depth estimation using video cameras in the south-west...camera system (short) 3. depth estimation...

Depth estimation using video cameras in the South-West

Erwin Bergsma member of the Coastal Processes Research Group, Plymouth University

Thursday 15 October 2015, Annual partner’s meeting, Newton Abbot

Contents

Personal background 1. Study site (short)

2. Camera system (short)

3. Depth estimation using cameras (long)

• MSc. at Delft University of Technology – Environmental Fluid Mechanics

• Turbulence in shallow jet flows – Coastal Engineering:

• Inlet breaching (2010 Xyntia storm) modelling; Pertuis de Maumusson, France

• PhD at Plymouth University – Coastal Oceanography

• Bathymetry estimation using video cameras

• Freelance Coastal Engineer, Hydraulic and morphodynamic modeller and Programming Specialist:

– Modelling: [Delft3D, Xbeach, SWAN, SWASH and Unibest package] – Programming: [MatLAB, Python] – Africa : Gabon, Senegal. Asia : Fujairah (UAE), Abu Dhabi, Bahrain, Iraq.

Personal Background

Study Site

Study Site Porthtowan, Cornwall

Geographical orientation

Camera system

Camera system

Camera system

Camera system

Why do we want to estimate depths?

Residual coastal strength: first defence against flooding,

Why do we want to estimate depths?

Residual coastal strength: first defence against flooding, depth estimates between storms to access residual strength

Why do we want to estimate depths?

Residual coastal strength: first defence against flooding, depth estimates between storms to access residual strength

Calm wave conditions are not required to get an

accurate depth estimation with the video cameras

Why do we want to estimate depths?

Beach user /Swimmer’s Safety:

Why do we want to estimate depths?

Beach user /Swimmer’s Safety:

Where those patterns on the left before?

Why do we want to estimate depths?

Beach user /Swimmer’s Safety:

Where those patterns on the left before?

How does it works?

nearshore

offshore

How does it works?

Relation between water waves to get a depth:

𝝈2 = 𝑔𝑔 tanh 𝒌𝒉 ⇔ c = 𝑔𝑘

tanh 𝑔𝒅

c = wave speed. Lower speed = shallower, higher speed = deeper

Summary

1. Relation between shallow water waves and depth

2. “Tracking” waves with video cameras to find the wave’s speed.

Results

Results

SURVEY

ESTIMATE

Results

SURVEY

ESTIMATE

Results

Results Hourly depth estimates:

Results

Depth estimate – mean cross shore

profile over 1,5 years

Results

Depth estimate – mean cross shore

profile over 1,5 years

Results

Depth estimate – mean cross shore

profile over 1,5 years

Results Pre stormy winter 2014 Post stormy winter 2014

Results Pre stormy winter 2014 Post stormy winter 2014

Results Pre stormy winter 2014 Post stormy winter 2014

Limitations? • Depth estimation in the shallowest parts less accurate:

linear dispersion relation is valid in lesser extend • The more off-shore the harder the waves are

distinguishable by the cameras

Further work and challenges

• Higher resolution cameras. Leads to more possibilities further offshore

• Mobile cBathy stations, short sampling periods • cBathy applied on X-band radar • cBathy from a UAV / drone • cBathy with satellite imagery • Large scale cBathy deployment over the world

Thank you

Coastal Processes Research Group: www.coastalprocesses.org daniel.conley@plymouth.ac.uk Erwin W. J. Bergsma: erwin.bergsma@plymouth.ac.uk or erwin@bergsma.in

Extra slides about cBathy

nearshore

offshore

How does it works?

Relation between water waves to get a depth:

𝝈2 = 𝑔𝑔 tanh 𝒌𝒉 ⇔ c = 𝑔𝑘

tanh 𝑔𝒅

We want to find wave frequency(𝝈) and wave number 𝒌

How does it works?

• Wave frequency σ = 1𝑤𝑤𝑤𝑤 𝑝𝑤𝑝𝑝𝑝𝑝

has to do with time (t) and period it takes to complete one wave

• Wave number k = 1𝑤𝑤𝑤𝑤 𝑙𝑤𝑙𝑔𝑙𝑙

has to do with space (x,y) and length of one complete wave

one complete wave

How does it works?

So how do we get the wave frequency and wave number?

How does it works? 2ND relation: Similarity of pixel intensities to water level elevation:

𝜼 𝒙,𝒚, 𝒕 = �𝐴𝑙 sin 𝝈𝒏𝑡 −𝒌𝒙,𝒏 𝑥 −𝒌𝒚,𝒏 𝑦 + 𝜑𝑁

𝑙=1

1𝑤𝑤𝑤𝑤 𝑝𝑤𝑝𝑝𝑝𝑝

1

𝑤𝑤𝑤𝑤 𝑙𝑤𝑙𝑔𝑡𝑙

How does it works?

To find the waveWe only have to estimate the distance between (for k) and the speed (𝝈𝒏) of

the waves in time with video cameras;