R.R.P. van Nooijen
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
12 records found
1
Improving groundwater–surface water exchange representation in WALRUS
Evaluation of direction- and temperature-dependent resistance parameterization for a managed dutch lowland catchment
Climate change is intensifying hydrological extremes in Dutch lowland catchments, where shallow groundwater tables, active water management, and strong groundwater–surface water (GW–SW) coupling pose particular modelling challenges. This thesis investigates whether targeted structural modifications to the Wageningen Lowland Runoff Simulator (WALRUS) can improve its performance for the managed lowland catchment of the Marswetering in the Netherlands.
Two adaptations are proposed and evaluated against the baseline model. The first is a temperature- and direction-dependent groundwater resistance parameterization (cG), which is intended to represent streambed clogging asymmetry as well as seasonal variability driven by changes in water viscosity. The second is an adapted quickflow partitioning formulation (fQS), which is designed to prevent the complete suppression of land-surface quickflow during extended dry periods. Five model variants are calibrated using a multi-objective framework combining six hydraulic signatures including discharge, flow duration curve, autocorrelation, seasonal runoff coefficient, and normalized groundwater dynamics, and evaluated across a calibration period and two independent testing periods.
All model variants reproduce peak discharge with reasonable skill, but consistently underperform on low flows and summer groundwater dynamics. A structurally recurring opposite-sign error in summer groundwater behaviour, present across all variants and both testing periods, points to systematic underestimation of actual evapotranspiration or outward seepage or overestimation of the equilibrium storage deficit. The adapted fQS formulation produces a physically consistent but negligible improvement, largely absorbed by compensating changes in the quickflow reservoir constant. The direction-dependent cG formulation successfully reproduces asymmetric GW–SW exchange resistance, but introduces elevated summer discharge that worsens seasonal water balance closure. The temperature-dependent component of cG shows no identifiable calibrated influence. Additionally, the spatially uniform representation of managed surface water levels is identified as a source of uncertainty, given the heterogeneous water management across the catchment's sub-catchments.
The results indicate that improving summer unsaturated zone dynamics represents a higher priority for WALRUS development than further GW–SW exchange complexity. The multi-objective calibration framework developed here is directly transferable to comparable lowland catchment applications. ...
Two adaptations are proposed and evaluated against the baseline model. The first is a temperature- and direction-dependent groundwater resistance parameterization (cG), which is intended to represent streambed clogging asymmetry as well as seasonal variability driven by changes in water viscosity. The second is an adapted quickflow partitioning formulation (fQS), which is designed to prevent the complete suppression of land-surface quickflow during extended dry periods. Five model variants are calibrated using a multi-objective framework combining six hydraulic signatures including discharge, flow duration curve, autocorrelation, seasonal runoff coefficient, and normalized groundwater dynamics, and evaluated across a calibration period and two independent testing periods.
All model variants reproduce peak discharge with reasonable skill, but consistently underperform on low flows and summer groundwater dynamics. A structurally recurring opposite-sign error in summer groundwater behaviour, present across all variants and both testing periods, points to systematic underestimation of actual evapotranspiration or outward seepage or overestimation of the equilibrium storage deficit. The adapted fQS formulation produces a physically consistent but negligible improvement, largely absorbed by compensating changes in the quickflow reservoir constant. The direction-dependent cG formulation successfully reproduces asymmetric GW–SW exchange resistance, but introduces elevated summer discharge that worsens seasonal water balance closure. The temperature-dependent component of cG shows no identifiable calibrated influence. Additionally, the spatially uniform representation of managed surface water levels is identified as a source of uncertainty, given the heterogeneous water management across the catchment's sub-catchments.
The results indicate that improving summer unsaturated zone dynamics represents a higher priority for WALRUS development than further GW–SW exchange complexity. The multi-objective calibration framework developed here is directly transferable to comparable lowland catchment applications. ...
Climate change is intensifying hydrological extremes in Dutch lowland catchments, where shallow groundwater tables, active water management, and strong groundwater–surface water (GW–SW) coupling pose particular modelling challenges. This thesis investigates whether targeted structural modifications to the Wageningen Lowland Runoff Simulator (WALRUS) can improve its performance for the managed lowland catchment of the Marswetering in the Netherlands.
Two adaptations are proposed and evaluated against the baseline model. The first is a temperature- and direction-dependent groundwater resistance parameterization (cG), which is intended to represent streambed clogging asymmetry as well as seasonal variability driven by changes in water viscosity. The second is an adapted quickflow partitioning formulation (fQS), which is designed to prevent the complete suppression of land-surface quickflow during extended dry periods. Five model variants are calibrated using a multi-objective framework combining six hydraulic signatures including discharge, flow duration curve, autocorrelation, seasonal runoff coefficient, and normalized groundwater dynamics, and evaluated across a calibration period and two independent testing periods.
All model variants reproduce peak discharge with reasonable skill, but consistently underperform on low flows and summer groundwater dynamics. A structurally recurring opposite-sign error in summer groundwater behaviour, present across all variants and both testing periods, points to systematic underestimation of actual evapotranspiration or outward seepage or overestimation of the equilibrium storage deficit. The adapted fQS formulation produces a physically consistent but negligible improvement, largely absorbed by compensating changes in the quickflow reservoir constant. The direction-dependent cG formulation successfully reproduces asymmetric GW–SW exchange resistance, but introduces elevated summer discharge that worsens seasonal water balance closure. The temperature-dependent component of cG shows no identifiable calibrated influence. Additionally, the spatially uniform representation of managed surface water levels is identified as a source of uncertainty, given the heterogeneous water management across the catchment's sub-catchments.
The results indicate that improving summer unsaturated zone dynamics represents a higher priority for WALRUS development than further GW–SW exchange complexity. The multi-objective calibration framework developed here is directly transferable to comparable lowland catchment applications.
Two adaptations are proposed and evaluated against the baseline model. The first is a temperature- and direction-dependent groundwater resistance parameterization (cG), which is intended to represent streambed clogging asymmetry as well as seasonal variability driven by changes in water viscosity. The second is an adapted quickflow partitioning formulation (fQS), which is designed to prevent the complete suppression of land-surface quickflow during extended dry periods. Five model variants are calibrated using a multi-objective framework combining six hydraulic signatures including discharge, flow duration curve, autocorrelation, seasonal runoff coefficient, and normalized groundwater dynamics, and evaluated across a calibration period and two independent testing periods.
All model variants reproduce peak discharge with reasonable skill, but consistently underperform on low flows and summer groundwater dynamics. A structurally recurring opposite-sign error in summer groundwater behaviour, present across all variants and both testing periods, points to systematic underestimation of actual evapotranspiration or outward seepage or overestimation of the equilibrium storage deficit. The adapted fQS formulation produces a physically consistent but negligible improvement, largely absorbed by compensating changes in the quickflow reservoir constant. The direction-dependent cG formulation successfully reproduces asymmetric GW–SW exchange resistance, but introduces elevated summer discharge that worsens seasonal water balance closure. The temperature-dependent component of cG shows no identifiable calibrated influence. Additionally, the spatially uniform representation of managed surface water levels is identified as a source of uncertainty, given the heterogeneous water management across the catchment's sub-catchments.
The results indicate that improving summer unsaturated zone dynamics represents a higher priority for WALRUS development than further GW–SW exchange complexity. The multi-objective calibration framework developed here is directly transferable to comparable lowland catchment applications.
In a world with accelerating climate change, rapid population increase and urbanization, urban water systems are under a growing stress. Thus precise short- and medium-term water demand forecasts are needed to optimize water supply operations. Water demand is influenced by human behavior and industrial activities which bring uncertainty, hence it is useful to utilize probabilistic methods to forecast water demand. This thesis provides an overview of probabilistic methods to predict water demand 24 hours ahead, highlighting their advantages, disadvantages, the accuracy of their interval and point forecasts. The case study uses the dataset from the Battle of Water Demand 2024, covering two years and two months of data across 10 districts in Ferrara, Italy, including residential, hospital, countryside, city center, and industrial districts. Three commonly used probabilistic extensions of neural networks were applied: QR (Quantile Regression), MDN (Gaussian Mixture Density Network), and an adapted CQR (Conformal Quantile Regression) method with online updating.
First, forecast models were developed to obtain probabilistic predictions. Three neural network architectures were investigated: a linear model, an MLP (multi-layer perceptron), and an LSTM (Long Short-Term Memory) model, along with a seasonal moving average as a benchmark. These neural network models were trained per district and collectively across districts. When trained per district, the linear model performed most accurately. When trained together, the MLP model performed best, but the linear model generalized the best overall, with the MLP generalizing second-best. The LSTM model had the worst performance. In districts with less heteroscedasticity in the demand pattern, the benchmark model performed on par with the neural networks in the end of the forecasting horizon, indicating that complex models are not always necessary. A categorical variable to determine the DMA did not improve the point forecasts.
Because the MLP with solely lagged features ultimately had the best performance for point forecasts, this model was used for probabilistic extensions to estimate the 0.95 prediction interval. The MLP was extended with the aforementioned probabilistic extensions. The probabilistic models were assessed in terms of reliability with where a probability was computed that tells whether the 0.95 prediction interval is reached as well as sharpness which tells how wide the interval is. Finally the Winkler Score is used that computes a trade-off between both.
Both models that fully learn the prediction interval (QR and MDN) were more difficult to calibrate, and further research is needed to calibrate them accordingly. By training these models jointly on the 10 DMAs, the coverages did vary per DMA. This can potentially be solved by training one model per DMA or by using regularization per DMA to push the model to have a similar coverage per DMA.
The MCD model had difficulty to adapt the prediction interval over the forecasting horizon, causing the coverage to reduce. The QR and MDN models also have trade off imbalances between reliability and sharpness over the forecasting horizon on the testing set, which have more random patterns. Interestingly the Conformal Prediction algorithm maintains its coverage best over the forecasting horizon and increases the sharpness, which is due to the online updating procedure. When allowing small decreases up to 0.02 probability in coverage on the testing set, the CQR model performs best according to the Winkler Score. The MDN performs best when larger drops of coverage are allowed.
Analyzing the rolling coverage over time shows improvements are possible, especially in late spring and summer periods there is under-coverage. There is also still a difference of coverage between weekdays and weekends. This indicates there is still epistemic uncertainty left to reduce, which is the uncertainty of data and model parameters. More features are recommended to reduce this which are categorical features as well as future weather data. This is recommended to investigate for non-industrial DMAs, by assuming a perfect forecast. A larger dataset may also be beneficial to obtain better performing models. ...
First, forecast models were developed to obtain probabilistic predictions. Three neural network architectures were investigated: a linear model, an MLP (multi-layer perceptron), and an LSTM (Long Short-Term Memory) model, along with a seasonal moving average as a benchmark. These neural network models were trained per district and collectively across districts. When trained per district, the linear model performed most accurately. When trained together, the MLP model performed best, but the linear model generalized the best overall, with the MLP generalizing second-best. The LSTM model had the worst performance. In districts with less heteroscedasticity in the demand pattern, the benchmark model performed on par with the neural networks in the end of the forecasting horizon, indicating that complex models are not always necessary. A categorical variable to determine the DMA did not improve the point forecasts.
Because the MLP with solely lagged features ultimately had the best performance for point forecasts, this model was used for probabilistic extensions to estimate the 0.95 prediction interval. The MLP was extended with the aforementioned probabilistic extensions. The probabilistic models were assessed in terms of reliability with where a probability was computed that tells whether the 0.95 prediction interval is reached as well as sharpness which tells how wide the interval is. Finally the Winkler Score is used that computes a trade-off between both.
Both models that fully learn the prediction interval (QR and MDN) were more difficult to calibrate, and further research is needed to calibrate them accordingly. By training these models jointly on the 10 DMAs, the coverages did vary per DMA. This can potentially be solved by training one model per DMA or by using regularization per DMA to push the model to have a similar coverage per DMA.
The MCD model had difficulty to adapt the prediction interval over the forecasting horizon, causing the coverage to reduce. The QR and MDN models also have trade off imbalances between reliability and sharpness over the forecasting horizon on the testing set, which have more random patterns. Interestingly the Conformal Prediction algorithm maintains its coverage best over the forecasting horizon and increases the sharpness, which is due to the online updating procedure. When allowing small decreases up to 0.02 probability in coverage on the testing set, the CQR model performs best according to the Winkler Score. The MDN performs best when larger drops of coverage are allowed.
Analyzing the rolling coverage over time shows improvements are possible, especially in late spring and summer periods there is under-coverage. There is also still a difference of coverage between weekdays and weekends. This indicates there is still epistemic uncertainty left to reduce, which is the uncertainty of data and model parameters. More features are recommended to reduce this which are categorical features as well as future weather data. This is recommended to investigate for non-industrial DMAs, by assuming a perfect forecast. A larger dataset may also be beneficial to obtain better performing models. ...
In a world with accelerating climate change, rapid population increase and urbanization, urban water systems are under a growing stress. Thus precise short- and medium-term water demand forecasts are needed to optimize water supply operations. Water demand is influenced by human behavior and industrial activities which bring uncertainty, hence it is useful to utilize probabilistic methods to forecast water demand. This thesis provides an overview of probabilistic methods to predict water demand 24 hours ahead, highlighting their advantages, disadvantages, the accuracy of their interval and point forecasts. The case study uses the dataset from the Battle of Water Demand 2024, covering two years and two months of data across 10 districts in Ferrara, Italy, including residential, hospital, countryside, city center, and industrial districts. Three commonly used probabilistic extensions of neural networks were applied: QR (Quantile Regression), MDN (Gaussian Mixture Density Network), and an adapted CQR (Conformal Quantile Regression) method with online updating.
First, forecast models were developed to obtain probabilistic predictions. Three neural network architectures were investigated: a linear model, an MLP (multi-layer perceptron), and an LSTM (Long Short-Term Memory) model, along with a seasonal moving average as a benchmark. These neural network models were trained per district and collectively across districts. When trained per district, the linear model performed most accurately. When trained together, the MLP model performed best, but the linear model generalized the best overall, with the MLP generalizing second-best. The LSTM model had the worst performance. In districts with less heteroscedasticity in the demand pattern, the benchmark model performed on par with the neural networks in the end of the forecasting horizon, indicating that complex models are not always necessary. A categorical variable to determine the DMA did not improve the point forecasts.
Because the MLP with solely lagged features ultimately had the best performance for point forecasts, this model was used for probabilistic extensions to estimate the 0.95 prediction interval. The MLP was extended with the aforementioned probabilistic extensions. The probabilistic models were assessed in terms of reliability with where a probability was computed that tells whether the 0.95 prediction interval is reached as well as sharpness which tells how wide the interval is. Finally the Winkler Score is used that computes a trade-off between both.
Both models that fully learn the prediction interval (QR and MDN) were more difficult to calibrate, and further research is needed to calibrate them accordingly. By training these models jointly on the 10 DMAs, the coverages did vary per DMA. This can potentially be solved by training one model per DMA or by using regularization per DMA to push the model to have a similar coverage per DMA.
The MCD model had difficulty to adapt the prediction interval over the forecasting horizon, causing the coverage to reduce. The QR and MDN models also have trade off imbalances between reliability and sharpness over the forecasting horizon on the testing set, which have more random patterns. Interestingly the Conformal Prediction algorithm maintains its coverage best over the forecasting horizon and increases the sharpness, which is due to the online updating procedure. When allowing small decreases up to 0.02 probability in coverage on the testing set, the CQR model performs best according to the Winkler Score. The MDN performs best when larger drops of coverage are allowed.
Analyzing the rolling coverage over time shows improvements are possible, especially in late spring and summer periods there is under-coverage. There is also still a difference of coverage between weekdays and weekends. This indicates there is still epistemic uncertainty left to reduce, which is the uncertainty of data and model parameters. More features are recommended to reduce this which are categorical features as well as future weather data. This is recommended to investigate for non-industrial DMAs, by assuming a perfect forecast. A larger dataset may also be beneficial to obtain better performing models.
First, forecast models were developed to obtain probabilistic predictions. Three neural network architectures were investigated: a linear model, an MLP (multi-layer perceptron), and an LSTM (Long Short-Term Memory) model, along with a seasonal moving average as a benchmark. These neural network models were trained per district and collectively across districts. When trained per district, the linear model performed most accurately. When trained together, the MLP model performed best, but the linear model generalized the best overall, with the MLP generalizing second-best. The LSTM model had the worst performance. In districts with less heteroscedasticity in the demand pattern, the benchmark model performed on par with the neural networks in the end of the forecasting horizon, indicating that complex models are not always necessary. A categorical variable to determine the DMA did not improve the point forecasts.
Because the MLP with solely lagged features ultimately had the best performance for point forecasts, this model was used for probabilistic extensions to estimate the 0.95 prediction interval. The MLP was extended with the aforementioned probabilistic extensions. The probabilistic models were assessed in terms of reliability with where a probability was computed that tells whether the 0.95 prediction interval is reached as well as sharpness which tells how wide the interval is. Finally the Winkler Score is used that computes a trade-off between both.
Both models that fully learn the prediction interval (QR and MDN) were more difficult to calibrate, and further research is needed to calibrate them accordingly. By training these models jointly on the 10 DMAs, the coverages did vary per DMA. This can potentially be solved by training one model per DMA or by using regularization per DMA to push the model to have a similar coverage per DMA.
The MCD model had difficulty to adapt the prediction interval over the forecasting horizon, causing the coverage to reduce. The QR and MDN models also have trade off imbalances between reliability and sharpness over the forecasting horizon on the testing set, which have more random patterns. Interestingly the Conformal Prediction algorithm maintains its coverage best over the forecasting horizon and increases the sharpness, which is due to the online updating procedure. When allowing small decreases up to 0.02 probability in coverage on the testing set, the CQR model performs best according to the Winkler Score. The MDN performs best when larger drops of coverage are allowed.
Analyzing the rolling coverage over time shows improvements are possible, especially in late spring and summer periods there is under-coverage. There is also still a difference of coverage between weekdays and weekends. This indicates there is still epistemic uncertainty left to reduce, which is the uncertainty of data and model parameters. More features are recommended to reduce this which are categorical features as well as future weather data. This is recommended to investigate for non-industrial DMAs, by assuming a perfect forecast. A larger dataset may also be beneficial to obtain better performing models.
Evaluating Constant Failure Rates in Storm Surge Barriers
A Statistical Framework Applied to Censored Component Lifetimes of the Oosterscheldekering
Master thesis
(2023)
-
J.H. Epema, M. Kok, G. Jongbloed, P. Chen, R.R.P. van Nooijen, A.M.R. Bakker, L.F. Mooyaart
This study examines the validity of constant failure rates in the reliability assessment of storm surge barriers, with a focus on the Stormvloedkering Oosterschelde (SVKO). Analysing a dataset of 1,501 malfunctions, including 87 critical incidents over six years, we employ Exponential and Weibull statistical models to assess failure rates. The research question—whether the assumption of constant failure rates over time is valid—is addressed with a nuanced perspective. The findings in this study reveal that neither model conclusively fits all failure scenarios, with some data supporting constant rates and other data indicating variability. The Weibull model better describes certain scenarios, suggesting variable failure rates, while in other instances, both models show comparable performance. The p-values from hypothesis testing and visual inspection of component data provide inconclusive evidence, leading to the suggestion that both constant and variable failure rates may exist in storm surge barrier components. The research contributes to the field by challenging the prevailing assumption of constant failure rates, developing a statistical framework, and by suggesting the need for a flexible, scenario-specific approach to modeling failure rates for improved reliability assessments.
...
This study examines the validity of constant failure rates in the reliability assessment of storm surge barriers, with a focus on the Stormvloedkering Oosterschelde (SVKO). Analysing a dataset of 1,501 malfunctions, including 87 critical incidents over six years, we employ Exponential and Weibull statistical models to assess failure rates. The research question—whether the assumption of constant failure rates over time is valid—is addressed with a nuanced perspective. The findings in this study reveal that neither model conclusively fits all failure scenarios, with some data supporting constant rates and other data indicating variability. The Weibull model better describes certain scenarios, suggesting variable failure rates, while in other instances, both models show comparable performance. The p-values from hypothesis testing and visual inspection of component data provide inconclusive evidence, leading to the suggestion that both constant and variable failure rates may exist in storm surge barrier components. The research contributes to the field by challenging the prevailing assumption of constant failure rates, developing a statistical framework, and by suggesting the need for a flexible, scenario-specific approach to modeling failure rates for improved reliability assessments.
Student report
(2022)
-
D.D. Dekker, M. Hrachowitz, M.A. Schleiss, R.R.P. van Nooijen, Abdulghani Hasan
This additional thesis project is performed as preliminary research for a bigger project that they are going to start at Lund University, to investigate whether the use of a different interpolation methods, to link the precipitation data to the sub-basins centers of the HYPE model, lead to improved model performance. In this report, previously performed research is summarised and the limitations in researching this question with the HYPE model are described. A start is made with investigating this question, by answering the question whether different interpolation methods result in a different discharge when it is assumed that all fallen precipitation ends up as discharge. This is investigated for the PO basin in Italy with 4 interpolation methods: NN, IDW, BIL and OK. The effects of the interpolation methods on the computed discharge time series are analysed with the use of the correlation, RE, NSE and KGE. It is shown that for the PO basin, with a 1000 km2 average sub-basin size and a gridded data set with a resolution of 50 km, the interpolation methods do not produce differences for which you would expect that it could lead to model improvement. However, based on findings from previous research, a next step is proposed, in which we can investigate if we do observe a difference at a different sub-basin scale or data resolution.
...
This additional thesis project is performed as preliminary research for a bigger project that they are going to start at Lund University, to investigate whether the use of a different interpolation methods, to link the precipitation data to the sub-basins centers of the HYPE model, lead to improved model performance. In this report, previously performed research is summarised and the limitations in researching this question with the HYPE model are described. A start is made with investigating this question, by answering the question whether different interpolation methods result in a different discharge when it is assumed that all fallen precipitation ends up as discharge. This is investigated for the PO basin in Italy with 4 interpolation methods: NN, IDW, BIL and OK. The effects of the interpolation methods on the computed discharge time series are analysed with the use of the correlation, RE, NSE and KGE. It is shown that for the PO basin, with a 1000 km2 average sub-basin size and a gridded data set with a resolution of 50 km, the interpolation methods do not produce differences for which you would expect that it could lead to model improvement. However, based on findings from previous research, a next step is proposed, in which we can investigate if we do observe a difference at a different sub-basin scale or data resolution.
The water balance model has been an indispensable tool for quantifying water supply and demand and regulating water resources for decades. By convention, hydrologists use in-situ measurements of river discharge for model calibration. Whereas hardly is it possible for areas suffering from data storage. Researchers have been working on methods dealing with this problem for many years. Satellite datasets are potential substitutes for the in-situ observation of hydrological variables. This research proposes a monthly strategy using satellite time series as model inputs. This research aims to apply this strategy in the absence of gauged data for discharge simulations and predictions. Instead of streamflow, the strategy uses variables like actual evapotranspiration (ETa) and/or water storage for calibration. The research challenges lie in the situation that there can be significant errors in data derived from satellite products. Also, hydrological models designed for discharge simulation cannot necessarily function well when calibrated on other terms. This boils down to the research questions as follow:
1. How large are water balance data errors and to what extent can they be reduced?
2. How large are water balance model errors and to what extent can they be reduced?
3. To what extent does quantifying and reducing data and model errors eliminate trade-offs in fitting multiple datasets?
The strategy is constructed based on an error estimation and water balance data fusion method and the original and the advanced version of the Water Partition and Balance model (Wapaba), then tested in the Smoky Hill River catchment. When using unprocessed data, the water balance is not closed for the basin. The discharge simulation has the fitting precision index the Box-Cox transformed root mean squared error (TRMSE) in the range of 0.70 - 1.43 for different datasets and in calibration and validation period. Indexes of discharge fitting σma$ exceed 1.66. After closing the water balance with the mean time series of all fluxes, TRMSE decreases to 0.55. Considering data uncertainties, TRMSE is further declined to 0.29 and σ drops to 0.46. After that, the model structure is also improved. When using the modified model to calibrate on only ETa and TWSA for calibration (TRMSE for discharge = 0.87), the performance is similar to that of using the original Wapaba on all fluxes (TRMSE for discharge = 0.86). The fitting precision index σ for TWSA also decreases.
The research demonstrates the effectiveness of the data fusion method in correcting satellite time series and sheds light on the potential of application of this strategy in the ungauged area through the comparison of different calibration cases. After modification, the strategy is able to reproduce the flow regime, without using in-situ data, to the same degree as all three hydrological components (discharge, actual evapotranspiration and water storage) are used for calibration.
...
1. How large are water balance data errors and to what extent can they be reduced?
2. How large are water balance model errors and to what extent can they be reduced?
3. To what extent does quantifying and reducing data and model errors eliminate trade-offs in fitting multiple datasets?
The strategy is constructed based on an error estimation and water balance data fusion method and the original and the advanced version of the Water Partition and Balance model (Wapaba), then tested in the Smoky Hill River catchment. When using unprocessed data, the water balance is not closed for the basin. The discharge simulation has the fitting precision index the Box-Cox transformed root mean squared error (TRMSE) in the range of 0.70 - 1.43 for different datasets and in calibration and validation period. Indexes of discharge fitting σma$ exceed 1.66. After closing the water balance with the mean time series of all fluxes, TRMSE decreases to 0.55. Considering data uncertainties, TRMSE is further declined to 0.29 and σ drops to 0.46. After that, the model structure is also improved. When using the modified model to calibrate on only ETa and TWSA for calibration (TRMSE for discharge = 0.87), the performance is similar to that of using the original Wapaba on all fluxes (TRMSE for discharge = 0.86). The fitting precision index σ for TWSA also decreases.
The research demonstrates the effectiveness of the data fusion method in correcting satellite time series and sheds light on the potential of application of this strategy in the ungauged area through the comparison of different calibration cases. After modification, the strategy is able to reproduce the flow regime, without using in-situ data, to the same degree as all three hydrological components (discharge, actual evapotranspiration and water storage) are used for calibration.
...
The water balance model has been an indispensable tool for quantifying water supply and demand and regulating water resources for decades. By convention, hydrologists use in-situ measurements of river discharge for model calibration. Whereas hardly is it possible for areas suffering from data storage. Researchers have been working on methods dealing with this problem for many years. Satellite datasets are potential substitutes for the in-situ observation of hydrological variables. This research proposes a monthly strategy using satellite time series as model inputs. This research aims to apply this strategy in the absence of gauged data for discharge simulations and predictions. Instead of streamflow, the strategy uses variables like actual evapotranspiration (ETa) and/or water storage for calibration. The research challenges lie in the situation that there can be significant errors in data derived from satellite products. Also, hydrological models designed for discharge simulation cannot necessarily function well when calibrated on other terms. This boils down to the research questions as follow:
1. How large are water balance data errors and to what extent can they be reduced?
2. How large are water balance model errors and to what extent can they be reduced?
3. To what extent does quantifying and reducing data and model errors eliminate trade-offs in fitting multiple datasets?
The strategy is constructed based on an error estimation and water balance data fusion method and the original and the advanced version of the Water Partition and Balance model (Wapaba), then tested in the Smoky Hill River catchment. When using unprocessed data, the water balance is not closed for the basin. The discharge simulation has the fitting precision index the Box-Cox transformed root mean squared error (TRMSE) in the range of 0.70 - 1.43 for different datasets and in calibration and validation period. Indexes of discharge fitting σma$ exceed 1.66. After closing the water balance with the mean time series of all fluxes, TRMSE decreases to 0.55. Considering data uncertainties, TRMSE is further declined to 0.29 and σ drops to 0.46. After that, the model structure is also improved. When using the modified model to calibrate on only ETa and TWSA for calibration (TRMSE for discharge = 0.87), the performance is similar to that of using the original Wapaba on all fluxes (TRMSE for discharge = 0.86). The fitting precision index σ for TWSA also decreases.
The research demonstrates the effectiveness of the data fusion method in correcting satellite time series and sheds light on the potential of application of this strategy in the ungauged area through the comparison of different calibration cases. After modification, the strategy is able to reproduce the flow regime, without using in-situ data, to the same degree as all three hydrological components (discharge, actual evapotranspiration and water storage) are used for calibration.
1. How large are water balance data errors and to what extent can they be reduced?
2. How large are water balance model errors and to what extent can they be reduced?
3. To what extent does quantifying and reducing data and model errors eliminate trade-offs in fitting multiple datasets?
The strategy is constructed based on an error estimation and water balance data fusion method and the original and the advanced version of the Water Partition and Balance model (Wapaba), then tested in the Smoky Hill River catchment. When using unprocessed data, the water balance is not closed for the basin. The discharge simulation has the fitting precision index the Box-Cox transformed root mean squared error (TRMSE) in the range of 0.70 - 1.43 for different datasets and in calibration and validation period. Indexes of discharge fitting σma$ exceed 1.66. After closing the water balance with the mean time series of all fluxes, TRMSE decreases to 0.55. Considering data uncertainties, TRMSE is further declined to 0.29 and σ drops to 0.46. After that, the model structure is also improved. When using the modified model to calibrate on only ETa and TWSA for calibration (TRMSE for discharge = 0.87), the performance is similar to that of using the original Wapaba on all fluxes (TRMSE for discharge = 0.86). The fitting precision index σ for TWSA also decreases.
The research demonstrates the effectiveness of the data fusion method in correcting satellite time series and sheds light on the potential of application of this strategy in the ungauged area through the comparison of different calibration cases. After modification, the strategy is able to reproduce the flow regime, without using in-situ data, to the same degree as all three hydrological components (discharge, actual evapotranspiration and water storage) are used for calibration.
Master thesis
(2021)
-
A.G. Stampoultzidis, M.W. Ertsen, R.R.P. van Nooijen, Daniele Morandi Bonacossi
This thesis is the first attempt in modeling Sennacherib’s (Neo-Assyrian king) colossal watercourse networks, with its main goal to uncover their hypothetical functionalities and operation, regarding irrigation needs and harvest yield production. Their total length surpasses 150 Km and was completed in four stages, with lack of data forcing stage two’s exclusion. The stages modeled are separated in a Local and Regional system, the latter is assumed to connect the Zagros mountains foothills and Bandawai area with the capital of the era (early 6th century BCE) Nineveh, through a mix of artificial channels, canalized rivers, and streams. Two models were used AquaCrop and Sobek, with the former for crop and latter for flow simulations. Two feeder channel widths (1- 2 m), along with three inflow (“Wet”- “Reference”- “Dry”) and control (Absent, Maximum, Limited) choices are modeled with Sobek, adding up to 14 hydraulic scenarios. Noteworthy is that “Dry” year agriculture requires irrigation throughout the hole region and both seasons (Spring-Autumn), with results presenting around 60% gains in harvest amounts when control is applied for the Regional
System. Although navigation feasibility was not thoroughly explored, “Reference”
(and consequently “Wet”) year inflows show water depths rising to or higher
than the accepted. Concluding a decisive answer to the “archeological debate” of
Sennacherib’s motivation for construction of such massive infrastructure is impossible with present data (Environment, canal, social) available. Circumstances favoring control and therefore planned rural irrigation (rather than capital-centered) for Nineveh’s hinterlands are analyzed and discussed. Lastly a few interventions seen as valuable for further modeling studies were suggest towards the members of the LoNAP team as a contribution to their upcoming field survey. ...
System. Although navigation feasibility was not thoroughly explored, “Reference”
(and consequently “Wet”) year inflows show water depths rising to or higher
than the accepted. Concluding a decisive answer to the “archeological debate” of
Sennacherib’s motivation for construction of such massive infrastructure is impossible with present data (Environment, canal, social) available. Circumstances favoring control and therefore planned rural irrigation (rather than capital-centered) for Nineveh’s hinterlands are analyzed and discussed. Lastly a few interventions seen as valuable for further modeling studies were suggest towards the members of the LoNAP team as a contribution to their upcoming field survey. ...
This thesis is the first attempt in modeling Sennacherib’s (Neo-Assyrian king) colossal watercourse networks, with its main goal to uncover their hypothetical functionalities and operation, regarding irrigation needs and harvest yield production. Their total length surpasses 150 Km and was completed in four stages, with lack of data forcing stage two’s exclusion. The stages modeled are separated in a Local and Regional system, the latter is assumed to connect the Zagros mountains foothills and Bandawai area with the capital of the era (early 6th century BCE) Nineveh, through a mix of artificial channels, canalized rivers, and streams. Two models were used AquaCrop and Sobek, with the former for crop and latter for flow simulations. Two feeder channel widths (1- 2 m), along with three inflow (“Wet”- “Reference”- “Dry”) and control (Absent, Maximum, Limited) choices are modeled with Sobek, adding up to 14 hydraulic scenarios. Noteworthy is that “Dry” year agriculture requires irrigation throughout the hole region and both seasons (Spring-Autumn), with results presenting around 60% gains in harvest amounts when control is applied for the Regional
System. Although navigation feasibility was not thoroughly explored, “Reference”
(and consequently “Wet”) year inflows show water depths rising to or higher
than the accepted. Concluding a decisive answer to the “archeological debate” of
Sennacherib’s motivation for construction of such massive infrastructure is impossible with present data (Environment, canal, social) available. Circumstances favoring control and therefore planned rural irrigation (rather than capital-centered) for Nineveh’s hinterlands are analyzed and discussed. Lastly a few interventions seen as valuable for further modeling studies were suggest towards the members of the LoNAP team as a contribution to their upcoming field survey.
System. Although navigation feasibility was not thoroughly explored, “Reference”
(and consequently “Wet”) year inflows show water depths rising to or higher
than the accepted. Concluding a decisive answer to the “archeological debate” of
Sennacherib’s motivation for construction of such massive infrastructure is impossible with present data (Environment, canal, social) available. Circumstances favoring control and therefore planned rural irrigation (rather than capital-centered) for Nineveh’s hinterlands are analyzed and discussed. Lastly a few interventions seen as valuable for further modeling studies were suggest towards the members of the LoNAP team as a contribution to their upcoming field survey.
Optimizing the use of Delta21 for flood prone areas not protected by flood defenses at Dordrecht
A study on the operation strategy and reliability of a new flood protection system
Master thesis
(2021)
-
M. Buijs, M. Kok, M.Z. Voorendt, R.R.P. van Nooijen, T. Botterhuis, H. van Waveren
High-water levels in the Rhine-Meuse delta are going to rise in the upcoming decades due to climate change with the accompanied sea level rise and the increase of the frequency of extreme Rhine discharges. The flood prone areas not protected by flood defenses at the Island of Dordrecht are directly influenced by the increase of these high-water levels, leading to an increase in the future flood risk. To lower this risk, the Delta21 project has been proposed. By pumping water from the Haringvliet, the high-water levels in the delta are lowered. It is unclear if an optimal operation of Delta21 for the flood prone areas of Dordrecht exists and if the reliability of the project impacts this high-water level reduction.
The main objective of this report is to determine if the inclusion of Delta21 to the flood protection system of the Rhine-Meuse delta can provide a significant reduction of the flood risk of the flood prone areas at Dordrecht. Furthermore, it should be determined if an optimal operational scheme is possible in which the flood protection system with Delta21 can comply with all the flood requirements of the flood prone areas at Dordrecht and limitations to the Europoort barrier and Delta21 project while considering the reliability of the Delta21 project and the climate scenarios of the year 2100.
It was determined that the present flood risk at the flood prone areas at the Island of Dordrecht is equal to €110,000 per year. For the minimum, medium and maximum scenario for the year 2100 this risk increases to €390,000, €1,300,000 and €8,100,000 per year respectively. The implementation of the optimal operation of Delta21 reduces this risk with 23, 15 and 64 %. To obtain a flood risk reduction of 64 % for the maximum scenario, the closure frequency of the Europoort barrier, which may not be larger than three times per year, is equal to ten times per year. Furthermore, the flood risk as a percentage of the average annual income per household at the flood prone areas may not exceed 1 %, but for the medium onward this limit is exceeded at the historical harbor (city center). The maximum allowable probability of failure per pump of the pumping station and per siphon of the spillway of Delta21 is about 0.5 if the correlation between the components of both these systems is smaller than 0.9.
The inclusion of Delta21 in the current flood protection system with the present Europoort closure level can provide a significant reduction of the future flood risk of the flood prone areas at Dordrecht. However, it is not possible to create an optimal operational scheme for all scenarios of the year 2100 in which the flood protection system with Delta21 complies with the flood requirements of the flood prone areas and the limitations of the Europoort barrier. Finally, the reliability of the new Delta21 components is non-decisive for the flood risk assessments that have been made as long as the components are not fully dependent.
...
The main objective of this report is to determine if the inclusion of Delta21 to the flood protection system of the Rhine-Meuse delta can provide a significant reduction of the flood risk of the flood prone areas at Dordrecht. Furthermore, it should be determined if an optimal operational scheme is possible in which the flood protection system with Delta21 can comply with all the flood requirements of the flood prone areas at Dordrecht and limitations to the Europoort barrier and Delta21 project while considering the reliability of the Delta21 project and the climate scenarios of the year 2100.
It was determined that the present flood risk at the flood prone areas at the Island of Dordrecht is equal to €110,000 per year. For the minimum, medium and maximum scenario for the year 2100 this risk increases to €390,000, €1,300,000 and €8,100,000 per year respectively. The implementation of the optimal operation of Delta21 reduces this risk with 23, 15 and 64 %. To obtain a flood risk reduction of 64 % for the maximum scenario, the closure frequency of the Europoort barrier, which may not be larger than three times per year, is equal to ten times per year. Furthermore, the flood risk as a percentage of the average annual income per household at the flood prone areas may not exceed 1 %, but for the medium onward this limit is exceeded at the historical harbor (city center). The maximum allowable probability of failure per pump of the pumping station and per siphon of the spillway of Delta21 is about 0.5 if the correlation between the components of both these systems is smaller than 0.9.
The inclusion of Delta21 in the current flood protection system with the present Europoort closure level can provide a significant reduction of the future flood risk of the flood prone areas at Dordrecht. However, it is not possible to create an optimal operational scheme for all scenarios of the year 2100 in which the flood protection system with Delta21 complies with the flood requirements of the flood prone areas and the limitations of the Europoort barrier. Finally, the reliability of the new Delta21 components is non-decisive for the flood risk assessments that have been made as long as the components are not fully dependent.
...
High-water levels in the Rhine-Meuse delta are going to rise in the upcoming decades due to climate change with the accompanied sea level rise and the increase of the frequency of extreme Rhine discharges. The flood prone areas not protected by flood defenses at the Island of Dordrecht are directly influenced by the increase of these high-water levels, leading to an increase in the future flood risk. To lower this risk, the Delta21 project has been proposed. By pumping water from the Haringvliet, the high-water levels in the delta are lowered. It is unclear if an optimal operation of Delta21 for the flood prone areas of Dordrecht exists and if the reliability of the project impacts this high-water level reduction.
The main objective of this report is to determine if the inclusion of Delta21 to the flood protection system of the Rhine-Meuse delta can provide a significant reduction of the flood risk of the flood prone areas at Dordrecht. Furthermore, it should be determined if an optimal operational scheme is possible in which the flood protection system with Delta21 can comply with all the flood requirements of the flood prone areas at Dordrecht and limitations to the Europoort barrier and Delta21 project while considering the reliability of the Delta21 project and the climate scenarios of the year 2100.
It was determined that the present flood risk at the flood prone areas at the Island of Dordrecht is equal to €110,000 per year. For the minimum, medium and maximum scenario for the year 2100 this risk increases to €390,000, €1,300,000 and €8,100,000 per year respectively. The implementation of the optimal operation of Delta21 reduces this risk with 23, 15 and 64 %. To obtain a flood risk reduction of 64 % for the maximum scenario, the closure frequency of the Europoort barrier, which may not be larger than three times per year, is equal to ten times per year. Furthermore, the flood risk as a percentage of the average annual income per household at the flood prone areas may not exceed 1 %, but for the medium onward this limit is exceeded at the historical harbor (city center). The maximum allowable probability of failure per pump of the pumping station and per siphon of the spillway of Delta21 is about 0.5 if the correlation between the components of both these systems is smaller than 0.9.
The inclusion of Delta21 in the current flood protection system with the present Europoort closure level can provide a significant reduction of the future flood risk of the flood prone areas at Dordrecht. However, it is not possible to create an optimal operational scheme for all scenarios of the year 2100 in which the flood protection system with Delta21 complies with the flood requirements of the flood prone areas and the limitations of the Europoort barrier. Finally, the reliability of the new Delta21 components is non-decisive for the flood risk assessments that have been made as long as the components are not fully dependent.
The main objective of this report is to determine if the inclusion of Delta21 to the flood protection system of the Rhine-Meuse delta can provide a significant reduction of the flood risk of the flood prone areas at Dordrecht. Furthermore, it should be determined if an optimal operational scheme is possible in which the flood protection system with Delta21 can comply with all the flood requirements of the flood prone areas at Dordrecht and limitations to the Europoort barrier and Delta21 project while considering the reliability of the Delta21 project and the climate scenarios of the year 2100.
It was determined that the present flood risk at the flood prone areas at the Island of Dordrecht is equal to €110,000 per year. For the minimum, medium and maximum scenario for the year 2100 this risk increases to €390,000, €1,300,000 and €8,100,000 per year respectively. The implementation of the optimal operation of Delta21 reduces this risk with 23, 15 and 64 %. To obtain a flood risk reduction of 64 % for the maximum scenario, the closure frequency of the Europoort barrier, which may not be larger than three times per year, is equal to ten times per year. Furthermore, the flood risk as a percentage of the average annual income per household at the flood prone areas may not exceed 1 %, but for the medium onward this limit is exceeded at the historical harbor (city center). The maximum allowable probability of failure per pump of the pumping station and per siphon of the spillway of Delta21 is about 0.5 if the correlation between the components of both these systems is smaller than 0.9.
The inclusion of Delta21 in the current flood protection system with the present Europoort closure level can provide a significant reduction of the future flood risk of the flood prone areas at Dordrecht. However, it is not possible to create an optimal operational scheme for all scenarios of the year 2100 in which the flood protection system with Delta21 complies with the flood requirements of the flood prone areas and the limitations of the Europoort barrier. Finally, the reliability of the new Delta21 components is non-decisive for the flood risk assessments that have been made as long as the components are not fully dependent.
The variability of the rootzone storage capacity in Austria
An exploration of its controls
The rootzone storage capacity (Sr) is a crucial part of the hydrological cycle. This storage provides water access for vegetation, in order to meet the atmospheric water demand through transpiration. The spatio-temporal variability of Sr is not well represented in current hydrological and climate models. The root zone storage capacity receives rapidly increasing interest from scientists. Recent studies developed a climate-based method to determine Sr. This method is based on the insight that ecosystems efficiently adapt their rootzone storage capacity to survive a drought with a certain return period. However, this method requires a vast amount of data. Long hydrological time-series with a rather fine temporal resolution are required. These time-series are not always available for many poorly gauged catchments. Therefore, it is important to explore what is exactly controlling this Sr, and if we can eventually predict it. This study aims to describe the spatial and temporal variability of Sr with a combination of climate and land cover variables in Austria for the study period 1982 - 2008. To anticipate on expected snowfall, a snowfall module is included and calibrated with the use of a MODIS satellite snow cover product. The most important climate and land cover variables are identified, using multiple linear regression analysis. The best performing regression models are selected with a diverse combination of variables. This is done by comparing 21 catchments across the central and eastern part of Austria. Additionally, a stationary time-series split method is used to explore how Sr is changing in time. Subsequently, another multiple linear regression analysis is performed to explore the controls of the dynamics of Sr for these catchments. According to this study, the Run-off coefficient describes Sr best in all studied regression models. A multiple linear regression model compiled out of the Run-off coefficient and the seasonality index performed best with an R2adj of 0.8. The seasonality index seems to be specific for this study since the highest fraction of precipitation and evaporation coincides in summer. Land cover seems of less importance for the estimation of Sr. However, no conclusion could be drawn for the importance of land cover types in the regression analysis considering the disputable applicability of the land cover data. Furthermore, the relations of fractional cropland cover and fractional forest cover with Sr are in contradiction with current literature. Apart from the spatial relationships, it is discovered that on average increased from the year 1992 onwards. However, no indisputable explanation is encountered (R2adj 0.55). The decreasing Run-off coefficient explains most of the increase in Sr. No conclusions could be drawn on the influence of land cover change on Sr, caused by an irregular land cover time-series. Since the catchments in this study are rather humid and have similar seasonal patterns, it would be interesting to investigate if the discovered relationships are also valid for more arid and seasonal varying catchments. Also, it would be useful to investigate the unexpected relationship between land cover and Sr further. ...
The rootzone storage capacity (Sr) is a crucial part of the hydrological cycle. This storage provides water access for vegetation, in order to meet the atmospheric water demand through transpiration. The spatio-temporal variability of Sr is not well represented in current hydrological and climate models. The root zone storage capacity receives rapidly increasing interest from scientists. Recent studies developed a climate-based method to determine Sr. This method is based on the insight that ecosystems efficiently adapt their rootzone storage capacity to survive a drought with a certain return period. However, this method requires a vast amount of data. Long hydrological time-series with a rather fine temporal resolution are required. These time-series are not always available for many poorly gauged catchments. Therefore, it is important to explore what is exactly controlling this Sr, and if we can eventually predict it. This study aims to describe the spatial and temporal variability of Sr with a combination of climate and land cover variables in Austria for the study period 1982 - 2008. To anticipate on expected snowfall, a snowfall module is included and calibrated with the use of a MODIS satellite snow cover product. The most important climate and land cover variables are identified, using multiple linear regression analysis. The best performing regression models are selected with a diverse combination of variables. This is done by comparing 21 catchments across the central and eastern part of Austria. Additionally, a stationary time-series split method is used to explore how Sr is changing in time. Subsequently, another multiple linear regression analysis is performed to explore the controls of the dynamics of Sr for these catchments. According to this study, the Run-off coefficient describes Sr best in all studied regression models. A multiple linear regression model compiled out of the Run-off coefficient and the seasonality index performed best with an R2adj of 0.8. The seasonality index seems to be specific for this study since the highest fraction of precipitation and evaporation coincides in summer. Land cover seems of less importance for the estimation of Sr. However, no conclusion could be drawn for the importance of land cover types in the regression analysis considering the disputable applicability of the land cover data. Furthermore, the relations of fractional cropland cover and fractional forest cover with Sr are in contradiction with current literature. Apart from the spatial relationships, it is discovered that on average increased from the year 1992 onwards. However, no indisputable explanation is encountered (R2adj 0.55). The decreasing Run-off coefficient explains most of the increase in Sr. No conclusions could be drawn on the influence of land cover change on Sr, caused by an irregular land cover time-series. Since the catchments in this study are rather humid and have similar seasonal patterns, it would be interesting to investigate if the discovered relationships are also valid for more arid and seasonal varying catchments. Also, it would be useful to investigate the unexpected relationship between land cover and Sr further.
The Canals Never Lie
MASSCOTE framework and modelling of operation strategies in the Gezira Irrigation Scheme
The Gezira irrigation scheme, in Sudan, stretches out south of Khartoum between the Blue Nile and the While Nile Rivers. For more than 95 years since its establishment in 1925, the scheme has been a significant asset to Sudan’s economy. It is considered one of the oldest and largest schemes in the world being served by one management body using surface gravity irrigation from a single source. For many years, the system has performed with high efficiency and was considered a good model example, both locally and internationally. In the last few decades the scheme has deteriorated. The water productivity records show declining trends, while more water is being withdrawn from Sennar dam. Farmers are complaining about water shortages and some fields are not reached at all. Authorities state severe damages in the system’s infrastructure, with about 85% of the system’s hydraulic works being damaged. The performance of operation is well below the required efficiency. The sedimentation issue has always been present in the system, especially in the minor canals, but has been severe in recent years. As a result, farmers tend to interfere in the operation of the system, as they are not satisfied with the management authorities. Therefore, the scheme’s operation situations changed accordingly, shifting from the original design approach into new – often undesired – situations. There have been many studies carried in the Gezira Irrigation scheme, in order to understand the reasons behind this deterioration and find sustainable solutions. Starting from the causes of the sedimentation issue and maintenance works, generating new methods of calculating the crop water requirements using remote sensing, influence of farmers’ practices on field level, change of management and institutions throughout the history, and many others. Yet, there is no sensible change seen on the ground. The main objective of this thesis is contribute to the researches done in the Gezira scheme by providing an understanding of how the canalization system’s respond to various water demands strategies. In recent years, there has been a noticeable improvement in methods used to determine crop water requirements using remote sensing. Coping with these improvements, a main question rises, to what extent these remote sensing approaches could be implemented in large irrigation schemes, taking the Gezira scheme as a case study. Determining water requirements is one thing, delivering the water through the canal system could be something else. In order to answer the above question, we first need to analyze the system we are studying. This step was carried through analyzing the current, general, performance of the Gezira scheme and comparing it to the initial design characteristics of the scheme using the MASSCOTE framework (FAO). Using the outcomes of the MASSCOTE evaluation, the thesis proceeded by generating a computer model of the canalization system, taking a major and minor canal as a representation for the water distribution system, and testing different water demands scenarios. This step provided insights into the operation of the canalization system on the level of major and minor canals. It was found that, within the boundaries of our model, implementing new remote sensing methods is practicable, assuming water is readily available at the offtake of the major canal. This conclusion takes into consideration the lengths of the canals and their locations regarding the overall scheme and the carrying capacity of each canals.
...
The Gezira irrigation scheme, in Sudan, stretches out south of Khartoum between the Blue Nile and the While Nile Rivers. For more than 95 years since its establishment in 1925, the scheme has been a significant asset to Sudan’s economy. It is considered one of the oldest and largest schemes in the world being served by one management body using surface gravity irrigation from a single source. For many years, the system has performed with high efficiency and was considered a good model example, both locally and internationally. In the last few decades the scheme has deteriorated. The water productivity records show declining trends, while more water is being withdrawn from Sennar dam. Farmers are complaining about water shortages and some fields are not reached at all. Authorities state severe damages in the system’s infrastructure, with about 85% of the system’s hydraulic works being damaged. The performance of operation is well below the required efficiency. The sedimentation issue has always been present in the system, especially in the minor canals, but has been severe in recent years. As a result, farmers tend to interfere in the operation of the system, as they are not satisfied with the management authorities. Therefore, the scheme’s operation situations changed accordingly, shifting from the original design approach into new – often undesired – situations. There have been many studies carried in the Gezira Irrigation scheme, in order to understand the reasons behind this deterioration and find sustainable solutions. Starting from the causes of the sedimentation issue and maintenance works, generating new methods of calculating the crop water requirements using remote sensing, influence of farmers’ practices on field level, change of management and institutions throughout the history, and many others. Yet, there is no sensible change seen on the ground. The main objective of this thesis is contribute to the researches done in the Gezira scheme by providing an understanding of how the canalization system’s respond to various water demands strategies. In recent years, there has been a noticeable improvement in methods used to determine crop water requirements using remote sensing. Coping with these improvements, a main question rises, to what extent these remote sensing approaches could be implemented in large irrigation schemes, taking the Gezira scheme as a case study. Determining water requirements is one thing, delivering the water through the canal system could be something else. In order to answer the above question, we first need to analyze the system we are studying. This step was carried through analyzing the current, general, performance of the Gezira scheme and comparing it to the initial design characteristics of the scheme using the MASSCOTE framework (FAO). Using the outcomes of the MASSCOTE evaluation, the thesis proceeded by generating a computer model of the canalization system, taking a major and minor canal as a representation for the water distribution system, and testing different water demands scenarios. This step provided insights into the operation of the canalization system on the level of major and minor canals. It was found that, within the boundaries of our model, implementing new remote sensing methods is practicable, assuming water is readily available at the offtake of the major canal. This conclusion takes into consideration the lengths of the canals and their locations regarding the overall scheme and the carrying capacity of each canals.
Gaussian Process Regression Models for Predicting Water Retention Curves
Application of Machine Learning Techniques for Modelling Uncertainty in Hydraulic Curves
An accurate representation of water retention curves is important for various reasons. Traditional models already exist for the representation of these curves, with one of them being the van Genuchten model. When soil parameters are available, the van Genuchten model can be used to plot water retention curves. However, when these soil parameters are not available, regression can be performed to estimate and predict the water retention curves. Commonly, the Non-Linear Least Squares regression method is used in combination with a certain water retention model. Problems arise for inhomogeneous soils as the traditional water retention models tend to break down. To improve the representation of water retention curves, Gaussian Process regression will be implemented. This method will be combined with the Non-Linear Least Squares method to obtain new representations of water retention curves. These new curves are better in terms of curve fit and uncertainty, when compared to the traditional method. These comparisons can be made visually, by observing the plots and their confidence intervals, as well as quantitatively by computing the log-likelihoods of the different methods. When comparing the results of the log-likelihood computations for both methods, it follows that the value of the log-likelihood is greater for water retention data with correlated residuals. In the case where the residuals are uncorrelated, the log-likelihoods are equal for both methods and no improvements are observed.
...
An accurate representation of water retention curves is important for various reasons. Traditional models already exist for the representation of these curves, with one of them being the van Genuchten model. When soil parameters are available, the van Genuchten model can be used to plot water retention curves. However, when these soil parameters are not available, regression can be performed to estimate and predict the water retention curves. Commonly, the Non-Linear Least Squares regression method is used in combination with a certain water retention model. Problems arise for inhomogeneous soils as the traditional water retention models tend to break down. To improve the representation of water retention curves, Gaussian Process regression will be implemented. This method will be combined with the Non-Linear Least Squares method to obtain new representations of water retention curves. These new curves are better in terms of curve fit and uncertainty, when compared to the traditional method. These comparisons can be made visually, by observing the plots and their confidence intervals, as well as quantitatively by computing the log-likelihoods of the different methods. When comparing the results of the log-likelihood computations for both methods, it follows that the value of the log-likelihood is greater for water retention data with correlated residuals. In the case where the residuals are uncorrelated, the log-likelihoods are equal for both methods and no improvements are observed.
The thermal recovery efficiency of High Temperature Aquifer Thermal Energy Storage (HT-ATES) systems can be limited due to the effect of buoyancy flow of the injected hot water. This thesis has researched the application of a Multiple Partially Penetrating Wells (MPPWs) as a well design method to counteract the effect of buoyancy flow and improve the performance of HT-ATES systems (>60°C). A MPPW is a well with more than one screen that allows injection and extraction of water at different depths in the aquifer.This method to counteract the effect of buoyancy flow was tested through numerical modelling with SEAWATv4. The modelled HT-ATES systems were running for four recovery cycles each including injection-storage-extraction-rest phases. The thermal recovery efficiency was determined over these cycles for 7 different scenarios and four different cases: a) A regular HT-ATES system with a fully penetrating screen, b) a regular HT-ATES system where buoyancy flow is neglected, c) an HT-ATES system with a MPPW with two screens, d) an HT-ATES system with four screens. The latter case was tested for three different control approaches based on data from four different locations in the aquifer.For the reference scenario where 90°C water was injected, a regular HT-ATES system had a thermal recovery efficiency of 0.61. With the application of MPPWs for both two or four screens this was 0.81 in the fourth recovery cycle, which approaches the case without buoyancy which had a thermal recovery efficiency of 0.88. The application of two or four screens did not show significant difference in thermal recovery efficiency after the first recovery cycle.A sensitivity analysis showed that the absolute increase in thermal recovery efficiency of an HT-ATES system with a MPPW compared to a regular HT-ATES is higher for larger buoyancy flow (i.e, high injection temperature and high (vertical) hydraulic conductivity), smaller injection volume and larger aquifer thickness. An applicability analysis showed that application of MPPWs is beneficial if the buoyancy flow (which is defined as the vertical hydraulic conductivity times the density ratio of the ambient groundwater and injected water) is greater than 0.1 meters per day.
...
The thermal recovery efficiency of High Temperature Aquifer Thermal Energy Storage (HT-ATES) systems can be limited due to the effect of buoyancy flow of the injected hot water. This thesis has researched the application of a Multiple Partially Penetrating Wells (MPPWs) as a well design method to counteract the effect of buoyancy flow and improve the performance of HT-ATES systems (>60°C). A MPPW is a well with more than one screen that allows injection and extraction of water at different depths in the aquifer.This method to counteract the effect of buoyancy flow was tested through numerical modelling with SEAWATv4. The modelled HT-ATES systems were running for four recovery cycles each including injection-storage-extraction-rest phases. The thermal recovery efficiency was determined over these cycles for 7 different scenarios and four different cases: a) A regular HT-ATES system with a fully penetrating screen, b) a regular HT-ATES system where buoyancy flow is neglected, c) an HT-ATES system with a MPPW with two screens, d) an HT-ATES system with four screens. The latter case was tested for three different control approaches based on data from four different locations in the aquifer.For the reference scenario where 90°C water was injected, a regular HT-ATES system had a thermal recovery efficiency of 0.61. With the application of MPPWs for both two or four screens this was 0.81 in the fourth recovery cycle, which approaches the case without buoyancy which had a thermal recovery efficiency of 0.88. The application of two or four screens did not show significant difference in thermal recovery efficiency after the first recovery cycle.A sensitivity analysis showed that the absolute increase in thermal recovery efficiency of an HT-ATES system with a MPPW compared to a regular HT-ATES is higher for larger buoyancy flow (i.e, high injection temperature and high (vertical) hydraulic conductivity), smaller injection volume and larger aquifer thickness. An applicability analysis showed that application of MPPWs is beneficial if the buoyancy flow (which is defined as the vertical hydraulic conductivity times the density ratio of the ambient groundwater and injected water) is greater than 0.1 meters per day.
Design of an adaptive weir
A case study of the replacement of weir Belfeld
In contrast with other Dutch rivers parts, human measures in the upstream part of the River Meuse did not mainly focus on discharging surplus water, but on retaining water in dry periods. Almost 100 years ago seven weirs were constructed in the River Meuse to enable transport of coals. The structures reach, due to concrete degradation, the end of their technical lifetime and the manual operation does not meet the current ARBO-legislation; both make replacement of the weirs required. Rijkswaterstaat, asset owner of the waterways and weirs, invited the civil engineering sector to collectively develop weir replacement strategies in 2015. The future developments of the river and surroundings were an important subject during the meetings. With this in mind, during the co-creation meetings under the title ‘Grip op de Maas’, one of the proposed weir replacement strategies was called the Adaptive Meuse (De Bouwcampus, 2015). This perspective took the future uncertainty into account by proposing adaptive designing of the weirs. This study builds on this perspective by the objective of designing an adaptive weir in the River Meuse according to the approach of adaptive delta management.This approach states that designs have to be flexible and able to switch between multiple strategies for future challenges concerning flood safety and freshwater storage. In this study, an overview of the measures required to deal with the development of specific purposes is provided by adaptation schemes. When and if a specific purpose applies, depends on which of the four Dutch delta scenarios, DRUK, STOOM, RUST and WARM, evolves. These scenarios are based on a unique combination of the rate of climate change and socioeconomic developments (Wolters, Van den Born, Dammers, & Reinhard, 2018). Adaptivity is obtained by regional adaptation measures over a particular stretch of the river and by weir adaptation measures. To address both the adaptivity of the river and the weirs, three design levels have been established. These levels are summed below:The global design level comprises the total dammed section of the Dutch River Meuse and two weir sections in Belgium. The series of weirs in the river still suffices the requirements; on this scale, no large adaptations have to be made now or in the future.The regional design level comprises the weir sections Roermond and Belfeld, since the adaptivity of these sections is the largest of all weir sections in the global design area.The local design level addresses the geometric design of weir Belfeld itself. The adaptivity of the designed weir enables the discard of regional adaptation measures with undesired implications.Since both dehydration and permanent flooding of the river valley have to be prevented and the water distribution over Belgium and the Netherlands is fixed in an agreement, the global adaptivity is limited. Weir removal or replacement of the current weir by a new weir at a different location is in most weir sections infeasible. Weir section Roermond forms an exception, since the commercial navigation uses the later dug parallel Lateral Canal, part of weir section Belfeld. Thus, the majority of weir section Roermond can be restricted to only recreational vessels. By modifications of weir Belfeld, it could possibly (partly) take over the functions of weir Roermond in future. In 2030 however, one-to-one weir replacement is selected to avert significant changes in groundwater table.For the regional design level, the future developments are split into the flexibility to groundwater changes, the use of freshwater for drinking water production, agriculture and industrial activities, the discharge of flood waves and the navigation on the River Meuse and to the port of Roermond. An adaptation scheme indicates what regional adaptation measures and weir adaptation measures are required to serve the purposes per time period in each of the four delta scenarios. On basis of this adaptation scheme, it is concluded that by designing an adaptive weir, measures along the entire weir section, for instance raising embankments or raising bridges, can be discarded in the future. The freedom of choice of the future waterway manager is preserved by an adaptive weir. In the last design level, an adaptive weir is proposed to replace weir Belfeld. The requirements to and the design of the adaptive weir are based on the regional adaptation scheme. The weir can be adapted by:constructing additional weir openings at the eastern embankment, enabled by the location of the new weir, which is just a couple of hundreds meter upstream of the current weir.adjusting the management of the weir gates, enabled by the choice of radial gates. The location in the weir section at which the dammed water level is independent of the river discharge can be shifted throughout the weir section. This allows the air clearance and the water level dynamics to be managed for container transport and ecological development, respectively. heightening the dammed water level, also enabled by the choice of radial gates. To withstand the larger water head, the initial investment increases by 30%. However, regional measures are saved in the future. Only the gates and height of the superstructure have to be adapted in the future to heighten the dammed water level.In conclusion, on global level, a change of the locations and numbers of weirs in the River Meuse is presently not desired and required, since the functions are met with the current weir layout. To guarantee this in the entire upcoming century, adaptations are required in the weir section Belfeld in the River Meuse. The resulting adaptation scheme after construction of the proposed adaptive weir presents the regional and weir adaptation measures that serve the mentioned scenario-dependent purposes. By the large adaptivity of proposed weir design, regional adaptation measures along the river with undesired implications can be discarded or minimized. Only if higher container vessels have to be accommodated on the Meuse River or the accessibility of the Prins Willem-Alexanderport has to be improved, regional measures are inevitable. The method used in this report can be used to set up adaptation schemes for all weir sections in the River Meuse. By involving Rijkswaterstaat and other stakeholders, the schemes can be turned into quantitative ones. Last, it is recommended to start structural calculations on the adaptive weir Belfeld.
...
In contrast with other Dutch rivers parts, human measures in the upstream part of the River Meuse did not mainly focus on discharging surplus water, but on retaining water in dry periods. Almost 100 years ago seven weirs were constructed in the River Meuse to enable transport of coals. The structures reach, due to concrete degradation, the end of their technical lifetime and the manual operation does not meet the current ARBO-legislation; both make replacement of the weirs required. Rijkswaterstaat, asset owner of the waterways and weirs, invited the civil engineering sector to collectively develop weir replacement strategies in 2015. The future developments of the river and surroundings were an important subject during the meetings. With this in mind, during the co-creation meetings under the title ‘Grip op de Maas’, one of the proposed weir replacement strategies was called the Adaptive Meuse (De Bouwcampus, 2015). This perspective took the future uncertainty into account by proposing adaptive designing of the weirs. This study builds on this perspective by the objective of designing an adaptive weir in the River Meuse according to the approach of adaptive delta management.This approach states that designs have to be flexible and able to switch between multiple strategies for future challenges concerning flood safety and freshwater storage. In this study, an overview of the measures required to deal with the development of specific purposes is provided by adaptation schemes. When and if a specific purpose applies, depends on which of the four Dutch delta scenarios, DRUK, STOOM, RUST and WARM, evolves. These scenarios are based on a unique combination of the rate of climate change and socioeconomic developments (Wolters, Van den Born, Dammers, & Reinhard, 2018). Adaptivity is obtained by regional adaptation measures over a particular stretch of the river and by weir adaptation measures. To address both the adaptivity of the river and the weirs, three design levels have been established. These levels are summed below:The global design level comprises the total dammed section of the Dutch River Meuse and two weir sections in Belgium. The series of weirs in the river still suffices the requirements; on this scale, no large adaptations have to be made now or in the future.The regional design level comprises the weir sections Roermond and Belfeld, since the adaptivity of these sections is the largest of all weir sections in the global design area.The local design level addresses the geometric design of weir Belfeld itself. The adaptivity of the designed weir enables the discard of regional adaptation measures with undesired implications.Since both dehydration and permanent flooding of the river valley have to be prevented and the water distribution over Belgium and the Netherlands is fixed in an agreement, the global adaptivity is limited. Weir removal or replacement of the current weir by a new weir at a different location is in most weir sections infeasible. Weir section Roermond forms an exception, since the commercial navigation uses the later dug parallel Lateral Canal, part of weir section Belfeld. Thus, the majority of weir section Roermond can be restricted to only recreational vessels. By modifications of weir Belfeld, it could possibly (partly) take over the functions of weir Roermond in future. In 2030 however, one-to-one weir replacement is selected to avert significant changes in groundwater table.For the regional design level, the future developments are split into the flexibility to groundwater changes, the use of freshwater for drinking water production, agriculture and industrial activities, the discharge of flood waves and the navigation on the River Meuse and to the port of Roermond. An adaptation scheme indicates what regional adaptation measures and weir adaptation measures are required to serve the purposes per time period in each of the four delta scenarios. On basis of this adaptation scheme, it is concluded that by designing an adaptive weir, measures along the entire weir section, for instance raising embankments or raising bridges, can be discarded in the future. The freedom of choice of the future waterway manager is preserved by an adaptive weir. In the last design level, an adaptive weir is proposed to replace weir Belfeld. The requirements to and the design of the adaptive weir are based on the regional adaptation scheme. The weir can be adapted by:constructing additional weir openings at the eastern embankment, enabled by the location of the new weir, which is just a couple of hundreds meter upstream of the current weir.adjusting the management of the weir gates, enabled by the choice of radial gates. The location in the weir section at which the dammed water level is independent of the river discharge can be shifted throughout the weir section. This allows the air clearance and the water level dynamics to be managed for container transport and ecological development, respectively. heightening the dammed water level, also enabled by the choice of radial gates. To withstand the larger water head, the initial investment increases by 30%. However, regional measures are saved in the future. Only the gates and height of the superstructure have to be adapted in the future to heighten the dammed water level.In conclusion, on global level, a change of the locations and numbers of weirs in the River Meuse is presently not desired and required, since the functions are met with the current weir layout. To guarantee this in the entire upcoming century, adaptations are required in the weir section Belfeld in the River Meuse. The resulting adaptation scheme after construction of the proposed adaptive weir presents the regional and weir adaptation measures that serve the mentioned scenario-dependent purposes. By the large adaptivity of proposed weir design, regional adaptation measures along the river with undesired implications can be discarded or minimized. Only if higher container vessels have to be accommodated on the Meuse River or the accessibility of the Prins Willem-Alexanderport has to be improved, regional measures are inevitable. The method used in this report can be used to set up adaptation schemes for all weir sections in the River Meuse. By involving Rijkswaterstaat and other stakeholders, the schemes can be turned into quantitative ones. Last, it is recommended to start structural calculations on the adaptive weir Belfeld.