The correct representation of depth-induced wave breaking is important for understanding coastal ... more The correct representation of depth-induced wave breaking is important for understanding coastal morphology and for design andmanagement in the coastal zone. Although numerous studies have demonstrated the applicability of a constant scaling of the Battjes and Janssen (1978) dissipationmodel for depth-induced breaking, recent studies have shown its inability to sufficiently reproduce wave dissipation over complex field cases. In the present study, we contrast the application of such a constant scaling to two alternative wave breaking parameterizationswith a variable scaling based on either thewave nonlinearity (the φparameterization) oronbothbottomslope and normalizedwavelength supplementedwithwave directionality (the β−kd parameterization).We consider three field data sets characteristic of a simple beach-bar profile, a bay partially protected by a shoal and a complex intertidal region.We demonstrate that in these cases the β−kd parameterization provides a better alternative to the use of a constant scaling or the φ parameterization. To illustrate the operational consequences, we up-scale the conditions over the case of the intertidal region to correspond to design conditions for the Dutch coast (storm conditions with a 4000 year return period). Under these extreme conditions, for locally generated waves both the β−kd and φ parameterizations predict qualitatively similar increased significant wave heights but the β−kd parameterization increased the waves twice as much as the φ parameterization. Under other conditions, when non-locally generated waves (swell) dissipates over a gently sloping bottom, the β−kd parameterization predicts lower significant wave heights compared to either the constant scaling or φ parameterization.
Wave breaking in shallow water is still poorly understood and needs to be better parameterized in... more Wave breaking in shallow water is still poorly understood and needs to be better parameterized in 2D spectral wave models. Significant wave heights over horizontal bathymetries are typically under-predicted in locally generated wave conditions and over-predicted in non-locally generated conditions. A joint scaling dependent on both local bottom slope and normalized wave number is presented and is shown to resolve these issues. Compared to the 12 wave breaking parameterizations considered in this study, this joint scaling demonstrates significant improvements, up to 50% error reduction, over 1D horizontal bathymetries for both locally and non-locally generated waves. In order to account for the inherent differences between uni-directional (1D) and directionally spread (2D) wave conditions, an extension of the wave breaking dissipation models is presented. By including the effects of wave directionality, rmserrors for the significant wave height are reduced for the best performing parameterizations in conditions with strong directional spreading. With this extension, our joint scaling improves modeling skill for significant wave heights over a verification data set of 11 different 1D laboratory bathymetries, 3 shallow lakes and 4 coastal sites. The corresponding averaged normalized rms-error for significant wave height in the 2D cases varied between 8% and 27%. In comparison, using the default setting with a constant scaling, as used in most presently operating 2D spectral wave models, gave equivalent errors between 15% and 38%.
This paper presents the application of new source terms in SWAN for the dominant water wave physi... more This paper presents the application of new source terms in SWAN for the dominant water wave physics in the coastal zone: depth-induced breaking and triad wave-wave interactions. We present results demonstrating increased modelling skill in the prediction of bulk wave parameters e.g. significant wave height and of the spectral shape compared to currently used defaults, particularly in cases with horizontal bathymetries. These preliminary results suggest a greater range of applicability of these source terms for operational applications.
The scaling of the energy dissipation of random, short-crested waves due to depth-induced breakin... more The scaling of the energy dissipation of random, short-crested waves due to depth-induced breaking is investigated with a numerical model and a large number of laboratory observations and generalized lake observations. The scaling is found to depend on the normalized water depth kd (where d is water depth and k is an average wave number) and bottom slope n . The model has been supplemented with a relaxation model to represent the persistence of wave breaking, in particular at the steep edge of reefs. The n-kd scaling does not improve the results for cases with gently sloping bathymetry (between 1:30 and 1:100, say; compared to using a commonly used fixed scaling). It does improve the results for flat bottom cases (lakes and reefs), reducing the rms-error in the significant wave height for these cases by over 50%.
The correct representation of depth-induced wave breaking is important for understanding coastal ... more The correct representation of depth-induced wave breaking is important for understanding coastal morphology and for design andmanagement in the coastal zone. Although numerous studies have demonstrated the applicability of a constant scaling of the Battjes and Janssen (1978) dissipationmodel for depth-induced breaking, recent studies have shown its inability to sufficiently reproduce wave dissipation over complex field cases. In the present study, we contrast the application of such a constant scaling to two alternative wave breaking parameterizationswith a variable scaling based on either thewave nonlinearity (the φparameterization) oronbothbottomslope and normalizedwavelength supplementedwithwave directionality (the β−kd parameterization).We consider three field data sets characteristic of a simple beach-bar profile, a bay partially protected by a shoal and a complex intertidal region.We demonstrate that in these cases the β−kd parameterization provides a better alternative to the use of a constant scaling or the φ parameterization. To illustrate the operational consequences, we up-scale the conditions over the case of the intertidal region to correspond to design conditions for the Dutch coast (storm conditions with a 4000 year return period). Under these extreme conditions, for locally generated waves both the β−kd and φ parameterizations predict qualitatively similar increased significant wave heights but the β−kd parameterization increased the waves twice as much as the φ parameterization. Under other conditions, when non-locally generated waves (swell) dissipates over a gently sloping bottom, the β−kd parameterization predicts lower significant wave heights compared to either the constant scaling or φ parameterization.
Wave breaking in shallow water is still poorly understood and needs to be better parameterized in... more Wave breaking in shallow water is still poorly understood and needs to be better parameterized in 2D spectral wave models. Significant wave heights over horizontal bathymetries are typically under-predicted in locally generated wave conditions and over-predicted in non-locally generated conditions. A joint scaling dependent on both local bottom slope and normalized wave number is presented and is shown to resolve these issues. Compared to the 12 wave breaking parameterizations considered in this study, this joint scaling demonstrates significant improvements, up to 50% error reduction, over 1D horizontal bathymetries for both locally and non-locally generated waves. In order to account for the inherent differences between uni-directional (1D) and directionally spread (2D) wave conditions, an extension of the wave breaking dissipation models is presented. By including the effects of wave directionality, rmserrors for the significant wave height are reduced for the best performing parameterizations in conditions with strong directional spreading. With this extension, our joint scaling improves modeling skill for significant wave heights over a verification data set of 11 different 1D laboratory bathymetries, 3 shallow lakes and 4 coastal sites. The corresponding averaged normalized rms-error for significant wave height in the 2D cases varied between 8% and 27%. In comparison, using the default setting with a constant scaling, as used in most presently operating 2D spectral wave models, gave equivalent errors between 15% and 38%.
This paper presents the application of new source terms in SWAN for the dominant water wave physi... more This paper presents the application of new source terms in SWAN for the dominant water wave physics in the coastal zone: depth-induced breaking and triad wave-wave interactions. We present results demonstrating increased modelling skill in the prediction of bulk wave parameters e.g. significant wave height and of the spectral shape compared to currently used defaults, particularly in cases with horizontal bathymetries. These preliminary results suggest a greater range of applicability of these source terms for operational applications.
The scaling of the energy dissipation of random, short-crested waves due to depth-induced breakin... more The scaling of the energy dissipation of random, short-crested waves due to depth-induced breaking is investigated with a numerical model and a large number of laboratory observations and generalized lake observations. The scaling is found to depend on the normalized water depth kd (where d is water depth and k is an average wave number) and bottom slope n . The model has been supplemented with a relaxation model to represent the persistence of wave breaking, in particular at the steep edge of reefs. The n-kd scaling does not improve the results for cases with gently sloping bathymetry (between 1:30 and 1:100, say; compared to using a commonly used fixed scaling). It does improve the results for flat bottom cases (lakes and reefs), reducing the rms-error in the significant wave height for these cases by over 50%.
Uploads
Papers by James Salmon
and for design andmanagement in the coastal zone. Although numerous studies have demonstrated the applicability
of a constant scaling of the Battjes and Janssen (1978) dissipationmodel for depth-induced breaking, recent
studies have shown its inability to sufficiently reproduce wave dissipation over complex field cases. In the
present study, we contrast the application of such a constant scaling to two alternative wave breaking parameterizationswith
a variable scaling based on either thewave nonlinearity (the φparameterization) oronbothbottomslope
and normalizedwavelength supplementedwithwave directionality (the β−kd parameterization).We
consider three field data sets characteristic of a simple beach-bar profile, a bay partially protected by a shoal and a
complex intertidal region.We demonstrate that in these cases the β−kd parameterization provides a better alternative
to the use of a constant scaling or the φ parameterization. To illustrate the operational consequences,
we up-scale the conditions over the case of the intertidal region to correspond to design conditions for the
Dutch coast (storm conditions with a 4000 year return period). Under these extreme conditions, for locally generated
waves both the β−kd and φ parameterizations predict qualitatively similar increased significant wave
heights but the β−kd parameterization increased the waves twice as much as the φ parameterization. Under
other conditions, when non-locally generated waves (swell) dissipates over a gently sloping bottom, the β−kd
parameterization predicts lower significant wave heights compared to either the constant scaling or φ
parameterization.
spectral wave models. Significant wave heights over horizontal bathymetries are typically under-predicted
in locally generated wave conditions and over-predicted in non-locally generated conditions. A
joint scaling dependent on both local bottom slope and normalized wave number is presented and is
shown to resolve these issues. Compared to the 12 wave breaking parameterizations considered in this
study, this joint scaling demonstrates significant improvements, up to 50% error reduction, over 1D horizontal
bathymetries for both locally and non-locally generated waves. In order to account for the inherent
differences between uni-directional (1D) and directionally spread (2D) wave conditions, an extension
of the wave breaking dissipation models is presented. By including the effects of wave directionality, rmserrors
for the significant wave height are reduced for the best performing parameterizations in conditions
with strong directional spreading. With this extension, our joint scaling improves modeling skill for significant
wave heights over a verification data set of 11 different 1D laboratory bathymetries, 3 shallow
lakes and 4 coastal sites. The corresponding averaged normalized rms-error for significant wave height
in the 2D cases varied between 8% and 27%. In comparison, using the default setting with a constant scaling,
as used in most presently operating 2D spectral wave models, gave equivalent errors between 15%
and 38%.
and for design andmanagement in the coastal zone. Although numerous studies have demonstrated the applicability
of a constant scaling of the Battjes and Janssen (1978) dissipationmodel for depth-induced breaking, recent
studies have shown its inability to sufficiently reproduce wave dissipation over complex field cases. In the
present study, we contrast the application of such a constant scaling to two alternative wave breaking parameterizationswith
a variable scaling based on either thewave nonlinearity (the φparameterization) oronbothbottomslope
and normalizedwavelength supplementedwithwave directionality (the β−kd parameterization).We
consider three field data sets characteristic of a simple beach-bar profile, a bay partially protected by a shoal and a
complex intertidal region.We demonstrate that in these cases the β−kd parameterization provides a better alternative
to the use of a constant scaling or the φ parameterization. To illustrate the operational consequences,
we up-scale the conditions over the case of the intertidal region to correspond to design conditions for the
Dutch coast (storm conditions with a 4000 year return period). Under these extreme conditions, for locally generated
waves both the β−kd and φ parameterizations predict qualitatively similar increased significant wave
heights but the β−kd parameterization increased the waves twice as much as the φ parameterization. Under
other conditions, when non-locally generated waves (swell) dissipates over a gently sloping bottom, the β−kd
parameterization predicts lower significant wave heights compared to either the constant scaling or φ
parameterization.
spectral wave models. Significant wave heights over horizontal bathymetries are typically under-predicted
in locally generated wave conditions and over-predicted in non-locally generated conditions. A
joint scaling dependent on both local bottom slope and normalized wave number is presented and is
shown to resolve these issues. Compared to the 12 wave breaking parameterizations considered in this
study, this joint scaling demonstrates significant improvements, up to 50% error reduction, over 1D horizontal
bathymetries for both locally and non-locally generated waves. In order to account for the inherent
differences between uni-directional (1D) and directionally spread (2D) wave conditions, an extension
of the wave breaking dissipation models is presented. By including the effects of wave directionality, rmserrors
for the significant wave height are reduced for the best performing parameterizations in conditions
with strong directional spreading. With this extension, our joint scaling improves modeling skill for significant
wave heights over a verification data set of 11 different 1D laboratory bathymetries, 3 shallow
lakes and 4 coastal sites. The corresponding averaged normalized rms-error for significant wave height
in the 2D cases varied between 8% and 27%. In comparison, using the default setting with a constant scaling,
as used in most presently operating 2D spectral wave models, gave equivalent errors between 15%
and 38%.