extend lead time and better quantify predictability. In addition, the existing operation forecast models predicting intensive localized rainfall have not until recently had sufficiently fine grid resolution to explicitly show individual convective storms or storm clusters (Schaake et. al., 2007). Nevertheless, ensemble rainfall data will make a high resolution forecast model compensate for the limitations. Also ensemble of the weather forecasts can be constructed from forecasts from many several forecast centers.
The disadvantages of ensemble flood forecasting models include:
1- The need for improvement, while the impact of these improvements on hydrological models is uncertain.
2- One of the biggest challenges in the models is to increase the resolution and identify the adequate physical representation on the respected scale.
3- Before applying EPS, learning how to use this in an operational setting. A difficult period of time will be required in order to build up the knowhow of practitioners making new operational flood forecasting tools.
4- Having enough computer power for running operational systems. The only possible solution is to keep improving computing resources wherever possible or to use them more efficiently such as by using clusters of inputs or model factors as a compromise to the full EPS cascade (Cloke and Pappenberger, 2009).
5- Ensemble rainfall forecasts available are not extensive enough to fully verify product performance (Schellekens et. al., 2011).
In summary, EPS forecasts can readily be applied as inputs to medium term flood forecasting systems, while it is clear that these precipitation predictions still need significant improved.
2.7 SHETRAN
The rainfall runoff model that has been used during this project is the physically based spatially distributed rainfall runoff model SHETRAN. SHETRAN shows all of the component processes within a hydrological cycle: snow melt, interception, evapotranspiration as well as surface and subsurface flows in the unsaturated and saturated zone (O Connell, 2000). In addition SHETRAN uses a 2D simulation of overland flow velocity field to estimate erosion and sediment transport. It also uses 3D simulation for representing topography, vegetation, soil and geology of the catchment (Bathurst, 2010).
SHETRAN can model flood generation mechanism such as snow melt in a range of catchments. The version of SHETRAN applied in this study uses catchment rainfall as input and makes outputs of flow (every 15 minutes) for every point in the SHETRAN grid.
One of the main features of the SHETRAN model is fully distributed in the spatial scale since the catchment area is shown by an orthogonal grid network in the horizontal direction with a column of horizontal layers underlining each grid square in the vertical direction. A cell size of 100m × 100m for detailed simulations may be used, on the other hand a 1km × 1km grid may be necessary for broad catchment simulations(Ewen et. al., 2006).Three main elements lie at the core of SHETRAN, one each for water flow, sediment transport and solute transport. It is assumed that flow is not affected by transport. Therefore the three components lie in a natural hierarchy. The constituents model physical processes (Table 2.1) are represented by physically based spatial distributed (PBSD) equations, most of which are partial differential equations and extensive data sets are needed for model parametrization (Ewen et. al., 2000).
Abbildung in dieser Leseprobe nicht enthalten
Table 2.1- Main Processes Represented in SHETRAN
2.7.1 Limitations of Rainfall Runoff Models
SHETRAN is a powerful tool for studying rainfall-runoff for river catchments, including hydrological and environmental impacts associated with land-use and climate change. However, physical process simulation models such as SHETRAN, have different limitations. These limitations are related mostly to the increased complexity of the physical process models. At first, the models of large catchments are computationally extremely “heavy”. In addition, in physical process simulation models, it is difficult and takes time to set up for new catchments .A Graphical User Interface (GUI) tries to improve this which allows SHETRAN to be set up quickly, even by non-expert users (Stephen et. al., 2009).
Another limitation is that data requirements are more extensive because of increased complexities. On the other hand, in some ways data requirements are simplified for physical process models in that the requisite data are more easily calculated and identified because of the physical process basis. The complexity of physical process models needs a broad knowledge of erosion and sedimentation, and watershed knowledge in general. Without enough background, it is uncertain whether physical process models can be applied to achieve accurate results (Professor Basson, 2010). Furthermore, all PBSD modeling has a scale problem. "The essence of the problem is that the values obtained by measuring a physical property (e.g., the saturated hydrolic conductivity) at several points in a region usually cannot simply by "averaged" to get a single value that properly reflects the physics of the relevant process viewed at scale of region"(Ewen et. al., 2000). SHETRAN uses de Saint Vanant equations for unsteady flow. The equations include some assumptions which may be violated during the time (Kutija, 2010).The main limitations in the physical process simulation models are, that the soils should be indicated as vertically uniform and the sediment and solute transport modules are switched off (Parkin, 1995). So SHETRAN does not have the capability to handle variable grids.
In addition, preferential flow through the unsaturated zone is not modeled in SHETRAN, yet it is known to be a vital process for the movement of water, sediments, and solute in the subsurface, and several methods for modelling preferential flow have been available for several years (Ewen et. al., 2000). However, SHETRAN cannot produce any base flow for situations where groundwater is below stream channels or separate base flow from interflow, perched aquifer or spring conditions (http://www.dwaf.gov.za/Geohydrology/). It is clear that preferential flow is important, and a capability to model preferential flow must be added to SHETRAN. The problem faced is choosing the right approach, because soil physics has not yet come to a conclusion about how best to model preferential flow (Ewen et. al., 2000). Despite these limitations, unsteady river models give a detailed assessment of flood risk that supports the approach of continuous simulation.
2.8 NOAH 1D
The river model that has been used during this project is the unsteady river model NOAH 1D. The NOAH 1D modelling system is the first in a range of highly innovative NOAH hydroinformatic tools. NOAH 1D is an advanced modelling system for hydraulic networks under unsteady, predominantly free surface flow conditions. NOAH 1D incorporates the latest development in information technology, best numerical algorithms and, for the first time in hydroinformatic tools, object oriented numeric (http://research.ncl.ac.uk/noah/images).
NOAH 1D is computationally very efficient and is also very easy to use and extend. In addition, as has already been mentioned the cooperation of some unusual features is made at minimal cost ( Kutija and Murray, 2000). NOAH 1D is based on the full de saint venant equations expressing one- dimensional free surface flow in each link(edge) of network and the Where bs , is top width or storage width (m), h is the water depth (m), t is time (s), Q is discharge (m3 /s), x is the distance along the channel (m), q is lateral inflow (m3 /s/m), p is the Boussinesq coefficient, A is cross-sectional area (m2 ), g is the gravitational constant (m/s2 ), iQ is bottom slope and K is conveyance (m3 /s).
Abbildung in dieser Leseprobe nicht enthalten
In NOAH 1D the differential equations are solved using the implicit Abbot- Ionescu finite difference scheme on a ‘staggered grid’ and a Generalized Elimination Algorithm (GEA). A highly efficient direct solution method for nearly banded matrices is used to solve the resulting difference equations (Cunge et al., 1980).
NOAH 1D allows a variety of computational speeds to be applied. When interested in tracing some event, one can run just one time step at a time or select some intermediate speed. All the runs, are integrated with the real time graph of the values of discharge and water levels at selected points. This is produced by the object oriented structure of the numerical algorithm that permits interaction with the individual objects. The same character permits setting up watches on the selected objects and record when their main property (discharge or water level) would get maximum value ( Kutija and Murray, 2003).
Another feature facilitated by the object-oriented structure is the capability to add a variety of cross sectional types with the minimum change in the existing code. A special result of this is NOAH’s ability to deal with sewer systems as well as with receiving water at the same time. So, NOAH’s computational efficiency together with the extensive batch facilities provide good simulations (Kutija and Murray, 2003).
2.8.1 Manning ’s n
Today, n is most often applied as a required input value by every user of one- dimensional (1D) modelling software such as NOAH 1D because Manning’s equation is the standard numerical solution to relating water levels, velocity and river bed characteristics together in the hydraulic models central to the highly political business of flood risk prediction and management (Whatmore and Landstrom, 2003).
Measurement of surface roughness is always difficult; so sensitivity analysis is performed in this study in order to evaluate the degree of effectiveness of the assigned roughness coefficient values for the river channel and floodplain. Usual roughness coefficient values for category of rivers expressed as clean, windy, some pools, some shoals, weeds and stone is 0.045 and for category of rivers described as sluggish reaches, weedy, or floodway’s is 0.7(Chow, 1959).
Overall, a sensitivity test of Manning’s n can aid to ascertain whether vegetation changes are likely to influence model predictions. Therefore, the final calibration values of a model partly compensate for uncertainties in other parts of the model (Arcement et al., 2011).
2.8.2 Limitations of the 1D River Model
The benefits of 1D models are that they are relatively easy to develop and they are much faster to run. Also one- dimensional models have excellent results for many tidal or river flow conditions provided that the 1D modelling assumptions are not violated. These assumptions are:
1. Flow perpendicular to the entire cross section.
2. Level water surface across the entire cross section.
3. Discharge is distributed within a cross section based on the conveyance distribution.
4. The energy slope is uniform across the entire cross section.
Therefore if hydraulic conditions become more complex, the 1D model assumptions will be violated and maybe 1D model cannot use (http://www.fhwa.dot.gov).
Furthermore, many river models cannot simulate small flow depths or dry beds during low flows. To compensate for this, the river model may produce an addition of water to the model to balk the river bed from drying out. But where the model adds too much water then the calibration may not be compromised (Cunge et al., 1980).
2.9 Previous study: Reconstruction of the Dynamics of a Flood Event Based on Publicly Sourced Information
The Parkin report considered reconstructing the water levels and flood extents of the 6th September 2008 Morpeth flood in as much spatial and temporal detail as possible based on all available documents such as photographs. Then, they applied it with other data including a Digital Terrain Model (DTM) derived from Lidar data, river cross section, and other qualitative information to reconstruct water depths and extents for a series of snapshots at 1hour intervals during the rising limb of the flood. The resulting data set provided quantitative information on dynamic water levels that were used to test new generation 2D flood plain hydraulic models and to understand interactions between fluvial and pluvial sources of flooding. The model extended from the Mitford road to below the town as shown in Figure 2.1. Hence, the data was used with other qualitative information to give evidence of hydraulic details of flood development, including control of water levels by built floodplain structures, rapid catastrophic overtopping of flood defences and description of flood response between pluvial and fluvial sources at different times within the same event.
Abbildung in dieser Leseprobe nicht enthalten
Figure 2.1 - Areas of Morpeth severely flooded on the 6th September 2008(Parkin, 2010)
2.10 Previous study: Continuous Simulation Approach for Flood Risk Estimation
The Ward report of 2008 considered that the main hypothesis of Continuous Simulation was that the FEH was unable to make predictions of change into the future and does not therefore offer a method to measure the possible impacts of climate and land use change. As a result the FEH focuses on the storm hydrograph for a given return period; it does not completely consider the effect of antecedent catchment conditions.
So, to overcome the limitations of the FEH, new methodology of Continuous Simulation, approach to flood risk estimation, with the Wansbeck catchment surveyed. The approach simulated 100 years of rainfall data and then modelled the rainfall in a calibrated rainfall-runoff model (SHETRAN) to generate 100 years of hourly flow. The method made the assumption that the one hundred year event was contained within the outputs. An unsteady 1D river model was used to simulate river levels for the most significant event. Also a conceptual model of light water storage area to simulate flows for a number of maximum outflow rates using the design event as shown in Figure 2.2. Then the value of peak flows at Mitford with and without light water storage was compared (Table 2.2).
Abbildung in dieser Leseprobe nicht enthalten
Figure 2.2 - 100 year flows at Mitford with and without Lightwater storage
Abbildung in dieser Leseprobe nicht enthalten
Table 2.2 - Peak flows at Mitford for various outflow rates from Lightwater
2.11 Study Area Description
Morpeth is an ancient market town located in a loop of the river Wansbeck in the northeast of England, about 15 miles north of Newcastle Upon Tyne and 12 miles West from the North sea. Morpeth lies in the Wansbeck catchment area, a relatively small river catchment which covers 331 km2 . The main reach of the Wansbeck has an active flood plain that is between 100m to 300m wide and the town itself is situated within this floodplain. The river has three tributaries: the Font, which drains the Simonside Hills and includes a local water supply reservoir (Fonthurn); the Hurt Burn, a catchment of similar dimensions but shallower in gradient, and the upper Wansbeck itself. The three rivers before flowing through the town of Morpeth, combine within a 10km reach just West of the i41trunk road. In Morpeth itself, the Wansbeck is joined by different small tributaries, including Cotting Burn, Charch Burn, and Postern Burn, all with catchments of less than 5 km2 (http://www.microdis-eu.be/content/morpeth-united-kingdom).
Previous significant flooding events happened in 1863, 1876, 1903, 1924, 1963 and 1968. Following the 1963 flood, a flood defence scheme was instituted. Flood walls were established on the north bank (Environmental Agency, 2007). When the Wansbeck submerged on the 6th and 7th September 2008, the flood water simply flooded over the top of the defences that were not high enough to hold back the volume of water.
This image was removed by the editorial team due to copyright issues.
Abbildung in dieser Leseprobe nicht enthalten
Figure 2.3 - Morpeth, Northumberland and location in UK (http://www.welcomenorthumberland.co.uk/map_morpeth.php)
This image was removed by the editorial team due to copyright issues.
Abbildung in dieser Leseprobe nicht enthalten
Figure 2.4 - Morpeth before and after the 2008 flooding (http://www.coolgeography.co.uk/GCSE/AQA/Water)
2.12 Flood risk characteristics – Wansbeck catchment
2.12.1 Rainfall
Currently floods are measured to have been a 1 in 115 year event. Prolonged rainfall overlapped with the flood peak from higher areas of the catchment and then caused a peak water level of 3.99 m in the river channel. It was the biggest flow recorded in the Wansbeck. The Environmental Agency recorded 150mm of rainfall in the Wansbeck catchment on the 5th and 6th of September as shown in Figure 2.5. This figure suggests that in Morpeth alone there was 86mm of precipitation compared to an average of 74mm for a month. The big volume of water led to the drainage system to back up, plus Oldgate Bridge obstructed the flow (www.environment-agency.gov.uk).
2.12.2 Physical Factors
The river Wansbeck valley is narrow and steep due to more surface runoff. According to storm events, a reliable flow gauge located upstream of Morpeth at the confluence of the Wansbeck and the Font representing 56% of rainfall is converted into surface runoff. Also because of the wet summer, the soil was already saturated and caused increase surface runoff. In addition, there has been increased urbanisation since the 1960s in Morpeth and due to this a huge volume of water falling on the town would have drained directly to the river channel. Other estimates of the catchment lag time (time lapse between the midpoint of storm rainfall and peak river level) shows that the Wansbeck has a lag time of only 8 hours. Therefore, any water falling in the catchment area would have been rapidly converted into channel flow by surface runoff and to a lesser extent by the flow. This resulted in the steepness of the valley and the soil composition (JBA, 2008).
This image was removed by the editorial team due to copyright issues.
Abbildung in dieser Leseprobe nicht enthalten
Figure 2.6 - Exceed of surface runoff during 6 September 2008, Morpeth (http://www.bbc.co.uk/tyne/content/images/2008/12/23/floods)
2.12.3 Impact of Flooding
More than 400 residents were affected by flooding on the 6th September 2008. A huge number of residents had to be forced from their homes, and lived in car vans or had to rebuild some places. When the river Wansbeck burst banks on the 6th September 2008, more than 1000 houses were affected. 995 properties in Morpeth centre were affected by the flood water directly. Estimations show that damage could exceed £10 million. Morpeth High Street (Bridge Street) was under 60 centimetres of water, at the peak of the flood. The main street had not been flooded since 1963. The library suffered structural damages that caused heavy debris to be transported by the river. Such was the extent of the damage that structural engineers were needed in order to test it’s safety (Maureen Fordham, 2008).
This image was removed by the editorial team due to copyright issues.
Figure 2.7 - Impacts of 2008 flooding- Morpeth (http://www.bbc.co.uk/tyne/content/images/2008/12/23/floods)
Abbildung in dieser Leseprobe nicht enthalten
2.13 Existing Defences
There are different flood defences in Morpeth that are summarised in Table 2.3.
Abbildung in dieser Leseprobe nicht enthalten
Table 2.3 - Existing flood defences in Morpeth
This image was removed by the editorial team due to copyright issues.
Figure 2.8 - Flood cells in Morpeth (http://www.coolgeography.co.uk/GCSE/AQA/)
2.14 Future Management Proposed Flood Defences for Morpeth
Since November 2007 the Environment Agency has been working to improve the existing protection against flooding in Morpeth. So the Environment Agency has obtained two broad options for a flood alleviation scheme that are technically, environmentally and economically viable. These are:
- To build flood defences where none presently exist, and refurbish or raise existing walls
- To build one or more floodwater storage areas upstream of Morpeth, together with new and refurbished flood walls in the town
By creating flood water storage areas upstream, it would mean the defences in the town would not need to be as high. Also the Environment Agency has identified three possible sites upstream (Figure 2.9) to store water during flood conditions. These are:
- Upstream of Mitford Hall on the Mitford Estate
- Upstream of Rivergreen Mill
- Upstream of Meldon Park
- Upstream of Hart Burn confluences
- Upstream of Font Ford Lodge
- Upstream of Woodhouse
Floodwater storage options would be complemented by improvements to culverts in Morpeth. Increasing culvert capacity should reduce the risk of flooding from Cotting Burn, Church Burn and Postern Burn (Environment Agency, 2009). Possible areas are on the Upper Wansbeck and the Font, and the addition of an upstream storage pond will be considered as part of the continuous simulation approach. The method will compare the peak levels with and without upstream storage for the flood event. In this research, the upstream storage location that has been chosen is the same as that investigated in the Environment Agency (2009) report, which is the upstream of Mitford Hall storage pond on the Upper Wansbeck and upstream of the Font Ford Lodge storage pond on the Font. This could reduce the risk of flooding between 2% (1 in 50 year standard of protection) and 0.3% (1 in 300 year standard of protection) chance of occurring in any year. The possible storage areas have been assessed using NOAH data. This study has obtained the storage areas from the NOAH data over a series of elevations. These are then used to make an elevation-storage relationship in order to estimate the volume of storage available.
This image was removed by the editorial team due to copyright issues.
Abbildung in dieser Leseprobe nicht enthalten
Figure 2.9 - Location of potential storage areas in the Wansbeck catchment (http://maps.google.co.uk/)
3 Materials and Methods
1km high resolution 24 ensemble forecast system for generation of rainfall data (from Meteorological Office) was used during the event (Morpeth 2008 flood). Hence, a SHETRAN model was made and calibrated using the ensemble rainfall data. The peak flow from SHETRAN was then used as an input to the NOAH 1D model to get the water level in each section. Two simulations of the NOAH model were then run. One of them, the situation of the catchment during the event without storage was modeled and the other was the situation with storage modeled. An overview of the project methodology is given in Figure 3.2.
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.1- Wansbeck catchment and the tributaries (Ward, 2008)
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.2 - Overview of project methodology
3.1 Data Requirements
The data requirements for the project can be categorized as they are shown in Table 3.1. The data was applied to commence the model preparation, calibration and verification.
Abbildung in dieser Leseprobe nicht enthalten
Table 3.1- Data required to undertake the project
3.2 Model preparation and processing Introduction
Following Figure 3.2 , to approach the flood model, at first 24 ensemble rainfall data from 20th July to 8th September 2008 (which included the 2008 flood for every 5 minutes) was used to represent catchment rainfall (over the area of each tributary) and to determine spatial variation in the catchment rainfall. This data set was then put into the calibrated SHETRAN model. Also to run the SHETRAN, some others map and data were used including:
- 1km raster grid for the digital elevation model
- 1km raster grid for land use
- 1km raster grid for soil type
- 10km raster grid for the catchment area
- The number of accumulated grid squares to make the river
- Using calibrated parameters (at Mitford) in SHETRAN such as Strickeler overland flow coefficient, soil depth and the hydraulic conductivity of the most common soil type
- Using observed flow data at Mitford during the event
After running the SHETRAN, every 15 minutes, flow data at the Font, Upper Wansbeck (two main tributaries) and Mitford (downstream of the Wansbeck catchment) was produced. Next, the largest flow data was put into the calibrated NOAH 1D. The procedure to run the NOAH 1D is the same as that investigated in the Ward (2008) report which includes:
1. Steady state phase; open channel sections were modeled in NOAH 1D as IRREGULAR.
2. Unsteady phase; weir sections were put into NOAH 1D.
3. Bridge sections were loaded into NOAH 1D.
4. Oldgate Bridge in the town of Morpeth was modeled as a closed section with a spill.
5. Cross sections were extended onto the floodplain to alleviate glass walling in the model.
6. Levees were infilled. NOAH 1D cannot model ineffective flow areas.
7. Sections were changed to STORAGE sections and values of Manning’s were obtained to floodplain parts of each section (Ward, 2008).
In this project, calibration and validation for NOAH 1D is the same as that investigated in the Ward (2008) report including:
- The model was calibrated using flow and level data recorded during the 1992 flood
- The model was validated using flow and level data recorded during the 2000 flood, 2007 flood and 2008 flood
3.3 Model Linking and Application
Firstly, a 24 km model for the rainfall, an outline of generation of 24 members ensemble probabilistic simulations was produced every 5 minutes (between 20/07/2008 and 08/09/2008). The rainfall was then used as input the calibrated SHETRAN model to approach flows every 15 minutes in each tributary. Each ensemble rainfall data contains 275 columns (corresponding to the 275 grid squares in the catchment). For each column there is a time series of data from 20/7/2008 to 8/9/2008. After running the SHETRAN model, 24 flow data for every 15 minutes was generated but in this study just maximum, medium and minimum hydrographs were used. Then the largest flood simulated in the calibrated SHETRAN model was transferred to the NOAH 1D model. At first the NOAH 1D model was run without lateral inflow but after calculation lateral inflow, the model was run again to represent additional flows from overland flow, interflow and groundwater flow. Hence a conceptual model for the storage was made in Upper Wansbeck and Font Ford Lodge. Finally, peak flows and peak levels at Mitford were compared before and after the storage was built. Overview of Model linking and Application is outlined in Table 3.2.
Abbildung in dieser Leseprobe nicht enthalten
Table 3.2- Overview of Model linking & Application
3.3.1 Wansbeck Rainfall During the Event
According to The Climatological Observers Link, August 2008, rainfall in July and August 2008 was significantly more than the long-term monthly average recorded in the northeast and then, soils were wet at the start of the storm which caused flooding in September. A frontal weather system moved across Wales and central England from the early hours of September 5th bringing heavy rainfall and flood in parts of the northeast, including the Wansbeck catchment. This wide rain band reached the Wansbeck catchment early in the afternoon of the 5th and rainfall continued during the night. The low pressure system with northeast winds circulating round the northern margin of the system, moved into the southern North Sea. On the 6th September most other parts of England were experiencing only sporadic heavy rainfall. The band of rainfall came in lengthwise towards the Northumbrian coast, therefore some coastal areas including the lower Wansbeck catchment got rainfall from the full length of the front. Heavy lowland rainfall coincided with the arrival of the flood peak from the upper part of the catchment caused by the overnight rainfall, therefore, further exacerbating the flooding. Rainfall totals in the area were very high with between 200% and 300% of the September average falling between the 4th and 6th in parts of Northumberland including Morpeth(JBA, 2009). Table 3.3 shows rainfall statistics for a selection of weather stations in Northumberland (JBA, 2009).
Abbildung in dieser Leseprobe nicht enthalten
Table 3.3- Rainfall statistics 4-6th September 2008
Numerical Weather Prediction (NWP) model forecasts, even if skilful, have not had enough resolution to provide information on the scale required by hydrological models. So, in this project with horizontal and vertical high resolution of NWP model forecasts, river flow for this particular event was surveyed (Roberts et. al., 2009). This study has used the Meteorological Office's 1km high resolution 24-ensemble forecast system to produce rainfall data in the context of localized pluvial flood predictions. The ensembles consist of one control run and 23 additional members. Figure 3.3 shows comparison observed data and average forecast ensembles rainfall data during the event (30 hours). In addition the observed hourly rainfall time- series for the September 2008 flood event from the available raingauges (Figure 3.5) in or near the Wansbeck catchment is show in Figure 3.4. Also, the spatial distributions of rainfall data are shown in Figure 3.6, which indicates interpolation of the daily rainfall totals across the area for 9am on the 5th September to 9am on the 6th September(Figure A) and 9am on the 5th September to 9am on the 6th September (Figure B). These indicate how on the first day the heavier rainfall was concentrated in the north-west of the Wansbeck catchment. On the other hand, on the second day the heaviest rainfall was concentrated towards the south-east of the catchment (Parkin et. al., 2011).
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.3- Compare observed data and average forecast ensembles rainfall data during the event (30 hours) in Wansbeck catchment
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.4- 48 hour total rainfall over Northumberland for 0:00 5 September to 24:00 6 September 2008, showing the Wansbeck catchment to the Mitford gauging station (station 22007)
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.5-Location of rain gauges around the Wansbeck catchment (Ward, 2008)
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.6-24 hour rainfall totals for A) 5 Sept 2008 (9am 5th to 9am 6th) and B) 6 Sept 2008 (9am 6th to 9am 7th) interpolated from measured daily rain gauge totals
3.3.2 SHETRAN Model Development
The SHETRAN model was built using a 1km2 map for elevation (DEM), vegetation and soil type as well as Ward used in 2008 (for more detail look at Table 3.4). Also 24 forecasting ensemble rainfall data (between 20/07/2008 and 06/09/2008) was used as inputs, to produce simulated flow at two tributaries (the Font and Upper Wansbeck) and the downstream of the Wansbeck catchment (Mitford). Ward (2008) calibrated the SHETRAN model by changing the parameters for Strickler overland flow, soil depth and the hydraulic conductivity of the predominant soil type. Also, calibrations were achieved by using the catchment rainfall inputs and by comparing observed and modelled flows at Mitford (between 1999 and 2000). In this study the previous calibration model was used because SHETRAN was not been calibrated for the big event.
Abbildung in dieser Leseprobe nicht enthalten
Table 3.4- The maps that were used as inputs for SHETRAN (Ward, 2008)
3.3.2.1 Elevation, Land use and Soil type maps
A digital elevation model (DEM) of 25 m had been obtained from the Landmap Archive (www. Landmap.ac.uk). But for using the DEM in SHETRAN, the data had been converted to a 1km2 dataset by Arc GIS that is shown in Figure 3.7.
In addition, the “Land Cover Map 1990” from the Centre of Ecology and Hydrology with 1km resolution was used for the dataset in the SHETRAN model (Ward, 2008). Because the maximum number of land use in SHETRAN is seven, so the 25 vegetation classes in Land Cover 1990 were divided into a sub-group shown in Table 3.5.
Abbildung in dieser Leseprobe nicht enthalten
Table 3.5- Categories of land use types for use in SHETRAN (Ward,2008)
For identification of soil types for the Wansbeck, the Soil Site Reporter of National Soil Resources Institute corresponding to superficial geology locations was used (Ward, 2008). The four main soil types in and around the Wansbeck catchment were obtained and linked to the spatial location of superficial geology in the 1km raster dataset (Table 3.6).
Abbildung in dieser Leseprobe nicht enthalten
Table 3.6- Identification of soil types , using the Soil Site Reporter for Wansbeck catchment (Ward, 2008)
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.7- 1km DEM map for the catchment area (Ward, 2008)
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.8- 1km land use map for the catchment boundary at Bothal Mill (Ward, 2008)
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.9- 625km superficial geology map for the catchment (Ward, 2008)
3.3.3 Calibration of the SHETRAN model
In this project calibration of the SHETRAN model, is the same as calibration of Ward's report in 2008. As there was no SHETRAN calibration for the big event the SHETRAN model of the Wansbeck catchment was calibrated for the period from the 1st October 1999 to the 1st October 2002 as shown in Figure 3.10. This figure shows that there is a good fit between measured discharge and simulated discharge. On the other hand there are small differences between them which may be due to rainfall and flow measurement errors. The calibration was achieved by changing the parameters for hydraulic conductivity of the predominant soil (Dunkeswick), soil depth and Strickler overland flow as shown in Table 3.7.
Abbildung in dieser Leseprobe nicht enthalten
Table 3.7- Final parameter values for calibration of SHETRAN model (Ward, 2008)
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.10- SHETRAN calibration with time series between 1999 and 2000
3.3.4 Continuous Simulation for 2008 flood
The calibrated model was used to generate total runoff (for the Font, Upper Wansbeck and Mitford) for the period 20/07/2010 to 08/09/2010 using rainfall from each of the 24 ensemble members with 1km resolution and also using observed rainfall data from the daily rain gauges in Morpeth, disaggregated using the hourly rainfall records from the rain gauges near or in the Wansbeck catchment (Figure 3.11, Figure 3.13 and Figure 3.15). The simulation period began on the 20/07/2008, to ensure appropriate antecedent soil moisture conditions. Also, the parameter for “Strickler coefficient” was changed to 1000 to make the channel smooth with higher coefficient. As with previous coefficients, tributaries flow during some periods of time (the Font and Upper Wansbeck in upstream) simulated higher than flows in Mitford (in downstream). In addition the simulated peak flow was compared with the observed flow at Mitford (Table 3.8). The simulated peak flow is substantially higher than the observed flow (especially for flow simulated by maximum and medium ensemble rainfall). So the range of simulated hydrographs (for the tributaries) during the flood for the Wansbeck catchment were produced, but in this research just hydrographs simulated by maximum, medium and minimum ensemble rainfall were used which are represented in Figure 3.12, Figure 3.14 and Figure 3.16.
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.11- SHETRAN hydrographs from all 24 ensemble rainfall at the Font
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.12- SHETRAN hydrographs from maximum, medium and minimum ensemble rainfall at the Font
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.13- SHETRAN hydrographs from all 24 ensemble rainfall at Upper Wansbeck
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.14- SHETRAN hydrographs from maximum, medium and minimum ensemble rainfall at Upper Wansbeck
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.15- SHETRAN hydrographs from all 24 ensemble rainfall at Mitford
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.16- SHETRAN hydrographs from maximum, medium and minimum ensemble rainfall at Mitford
Abbildung in dieser Leseprobe nicht enthalten
Table 3.8- SHETRAN peak flow simulated by maximum, medium and minimum ensemble rainfall for Wansbeck tributaries
3.3.5 NOAH 1D model development
The NOAH 1D model was built using surveyed cross sections that were supplied by Ward's report in 2008(Figure 3.17). The cross sections were made from Bothal Mill Bridge at the downstream end of the lower Wansbeck, through the town of Morpeth and up in to the tributaries of the Font and Upper Wansbeck. The available data for use in the NOAH 1D model were obtained such as the cross section data and flow data during the event in the catchment. The overall schematization of the Wansbeck catchment river model is represented in Figure 3.18. The figure shows that the Wansbeck river model was made up for the two main tributaries (the Font and Upper Wansbeck) that then converge around Mitford at the start of Edge WANS06. The river Wansbeck was schematized so that the Edges corresponded to the river flow reaches. Also the chainage of these Edges was given by using the cross section locations in AutoCAD (Table 3.9).
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.17- Schematization of cross sections along the Wansbeck and the location of the study area
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.18- Image of NOAH 1D model (Ward, 2008)
Abbildung in dieser Leseprobe nicht enthalten
Table 3.9- Description of cross section reaches and chainages
The procedure to make NOAH 1D model for the Wansbeck catchment is shown in table 3.10 (for more detail look at Appendix).
Abbildung in dieser Leseprobe nicht enthalten
Table 3.10- Procedure to approach NOAH 1D model for Wansbeck
3.3.6 Calibration and Validation of NOAH 1D Model
Calibration for the NOAH 1D model was achieved by changing the Manning's roughness parameter for the in-channel sections. The final calibrated values are represented in Table 3.10. In a previous study of the Wansbeck catchment, the model was calibrated for the 1992 event using flows and levels at Mitford gauging station (Figure 3.19 and Figure 3.20). The figures show the model overestimates the peak flow and peak level. Because maybe the NOAH 1D model does not consider the effect of bridge blockage which has been observed in the event (Ward, 2008).
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.20- NOAH 1D observed and modelled levels during 1992 flood (Ward, 2008)
Also in this study observed flows and water levels (using Mitford gauging station) are compared with modelled flows and water levels for the 2008 flood that are represented in Figure 3.21 and Figure 3.22. Figure 3.21 shows differences between observed and modeled flows from different ensemble rainfall. Peak flows modelled by maximum ensemble rainfall are higher than observed flows. It has been observed in the data obtained during the event that debris has been accumulated around the bridges, but not in NOAH modeled. Also, on the recession limb, observed peak flow is greater than simulated peak flows. This would suggest that the accumulation of debris was less during this time. In addition peak flows simulated by medium and minimum ensemble rainfall lower than observed flow, may be due to flow measurement errors. Overall, there is the best fit for peak flows simulated by medium ensemble rainfall and observed flows (modelled peak flow is 321(m3 /s) and observed peak flow is 357(m3 /s).
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.22- NAOH 1D observed and modelled levels during 2008 flood
The main aim of the validation test is to assess the sensitivity of the model to parameter values, and its ability to show new and unseen data. In a previous study, flood events in 2000 and 2007 were used for validation of the model for the Wansbeck. The flood event in 2000 had a peak flow of 194(m3 /s) and is therefore within the calibrated range of 204(m3 /s) established for the 1992 event. Figure 3.23 and 3.24 indicate that the NOAH 1D model underestimates the flow and river level due to the debris accumulation as being less significant than the 2000 flood event. This may be obtained as a result of the lack of debris as suggested in the flow results. It may be that the calibrated Manning's n values show the debris accumulation of the 1992 event, when the conveyance of the channel was lower (Ward, 2008).
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.23- NOAH 1D observed and modelled flows during 2000 flood
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.24- NOAH 1D observed and modelled levels during 2000 flood
Also the validation results for the 2007 event represent that overall there is a good agreement between the observed and modelled levels at Mitford gauging station (Figure 3.25), which mentions that the model is able to respond to new events well (e.g. flood event in 2008). The results for the 2007 event include a sensitivity test with results of different calibration runs giving similar levels. In addition, validation results for the 2008 flood for different hydrographs are shown in Figure 3.26.
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.25- NOAH 1D observed and modelled levels at Mitford during 2007 (Ward, 2008)
Abbildung in dieser Leseprobe nicht enthalten
Figure 3.26- NOAH 1D observed and modelled levels at Mitford during 2008
To summarise, the methodology to date has calibrated and validated a catchment rainfall runoff model and hydraulic river model for Wansbeck catchment. Thus, the results confirm that the model performs good agreement between observed and modelled peak flows and peak levels at Mitford gauging station during several events. The following section will initially consider the model linking and application for the 2008 flood. This is followed by a simulation of peak flows and peak levels (using NOAH 1D) with lateral inflow from the largest peak flow data of SHETRAN. The event has then been used to consider flood volume in relation to the use of upstream storage areas near Mitford Hall and Font Ford Lodge.
4 MODEL LINKING and APPLICATION
4.1 Design 2008 Flood without Storage
At first in order to define the linking of the SHETRAN and NOAH 1D models, the flow simulated (from the Font and Upper Wansbeck) in SHETRAN during the 2008 flood was modelled in NOAH 1D without measurement of lateral inflow and then with lateral inflow to get flows at Mitford. Flows at Mitford were ratioed to the relevant tributaries and using ensemble rainfall. Linking the SHETRAN model to the NOAH 1D model for the event was obtained to demonstrate the continuous simulation method and to compare the flows from the linked models before upstream storage was considered. The linkage of the two models is represented in Figure 4.1, 4.2 and 4.3, and the peak flows from the model are shown in Table 4.1. Comparison between the modeled SHETRAN peak flows and the peak flows in NOAH 1D shows that flows were underestimated by the NOAH 1D model.
Abbildung in dieser Leseprobe nicht enthalten
Figure 4.1- Observed, SHETRAN and NOAH flows using maximum ensemble rainfall at Mitford during 2008 flood
Abbildung in dieser Leseprobe nicht enthalten
Figure 4.2- Observed, SHETRAN and NOAH flows using medium ensemble rainfall at Mitford during 2008 flood
Abbildung in dieser Leseprobe nicht enthalten
Figure 4.3- Observed, SHETRAN and NOAH flows using medium ensemble rainfall at Mitford during 2008 flood
Abbildung in dieser Leseprobe nicht enthalten
Table 4.1- Peak flow observed and simulated at Mitford for the 2008 flood
Lateral inflow is the addition of water to a stream, river, or lake from the sides of the channel or reservoir. Lateral inflow can calculate ground water flow, overland flow or interflow. The lateral inflow along each Edge was calculated by subtracting the flow at the Edge outflow from the flow at the Edge inflow, and divided by the total length of the Edge as shown in Table 4.2. Also, the values of lateral inflow with different flows that were simulated by maximum, medium and minimum ensemble rainfall are shown in the Appendix.
Abbildung in dieser Leseprobe nicht enthalten
Table 4.2- Total length of each Edge
4.2 Storage near Mitford Hall and Font Ford Lodge
In this project the location of the storage is the same as the Environment Agency obtains (e.g. in Upper Wansbeck near the Mitford Hall and another one in the Font near the Font Ford Lodge) and is shown in Figure 4.4. The exact location was found by using AutoCAD and GIS software. The storage in this study was complemented by improvements to culverts in Upper Wansbeck and the Font. Actually, the storage areas were created by constructing a grassed earth embankment across the river. So, during normal and low flows, the river would pass through the embankment in the culverts with a natural bed. These culverts would be closed during high flows. During periods of high flow, the second 'high flow' culverts would limit the flow through the embankment, with the remainder of the water being impounded in the storage areas. In addition it would provide some form of flow control structure to the high flow culverts such as a moving gate if necessary (Environment Agency, 2009)
This image was removed by the editorial team due to copyright issues.
Abbildung in dieser Leseprobe nicht enthalten
Figure 4.4- Location of proposed upstream storage areas near the Mitford Hall and Font Ford Lodge
In order to model the effect of the upstream storage as the culverts, it was decided to change the Edge of WANS07 (Upper Wansbeck) and FONT 01 (Font) to be wider. Two cross sections were added to the Edge in order to design upstream culverts and downstream culverts in the NOAH 1D model. A series of 7 culverts side by side were designed as BOX RECTANGULAR (3m×3m) sections. So, the capacity of water hold has been increased by the provision of additional culverts capacity. The value of Manning’s coefficient obtained was 0.03 for all culverts. In addition, the Environment Agency obtains that the culverts in Upper Wansbeck will be built through the dam along the line of the existing River Wansbeck. More details are shown in Figure 4.5 and 4.6. Also Figure 4.7 shows the Schematise model of the storage in the Font near the Font Ford Lodge. In addition, because of the high cost of the big storage, another smaller storage (e.g. 3 culverts with the size 1m×0.6m and 35mAOD for the invert) instead of a bigger size was modelled to compare peak flows.
Abbildung in dieser Leseprobe nicht enthalten
Figure 4.5- Schematise location of the cross sections for the culverts at Upper Wansbeck
Abbildung in dieser Leseprobe nicht enthalten
Figure 4.6- Perceptual model of the storage at Upper Wansbeck
Abbildung in dieser Leseprobe nicht enthalten
Figure 4.7- Schematise location of the cross sections for the culverts at the Font
In this model, it was assumed that the inflow for the storage areas was equal to the river flow with maximum discharge. On the other hand the outflow was limited by storing the water. As the inflow rates would be below the maximum discharge rate, the water in storage was allowed to drain away at a limited rate so that the maximum outflow would not be exceeded at any time.
The available storage in the culverts and the cross sections before the culverts was obtained in this project from the NOAH 1D model. This enables a picture of the relationship between the storage areas, and from that available storage at each elevation (as shown in Table 4.3 and Table 4.4). This could then be related back to the water level and thus the dam crest height at Upper Wansbeck (that Environment Agency obtains dam crest height.) Therefore, the appropriate design for the storage area (culverts in Upper Wansbeck) is dependent on the water level in the storage that should be lower than the dam crest height. If the water level in the storage becomes higher than the dam crest height, overtop flow will occur and it will show that the dam crest that the Environment Agency proposed is not high enough. In this study overtop flow does not occur.
Abbildung in dieser Leseprobe nicht enthalten
Table 4.3- Heights and areas for the Upper Wansbeck storage and cross sections before
Abbildung in dieser Leseprobe nicht enthalten
Table 4.4- Heights and areas for the Font storage and cross section before
4.3 Design Event with Storage 4.3.1 NOAH 1D Flows with Storage
In order to simulate the effect of the storage pond, the NOAH 1D model was run with the SHETRAN peak flow produced by maximum, medium and minimum ensemble rainfall again, but this time with the addition of storage as shown in Figure 4.8, 4.9 and 4.10. From the peak flows from each scenario it is possible to estimate the reduction in flow at Mitford and the reduction in flood risk that may result from the inclusion of a storage area near Mitford Hall and Font Ford. In all different scenarios the storage area situated on the River Font would not be significantly effective at reducing the flow downstream in Morpeth. This site reduced flows lower than the site near Mitford Hall. Because flows in Font River are significantly lower than flows in the UpperWansbeck River. In addition with a smaller size of storage the flows reduction is not considered especially for peak flows simulated by maximum ensemble rainfall. Although this project does not calculate or focus on the return period for the peak flow and rainfall, it is still possible to demonstrate an overall trend in reduction of peak flows. It is also easy to assess the performance of the storage in several scenarios (Table 4.5), and the effect on the resulting recession limb as the storage water drains away. In addition, through the continuous simulation approaches the effect of phasing the flood flowing through each tributary was considered. This is undertaken by using the distributed outflows from SHETRAN as input to the NOAH 1D.
Abbildung in dieser Leseprobe nicht enthalten
Table 4.5- Peak flows at Mitford with and without the storage
4.3.2 NOAH 1D Levels with Storage
The storage areas at Upper Wansbeck and the Font have declined the peak levels at Mitford for all the scenarios, as represented in Table 4.6. Furthermore, the use of the 1D model explains the release of stored water not appearing to cause flooding later on as shown in Figures 4.11, 4.12 and 4.13. Also the water volume with and without storage is shown in Table 4.7. River levels continue to fall even as the additional water is released from storage. On the other hand, the NOAH 1D model does not include the effect of debris on bridges. So, in some situations it may be that the blockage of the bridges in Morpeth significantly reduces the conveyance of the channel, especially during the recession limb of hydrographs. In these situations, the release of additional water from storage may be due to water levels increasing relative to the results from the NOAH 1D model. Also, because the channel was changed so as to be wider for making the storage, in rising limb and recession limb. It seems that the water level increased, but peak flows decreased.
Abbildung in dieser Leseprobe nicht enthalten
Figure 4.11- Peak level simulated by maximum ensemble rainfall at Mitford
Abbildung in dieser Leseprobe nicht enthalten
Figure 4.12- Peak level simulated by medium ensemble rainfall at Mitford
Abbildung in dieser Leseprobe nicht enthalten
Figure 4.13- Peak level simulated by minimum ensemble rainfall at Mitford
Abbildung in dieser Leseprobe nicht enthalten
Table 4.6- Peak levels at Mitford with and without the storage
Abbildung in dieser Leseprobe nicht enthalten
Table 4.7- Water volume with and without storage
5 DISCUSSION
There are different issues from the Wansbeck study that have a relationship with the methodology that is used in this project including:
The derivation of catchment rainfall using ensemble members rainfall data is an uncertaint model that still requires improvement. Also, for each of the 24 ensemble members only 30 hours of data were available for the event The return period of the heavy rainfall during the event was not obtained The spatial and temporal resolution of the rainfall inputs must be at a fine enough resolution to simulate flood generating a rainfall event The resolution of observed rainfall data affects the optimal parameter values therefore the spatial and temporal resolution of the observed rainfall should be the same as that of the forecast ensemble rainfall model The SHETRAN model was on a very coarse grid of 1km2 and therefore may not show the catchment properties SHETRAN was not calibrated for the big event, because it needs more data recorded So, the previous calibrated SHETRAN model used may not have had enough flood events - For the purpose of comparison observed flows and SHETRAN flows, only the flows simulated by maximum, medium and minimum rainfall were used that may not have very good fit to observed flows The rainfall-runoff model structure should be selected for its ability to simulate a flood event, although this may cause uncertainties in antecedent baseflow simulation The return period for SHETRAN flows is unknown (because at least 100 years of rainfall and evaporation data is needed to run the model using this data but they are not available). So suitability of the storage with respect to the cost, is uncertain
- NOAH 1D does not model energy losses at bridges or weirs and therefore cannot represent the numerous river structures
- The NOAH 1D model did not include the impact of debris accumulation, so it overestimates flows against observed data(especially flows simulated by maximum ensemble rainfall)
- NOAH 1D calibration was done to give a good fit to peak flows and levels. On the other hand, this may have not been at the expense of the overall model fit
- The NOAH 1D model with the storage is a hypothesis model, so it is not the only solution to reduce flood risk in the town of Morpeth
Also there are potential positive and negative impacts and issues with built upstream storage in Upper Wansbeck and the Font that are listed below (Environment Agency, 2006 &2009).
Abbildung in dieser Leseprobe nicht enthalten
Table 5.1- Positive and negative impacts of upstream storage
6 CONCLUSION
The main aim of the project was to develop a continuous simulation approach to flood risk estimation and to apply it to the Wansbeck catchment. This aim has been reached successfully by the linking of a forecast ensemble rainfall model to a calibrated rainfall-runoff model and 1D river model. Also the continuous simulation approach has been used to derive the event and assess the possibility of using upstream storage (in Upper Wansbeck and the Font) to prevent flood risk in the town of Morpeth. The use of the 1D river model has explained the effects on the flood hydrograph at Mitford that might result if the flow from the storage was limited to a fixed discharge during the flood event.
Overall, the continuous simulation approach adopted here has automatically made an estimate of the flood duration, volume and antecedent base flow (Cameron et.al., 2002). This has the benefit of making the method less subjective and of simplifying the quantification of errors compared with the FEH approach (DEFRA and EA, 2005). Also, the approach is more flexible than the FEH method in the face of climate change (Kilsby, 2011).
The Parkin (2010) study used a 2D river model around Mitford. So, the assumption of 1D flow in this study may have been inappropriate where there were significant frictional impacts in Morpeth (Cunge et. al., 1980). Furthermore, the design of the storage in the Font, especially the smaller one in the NOAH 1D model had no significant effect to reduce the flows during the event. Therefore building storage in the Font would probably not be considered in the future, but the model still needs improvement. Although the model was calibrated in an unsteady state and has regenerated a range of flows in 1D.
Some of the specific problems encountered during this study relate to the forecast ensemble rainfall model. There are uncertainties in the forecast systems. A full uncertainty analysis has a high computational burden and moreover current forecasts ensemble data do not result in true probabilities of flooding, as uncertainties are not treated fully and assumptions of some of the approaches are violated. Any optimal framework will be inevitably a mixture of formal statistical treatments on top of the informal treatment of some parts of the modelling cascade.
Also, current data assimilation techniques in flood ensemble prediction systems contain soil moisture, snow cover or discharge, and some impacts on hydrological skill can be seen. On the other hand, hydrological data assimilation deserves much more interest especially when the initial hydrological state is presented as making an important impact on the anticipated lead time (Cloke and Pappenberger, 2009).
In addition, comparison of the timing and volumes of runoff against the observed flood water volumes on the floodplain explained that only a few of the ensemble members generated runoff that is consistent with the observed data, probably due to the positioning of the frontal system over the region in most of the ensemble members (Parkin et. al., 2011).
In conclusion, the study has succeeded in applying a continuous simulation approach of flood risk estimation to the Wansbeck. Although there are many advantages of using a rainfall runoff model and river model, the results here are largely comparative. As a result the return period of flow during the flood event used in this study, is unknown and this is just a hypothesis model. Therefore, the storage areas have not been tested in more different flood conditions. So, the results should only be used to highlight the possibility of the continuous simulation approach. In order to derive more meaningful results, the study needs to continue to re-examine the rainfall and rainfall-runoff models.
7 RECOMMENDATIONS FOR FUTURE STUDY
Good quality of input data is necessary for models to be able to provide good simulation results that enable proper decisions to be made for flood estimation and also for reduction of the current flood risk in the Wansbeck. The following points suggest what should be done to make good simulation results for decision makers to provide better decisions.
- The evidence supporting using an ensemble rainfall model for flood forecasting is still weak, and many more case studies are needed. Reports of future case studies should be more quantitative evidence for false flood alarms and contributions to uncertainty
- Improvement of the uncertainties of catchment response time, catchment characteristics and resolution of forcing data
- Ensemble rainfall models are not the magic solution to estimating the uncertainty of future rainfall and many further improvements are needed, including the ensemble simulation inputs themselves
- To establish whether the number of ensemble members used to derive flood forecasts is adequate or not
- Return period of the catchment rainfall in this project is not considered, so it should investigate the effect of decisions to make upstream storage
- Using a smaller grid for the SHETRAN model to better represent elevation, land use and soil type in future is necessary
- More knowledge of hydraulic and hydrology is quite necessary so as to be able to decide on data which are good representatives of river cross sections and which affect the flow dynamics and flood propagation
- Consider the effects of debris around the bridges in the Wansbeck catchment
- Investigate the effects of other flood storage upstream of the NOAH 1D model
- Run the calibrated NOAH 1D model with designing the dam in Upper Wansbeck and consider the effects of this on flows and water levels at Mitford
- Run the calibrated NOAH 1D model with designing all the proposed storage and consider the effects of this on flows and water levels at Mitford
- Consider the possibility of allowing uncontrolled spill later on in a flood event to allow for the culvert overtopping in very different floods
- Investigate the possibility of utilising an alternative storage area on the Font as suggested in the Environment Agency report (2009) near Woodhouse and survey the flows and water levels
- Flows at Mitford when the Font Reservoir is drawn down should be compared with the scenario where the reservoir is full and spill is uncontrolled
- Investigate the performance of existing flood mitigation schemes in the study area under climate change and land use change scenarios using integrated hydrologic and hydraulic models and make decisions for suitable flood mitigation schemes in the future
- Simulation of the 2008 flood and previous flood events in Morpeth can also be done by coupling the 1D model and the 2D model to approach a more accurate representation of flows and water levels in Morpeth and analyse the results to help in decision making based on previous flooding events
8 REFERENCES
ADAMOWSKI, K., (2000). Regional analysis of annual maximum and partial duration flood data by nonparametric and L-momentum methods, Journal of Hydrology, 219-231
ASCE, (1997). Flood-Runoff analysis. New York, USA: ASCE Press, 176
ARCEMENT, G. J. & SCHNEIDER, V. R., (2011). Guide for selecting Manning's Roughness Coefficients for natural channels and flood plains, [online] http:// www.fhwa.dot.gov/bridge/wsp2339.pdf
ASHFAQ, A. & WEBSTER, P., (2007). Evaluation of the FEH Rainfall- Runoff Method for Catchments in the UK, Water and Environment Journal, 222-224
ASQUITH, W. H. & FAMIGLIETTI, J. S., (2000). Precipitation areal-reduction factor estimation using an annual-maxima centered approach, Journal of Hydrology, 55-69
BASSON, G., (2010). Erosion and sediment dynamics from catchment to coast, 13-19 [online] http: //www.irtces.org/isi
BATHURST, J., (2010). Physically Based Distributing Modelling: catchment flow and transport modelling. Newcastle: Newcastle University, 3-12
BELL, V.A., CARRINGTON, D., MOOR, R.J., (2001). Comparison of Rainfall-Runoff Models for flood forecasting,, Journal of Hydrology (Environment Agency), 153- 155
BEVEN, K., (2000). Rainfall-Runoff Modelling. Chichester West Sussex, UK: John Wiley & Sons Ltd, Baffins Lane
BIRKINSHAW, S., JAMES, P., EWEN, J. (2009). Graphical user interface for rapid set-up of SHETRAN physically-based river catchment model
BRITISH HYDROLOGICAL SOCIETY, [online] http: //www.bris.ac.UK/civilengineering/research
CALVER, A., CROOKS, S., JONES, D., KAY, A., KJELDSEN, T., REYNARD, N., (2005). National river catchment flood frequency method using continuous simulation, Journal of Hydrology (Environment Agency), 7- 114
CALVER, A., LAMB, R., (2000). Generalised flood frequency estimation using continuous simulation. Potsdam, Germany: Potsdam Institute for Climate Impact Research , 412-421
CAMERON, D., BEVEN, K., TAWN, J., NADEN, P., (2000). Flood frequency estimation by continuous simulation (with likelihood based uncertainty estimation). Hydrology and Earth system science, 23-24
CAMERON, D., KNEALE, P., SEE, L., (2002). An evaluation of a neural net modelling approach to flood forecasting for an upload catchment. Hydrological Processes, 1033-1046
CENTRE FOR ECOLOGY & HYDROLOGY, [online] http: //www.ceh.ac.uk/newe/press/press Release Major upgrade flood estimation software
CHOW, V. T., (1959). Open channel Hydraulics. New York: McGraw- Hill, 112
CLOKE, H.L., PAPPENBERGER, F., (2009). Ensemble flood forecasting: A review. Journal of Hydrology, 614-623
CUNGE, J. A., HOLLY, F.M. & VERWEW, A., (1980). Practical Aspects of Computational River Hydraulic, (1st Edition). Massachussetts: Lowa Institue of Hydraulic Research
DEFRA, 2005. Revitalisation of the FSR/FEH rainfall runoff method, R&D Technical report FD1912/TR
ENVIRONMENT AGENCY, 2006. River Wansbeck: Scheme Viability Note
ENVIRONMENT AGENCY, 2007. Catchment flood management plan :History for Wansbeck at Mitford
ENVIRONMENT AGENCY, 2009. Morpeth flood Alleviation scheme options consultation
ENVIRONMENT AGENCY, 2009. Work continuous to cut Morpeth flood risk
ENVIRONMENT STATEMENT FAIRE FORD LAKES, FLOOD RISK ASSESMENT, 2008, [online] http: //www.fairford lakes.info/privately Es/Es .../11 .../Appendix-11.1-FRA.pdf
EWEN, J., O DONNELL, G., BURTON, A., ENDA O CHANNEL, P., (2006). Errors and uncertainty in physically based rainfall-runoff modelling of catchment change effects , Journal of Hydrology, 641-650
EWEN, J., PARKIN, G., ENDA O CHANNEL, P., (2000). Distributed River Basin flow and Transport Modelling system, 250-251 [online] http: //www.research.ncl.ac.UK/shetran/SHETRAN-ASCE-paper.pdf
FORDHAM, M., (2008). Preliminary finding from the MICRODIS project. Newcastle Upon Tyne: Northumbria university
GVUS 19th www use survey (n.d.). Retrieved 16th April, 2011, from World Wide Web http://www.bbc.co.uk/tyne/content/images/2008/12/23/floods N.B. "n.d." stands for "no date"
GVUS 19th www use survey (n.d.). Retrieved 16th April, 2011, from World Wide Web http://www.microdis-eu.be/content/morpeth-united-kingdom N.B. "n.d." stands for "no date"
GVUS 19th www use survey (n.d.). Retrieved 20th April, 2011, from World Wide Web http://www. Coolgeography.co.uk/GCSE/AQA N.B. "n.d." stands for "no date"
GVUS 19th www use survey (n.d.). Retrieved 16th April, 2011, from World Wide Web http://www. Dwaf.gov.za/Geohydrology/gra 2/3bE Final Report.pdf N.B. "n.d." stands for "no date"
GVUS 19th www use survey (n.d.). Retrieved 16th June, 2011, from World Wide Web http://www. Research.ncl.ac.uk/noah/images N.B. "n.d." stands for "no date"
JEREMY BENN ASSOCIATES, 2005 , River Wansbeck Strategy: River Modelling for Wansbeck Strategy & Wansbeck PAR
JEREMY BENN ASSOCIATES, 2008, flooding in Castle Morpeth 6&7 September 2008. Castle Morpeth Borough Council
JEREMY BENN ASSOCIATES, 2009, flooding in Castle Morpeth –Independent Review Final Report v1.0. Castle Morpeth Borough Council
JEREMY BENN ASSOCIATES, 2008, flood estimation handbook
KILSBY, C., (2011). Continuous simulation modelling of flood frequency distributions. Newcastle Upon Tyne: Newcastle university, 11
KILSBY, C., (2011). Statistical approach to flood.. Newcastle Upon Tyne: Newcastle university, 1-16
KILSBY, C., (2011). The FEH Rainfall-Runoff Method. Newcastle Upon Tyne: Newcastle university, 2-3
KJELDSEN, T., JONES, D., BAYLISS, A., SPENCER, P., SURENDRAN, S., LAEGER, S., WEBSTER, P., MCDONALD, D.,(2008). Improving the FEH statistical method. In:Flood &coastal management conference. Manchester, Manchester university
KOKKONEN, T., KOIVUSALO, H., KAROVEN, T., CROKE, B., JAKEMAN, A., (2004). Exploring streamflow response to effective rainfall across event magnitude scale, Hydrological processes, 1467-1486
KUTIJA, V. & MURRAY, M. G., (2000). A fast hydrodynamic network modelling system using object oriented numerics, Newcastle Upon Tyne: Newcastle university, 201-205
KUTIJA, V. & MURRAY, M. G., (2003). NOAH 1D- use of object oriented numerics in modelling of free surface flows in networks of pipes and channels, Newcastle Upon Tyne: Newcastle university, 201-205
LITTLEWOOD, L., (2002). Continuous river flow simulation: methods, applications and uncertainties . British Hydrological Society, 53-54
MASCARENHAS, F.C.B, (2005). Flood risk simulation . Southampton, UK: WIT press, 130-132
O CONNELL, P.E., (2000). Overview of computational models . Newcastle Upon Tyne: Newcastle university, 5-6
PARKIN, G., BIRKINSHAW, S., KILSBY, C., ROBERTS, N., ASFAW., A.S., (2008). The September 2008 Morpeth flood: performance of ensemble rainfall forecasts in relation to pluvial flooding impacts , Journal of Hydrology, 1-9
PARKIN, G., (1995). SHETRAN water flow component, Equations and Algorithms . Newcastle Upon Tyne: Newcastle university, 1-11
PARKIN, G., (2010). The September 2008 Morpeth flood reconstruction . Newcastle Upon Tyne: Newcastle university, 1-6
RAHMAN, A., HOANG, T.M.T., WEINMANN, P.E., LAURENSON, E.M., (1998). Joint probability approaches to design flood estimation: a review Report 98/8. Clayton, Victoria, Australia, Cooperation Research Center for Hydrology, Monash university, page 70
RESEARCH GROUP OF NORTHUMBRIA UNIVERSITY, (2010). Social and Economic Impacts of extreme events: Evidence, Methods and tools, 3-10
ROBERTS, M., COLE, J., FORBES, M., MOORE, J., BOSWELL, D., (2009). Use of high resolution NWP rainfall and river flow forecasts for advance warning of the Carlisle flood, north-west England , Royal Meteorology Society, 23-27
RODDING, T., (2007). The revitalised FSR/FEH rainfall-runoff method. Centre of Ecology & Hydrology, 1-4
SCHAAKE, C., HAMILL, M., BUIZZA, R., CLARK, M., (2007). The Hydrological ensemble prediction experiment , American Meteorological Society, 1541-1547
SCHELLEKENS, J., WEERTS, A.H., MOORE, R.J., PIERCE, C.E., HILDON, S., (2011). The use of MOGREPS ensemble rainfall forecasts in operational flood forecasting systems across England and Wales , Advances in Geosciences, 77-83
SHAW, E.M, (2004). Hydrology in practice. (3rd Edition), Routledge, Oxon, 321 pp.
SPENCER, P., HIGGINSON, N., PALMER, T., WASS, P., (2004). HIFlows-UK , [online] http://www.environment agency.gov.UK/hiflows-UK
WARD, E., (2008). Development of a continuous simulation approach for flood risk estimation applied to the Wansbeck catchment, Newcastle Upon Tyne: Newcastle University
WAGENER, T. & HOWARD, S., (2006). Parameter estimation and regionalization for continuous rainfall-runoff models including uncertainty, Journal of Hydrology, 132-154
WHATMORE, J. & LANDSTROM, C., (2003). Manning's N –putting Roughness to work, Oxford: Oxford university, 1-3
WILBY, R.L., BEVEN, K.J., REYNARD, N.S., (2007). Climate change and fluvial flood risk in the UK: more of the same, , Hydrological processes,10.1002/hyp.6847
XUAN, Y., CLUCKIE, I.D., WANG, Y., (2009). Uncertainty analysis of hydrological ensemble forecasts in a distributed model utilising short-rang rainfall prediction, , Hydrology and Earth system Sciences, 293-303
9 APPENDICES
Appendix 1- List of Data for SHETRAN Model
Appendix 2- List of Cross-Section Data used in NOAH 1D
Appendix 3- Original weir sections and interpolated sections added
Appendix 4- Bridge sections and upstream and downstream
Appendix 5- List of NOAH 1D boundary & lateral inflow files
Appendix 6- List of NOAH 1D Models
Appendix 1- List of Data for SHETRAN Model
Land use, geology and DEM data for the SHETRAN model was used from a previous study of the Wansbeck catchment. The data is given below and the data is on the CD.
1. Catchment Areas
- Bothal Mill catchment area
- Mitford 1km catchment area
2. Map Data
- SHETRAN 1km DEM
- SHETRAN land 7 classes
- SHETRAN 625k geol
3. Hydrometry Data
- Input, output and rundata
- Observed flows and water levels at Mitford
- Rainfall data
4. SHETRAN simulation results for all 24 ensemble rainfall
- Event- discharge for the tributaries
Appendix 2- List of Cross-Section Data used in NOAH 1D
Cross section data for the NOAH 1D model was used from a previous study of the Wansbeck catchment. The data is given below and the data is on the CD.
- Original sections
- Interpolated Weir Sections
- Altered Bridge Sections
- Deletion of Levees
- X-Section Floodplain Extension
Abbildung in dieser Leseprobe nicht enthalten
Appendix 5- List of NOAH 1D Boundary & lateral inflow files
The runoff generation for the event were approached by the SHETRAN model for the Upper Wansbeck and the Font that was simulated by maximum, medium and minimum ensemble rainfall. Then the data was used as boundary conditions. In addition, the lateral inflow was calculated and was put into the model. The data are listed below and are attached to the CD:
- Boundary conditions for Upper Wansbeck and the Font simulated by maximum, medium and minimum ensemble rainfall
- Lateral inflow for maximum, medium and minimum ensemble rainfall
Appendix 6- List of NOAH 1D Models
All NOAH 1D models that were used in this project are listed below and they are on the CD. The models will not run on the CD because NOAH 1D software is compulsory.
List of Folders on CD:
- NOAH 1D model simulated by maximum ensemble rainfall without storage
- NOAH 1D model simulated by maximum ensemble rainfall with storage at Upper Wansbeck
- NOAH 1D model simulated by maximum ensemble rainfall with storage at the Font
- NOAH 1D model simulated by maximum ensemble rainfall with smaller storage at the Font
- NOAH 1D model simulated by medium ensemble rainfall without storage
- NOAH 1D model simulated by medium ensemble rainfall with storage at Upper Wansbeck
- NOAH 1D model simulated by maximum ensemble rainfall with storage at the Font
- NOAH 1D model simulated by maximum ensemble rainfall with smaller storage at the Font
- NOAH 1D model simulated by minimum ensemble rainfall without storage
- NOAH 1D model simulated by minimum ensemble rainfall with storage at Upper Wansbeck
- NOAH 1D model simulated by maximum ensemble rainfall with storage at the Font
- NOAH 1D model simulated by maximum ensemble rainfall with smaller storage at the Font
- NOAH 1D model results simulated by maximum ensemble rainfall (Excel file)
- NOAH 1D model results simulated by medium ensemble rainfall (Excel file)
- NOAH 1D model results simulated by minimum ensemble rainfall (Excel file)
- Location of upstream storage (AutoCAD file)
[...]
- Quote paper
- Safieh Javadinejad (Author), 2011, Continuous Simulation Model for the Wansbeck Catchment and Analysis of the Flood Risk Estimation, Munich, GRIN Verlag, https://www.grin.com/document/902904
-
-
-
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X. -
-
Upload your own papers! Earn money and win an iPhone X.