DSD, Nov. 2015
Model-supported monitoring of coastal turbidity during
extension of the Port of Rotterdam
Meinte Blaas1, Marieke Eleveld2, Ghada el Serafy1, Hans van der Woerd2,
Sandra Gaytan1, Katherine Cronin1, William Borst3, Onno van Tongeren3
1) Deltares, 2) IVM VU University Amsterdam, 3) Port of Rotterdam,
DSD, Nov. 2015
Port of Rotterdam case (2007-2013)
Monitor the impact of Maasvlakte-2 land reclamation
Impact assessment extension of land reclamation by Port of Rotterdam (2008-2013).
Compliance to maximum increase in turbidity/SPM in large regions (N2000)
2003-2007: reference period for trend detection
2009-2012: 200.106 m3 sand extracted from sea floor (1.5 -2.5% silt)
DSD, Nov. 2015
Monitoring strategy: costs vs. insights
Traditional, ship-borne in situ monitoring
too limited in resolution & coverage
∆SPM
(%)
∆
t=7d
∆
t=15d
∆
t=30d
10 78 97 158
15 11 14 22
20 5 6 11
25 3 4 6
Noordwijk 20 km offshore site: D
min=8.9%, AC=7 d, 95% conf.
DSD, Nov. 2015
Monitoring strategy: costs vs. insights
Buoy monitoring resolves time
resolution issues, but still too limited in
spatial coverage
DSD, Nov. 2015
Is remote sensing the answer?
But…
•Only daytime & surface observations
• Seasonal biases (weather, sun angle)
• Intermittent gaps due to clouds & flagged pixels
Spatiotemporal resolution & coverage improve
trend detection.
DSD, Nov. 2015
MoS2: Model-Supported Monitoring of SPM
•VU IVM: MERIS Remote
Sensing SPM & Kd
•Port of Rotterdam: siltprofiler, bottom samples
•Deltares: model &
assimilation of SPM & Kd
DSD, Nov. 2015
SPM Modelling
Southern North Sea model
(Delft3D ‘ZUNO-DD’)
Hydrodynamic & sediment transport model:
• 3D SWE & adv. diff. eq for SPM
• 12 σ-layers, curvilinear horizontal grid,
• 3 HD domains, 2 WQ domains
• 2-layer sediment bed with buffering
• 3 silt size fractions
• lateral sources (cliffs, banks, rivers)
• resuspension of SPM due to currents &
surface waves
Van Kessel, Winterwerp, et al. 2011.
CSR 31(10), doi:10.1016/j.csr.2010.04.008
For example: Impact scenario
DSD, Nov. 2015
2010 : 92.479.569 m3
0
500000
1000000
1500000
2000000
2500000
3000000
3500000
4000000
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53
Weeks
m^3
Surface layer Bottom layer
annual mean increase [mg/l]
Baseline Scenario
DSD, Nov. 2015
EnKF to ‘interpolate’ RS data
DSD, Nov. 2015
Comparison against in situ profiles
DSD, Nov. 2015
Evaluation, e.g. Noordwijk site 10 km offshore
Surface salinity (magenta) & bottom stress (black)
Expand the forecast horizon
• Despite successful assimilation trend detection still cumbersome
with Kalman filtered model
• Revised BACI method: apply model as reference of autonomous
state of coastal system during entire construction phase
• Minimize RS vs. Model in T0
• Analyse RS vs. Model in T1
DSD, Nov. 2015
2003 2007 2008 2011
In Situ
RS
Model
Sand mining
}
Regional parameter optimisation
• Sensitivity analysis (Morris method)
10 parameters in 5 pairs for 19 regions:
• Settling Velocity, SettlingFluxRatios (3 silt fractions)
• TauShields & PickupFactor from sandy bed
• Resusp. rate, TauCrit of fluff layer (intermed. silt fraction)
• Parallel optimisation: minimize difference
between model output & remote sensing
(Simulated annealing)
• 2003-2008 (excl. 2007, used for evaluation)
• 84 parallel computational nodes
• Many days of iterations…
DSD, Nov. 2015
-300 -200 -100 0 100 200 300 400 km
200
300
400
500
600
700
800
900 km1
23
4
56
7
8
9
10
11
1213
14
15
1617
18
19
20
-300 -200 -100 0 100 200 300 400 km
200
300
400
500
600
700
800
900 km1
23
4
56
7
8
9
10
11
1213
14
15
1617
18
19
20
Validation
DSD, Nov. 2015
Trend analysis
DSD, Nov. 2015
Effect E E-σ E+σ
Depth-averaged effect term
[mg/l] Surface Depth Av. Bottom
E-σ -0.95 -1.10 -1.18
E 0.13 0.50 0.87
E+σ 1.38 2.37 3.29
Reference 9.30 13.19 16.61
Effect (trend) in Voordelta (2008:2011) – (2003:2007)
DSD, Nov. 2015
Summary
• Coastal management view on SPM in the North Sea
• Traditional monitoring data too low in resolution & coverage in time & space
> Too smooth perception of natural state of system
> Unfeasible requirements for environmental thresholds & trends and their detection
(significance)
• Additional sensors & platforms
• Moorings demonstrate temporal variations at tidal, wave (weather) & intrinsic
hydrodynamic time scales
• Remote Sensing demonstrated associated spatial structures
• Profiler & shipborne tracks show the local vertical gradients
• Model-supported monitoring:
• Direct interpretation of the various data sources is complicated (differences of
scale)
• Numerical model serves as common ground to refer these diverse data to
• Model used to create 3D extension -> insight in pathways, cause & effect
• Integrated approach gained trust from authorities