1
Shear Stress Data Mining (SSDM) method: national scale prediction of geomorphological processes Dr Barry Hankin, Dr Natasha Todd-Burley, Rebecca Ing, Matthew Hemsworth, JBA Consulting JBA Consulting was commissioned by the Environment Agency to carry out a research project looking at ways to predict hotspots of channel change. One of the short-listed methods, the Shear Stress Data Mining method, was developed by Barry Hankin from JBA Consulting. This approach makes use of existing large scale, high resolution national surface water flood mapping datasets. There is considerable depth and velocity information available with national coverage for a range of probability events nationally, and data-mining this information to understand shear stresses and likely sediment risk has not been undertaken at the national scale, to our knowledge. The approach relies on an efficient ArcGIS model builder code that computes local effective shear stress based on average velocity, depth and roughness. This is compared with critical shear stress for entrainment and erosion, using three assumptions on deposition, erosion and sediment grain size distribution. It results in a zonal classification of sediment erosion, transition and deposition based on a comparison of local effective shear stress and critical shear stress. The science underpinning this method is based on the physics of fluid flow and critical shear stress for entrainment and deposition. Predicting hotpots of channel change What is the Shear Stress Data Mining Method? What datasets does it use? How is it calculated? Effective shear stress can be derived from the following quadratic expression (Lane and Ferguson; 2005): Where is effective shear stress (N/m 2 ), is density of water (kg/m 3 ), g is acceleration due to gravity (m/s 2 ), n is the Manning’s coefficient (s/m 1/3 ), d is depth (m) and v is depth averaged velocity (m/s). This relation gives a very similar functional relationship to shear stresses derived on integrating flows assuming a logarithmic law of the wall. This quadratic equation can then be used and compared against critical shear stress and identify where erosion is more likely. Areas where the calculated shear stress is greater than the critical shear stress are classified as erosional. A model parameter ‘deposition threshold’ is then applied, which is a user-defined ratio defining the critical shear stress for deposition as a factor of critical shear stress. Areas where the calculated shear is less than this threshold are classified as depositional. If this parameter is set too high, deposition will occur wherever there is no erosion, and if too low there will hardly be any deposition shown. Examples below illustrate model outputs for the River Kent catchment, Cumbria. What do the outputs look like? JBA’s preliminary investigation using the River Kent Catchment has led to a series of recommendations for method improvements to be explored further. These include the following: Next Steps Filter out a buffer of riparian wooded areas because woodland promotes bank stability. It may be possible to remove some of the areas predicted to be high in erosion potential by 'cookie-cutting' out areas of woodland. Forest Inventory and OS 'Open Woods' layers could be used for this purpose. Apply a 'majority filtering’ to remove rapid spatial variations and isolated islands of one particular value. Develop a scenario library of SSDM outputs within a geodatabase such that users can pull in the most appropriate local conditions. The scenarios would consist of different combinations of Manning’s Roughness and D50 grain size, for the 3.33%, 1% and 0.1% Annual Exceedance Probability Flood Events. ArcGIS Majority Filter Lane, S.N. and Ferguson, R.I. (2005), Modelling reach-scale fluvial flows. In Computational Fluid Dynamics : Applications in Environmental Hydraulics. Paul. D. Bates, Stuart N. Lane, Robert I. Ferguson (eds). John Wiley and Sons Ltd. ISBN 13 978-0-470-84359-8 (HB) National DTM e.g. Integrated Height Model 2m resolution Digital Terrain Model (Environmen Agency, Licensed Data) Risk of Flooding from Surface Water complex model outputs (Environment Agency, Open Data) - continuous rasters of depths and velocities Landcover - Land Cover Map 2007 (Centre for Ecology and Hydrology Licensed Data) or CORINE Land Cover Map 2012 (Centre from Ecology and Hydrology, Open Data) with a landcover/roughness assumption Assumption on D50 grain size Where the D50 is the median diameter of the particle size distribution Depositional zone shown in Kentmere Reservoir Erosive reach modelled in the River Kent adjacent to Kentmere, with some localised depositional zones downstream To find out more, please contact: [email protected] C M Y CM MY CY CMY K poster 2 for print.pdf 1 30/08/2019 15:06

Dr Barry Hankin, Dr Natasha Todd-Burley, Rebecca Ing ... · Probability Flood Events. • ArcGIS Majority Filter Lane, S.N. and Ferguson, R.I. (2005), Modelling reach-scale fluvial

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Dr Barry Hankin, Dr Natasha Todd-Burley, Rebecca Ing ... · Probability Flood Events. • ArcGIS Majority Filter Lane, S.N. and Ferguson, R.I. (2005), Modelling reach-scale fluvial

Shear Stress Data Mining (SSDM) method: national scale prediction of geomorphological processes

Dr Barry Hankin, Dr Natasha Todd-Burley, Rebecca Ing, Matthew Hemsworth, JBA Consulting

JBA Consulting was commissioned by the Environment Agency to carry out a research project looking at ways to predict hotspots of channel change. One of the short-listed methods, the Shear Stress Data Mining method, was developed by Barry Hankin from JBA Consulting.

This approach makes use of existing large scale, high resolution national surface water flood mapping datasets. There is considerable depth and velocity information available with national coverage for a range of probability events nationally, and data-mining this information to understand shear stresses and likely sediment risk has not been undertaken at the national scale, to our knowledge.

The approach relies on an efficient ArcGIS model builder code that computes local effective shear stress based on average velocity, depth and roughness. This is compared with critical shear stress for entrainment and erosion, using three assumptions on deposition, erosion and sediment grain size distribution. It results in a zonal classification of sediment erosion, transition and deposition based on a comparison of local effective shear stress and critical shear stress. The science underpinning this method is based on the physics of fluid flow and critical shear stress for entrainment and deposition.

Predicting hotpots of channel change

What is the Shear Stress Data Mining Method?

What datasets does it use?

How is it calculated?

Effective shear stress can be derived from the following quadratic expression (Lane and Ferguson; 2005):

Where is effective shear stress (N/m2), is density of water (kg/m3), g is acceleration due to gravity (m/s2), n is the Manning’s coefficient(s/m1/3), d is depth (m) and v is depth averaged velocity (m/s). This relation gives a very similar functional relationship to shear stresses derived on integrating flows assuming a logarithmic law of the wall. This quadratic equation can then be used and compared against critical shear stress and identify where erosion is more likely.

Areas where the calculated shear stress is greater than the critical shear stress are classified as erosional. A model parameter ‘deposition threshold’ is then applied, which is a user-defined ratio defining the critical shear stress for deposition as a factor of critical shear stress. Areas where the calculated shear is less than this threshold are classified as depositional. If this parameter is set too high, deposition will occur wherever there is no erosion, and if too low there will hardly be any deposition shown. Examples below illustrate model outputs for the River Kent catchment, Cumbria.

What do the outputs look like?

JBA’s preliminary investigation using the River Kent Catchment has led to a series of recommendations for method improvements to be explored further. These include the following:

Next Steps

Filter out a buffer of riparian wooded areas because woodland promotes bank stability. It may be possible to remove some of the areas predicted to be high in erosion potential by 'cookie-cutting' out areas of woodland. Forest Inventory and OS 'Open Woods' layers could be used for this purpose.

Apply a 'majority filtering’ to remove rapid spatial variations and isolated islands of one particular value.

Develop a scenario library of SSDM outputs within a geodatabase such that users can pull in the most appropriate local conditions. The scenarios would consist of different combinations of Manning’s Roughness and D50 grain size, for the 3.33%, 1% and 0.1% Annual Exceedance Probability Flood Events.

ArcGIS Majority Filter

Lane, S.N. and Ferguson, R.I. (2005), Modelling reach-scale fluvial flows. In Computational Fluid Dynamics : Applications in Environmental Hydraulics. Paul. D. Bates,Stuart N. Lane, Robert I. Ferguson (eds). John Wiley and Sons Ltd. ISBN 13 978-0-470-84359-8 (HB)

National DTMe.g. Integrated Height Model 2m resolution Digital Terrain Model (Environmen Agency,

Licensed Data)

Risk of Flooding from Surface Water complex model outputs (Environment

Agency, Open Data) - continuous rasters of depths and velocities

Landcover - Land Cover Map 2007 (Centre for Ecology and Hydrology Licensed Data) or CORINE Land Cover Map 2012 (Centre from Ecology and Hydrology, Open Data) with a landcover/roughness assumption

Assumption on D50 grain sizeWhere the D50 is the median

diameter of the particle size distribution

Depositional zone shown in Kentmere Reservoir

Erosive reach modelled in the River Kent adjacent to

Kentmere, with some localised depositional zones downstream

To find out more, please contact: [email protected]

C

M

Y

CM

MY

CY

CMY

K

poster 2 for print.pdf 1 30/08/2019 15:06