Water Flow in the Inlets of the Grado-Marano Lagoon System, Northern Adriatic Sea

I. Mancero Mosquera 1, F. Arena 1 , V. Kovacevic 1, R. Villalta 2, M. Lipizer 2, A. Triches 2, G. Fontolan 3, S. Pillon 3 and A. Bezzi 3.

1. Istituto Nazionale di Oceanografia e di Geofisica Sperimentale OGS, Trieste, Italy

2. Autorità di Bacino Regionale del Friuli VeneziaGiuliaISMARCNR, Venice, Italy.

3. Dipartimento di Geoscienze, University of Trieste, Italy. GradoMarano Lagoon System

Fossalon Lignano di Grado Grado Flow sign convention:

Inflowing (-)

Fossalon Lignano di Grado Grado

Out flowing (+) Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Homogeneity along the water column:

7 July – 8 August 2010 Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Homogeneity along the water column:

EOF Analysis: • Applied to 34 cells out of 45 (8.5 m). • The Barotropic Mode accounts for about 99% of energy in Grado and 98% in Lignano. • Rest of energy is distributed in the remaining 33 modes Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Horizontal variability of velocity distributions: Near surface Mid waters Near bottom

Grado 8.9 m 5.9 m 3.15 m -

Lignano 8.3 m 5.3 m 2.55 m Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Horizontal variability of velocity distributions: PCA Analysis: • Applied to 34 cells out of 45 (8.5 m). • The Principal Component (PC1) accounts for more than 99% of the variance in both inlets. • Inclination of the Major Axis (PC1) is 80° (from East, positive anti clockwise) in Grado, 59° in Lignano. • Therefore it can be said flow is polarized. Current Velocities in the Grado-Marano Lagoon system: State of the knowledge • Further analyses were thus done over verticallyaveraged and PCdecomposed time series, taking the PC1 component as fairly representative of the variability in the inlets. Mean St.Dev. Min. (in flowing) Max. (out flowing) mm/s mm/s m/s date m/s date Grado 14.43 574.76 1.46 16 Feb.11 @18h 1.36 4 Dec.10 @11h Lignano 17.38 493.38 1.38 11 Aug.10 @08h 1.42 18 Mar.11 @11h Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Spectral Analysis:

SD SD D D

D = Diurnal SD = Semi-diurnal Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Spectral Analysis: Total signal (blue) + Tidal signal (red)

TD LF

TD = Third diurnal LF = Low frequencies Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Spectral Analysis: Total signal (blue) + Nontidal signal (black)

Seiches Seiches Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Grado signals: Total (up), Tidal (mid), Non-tidal (lo) with their contributions to the total variance.

100 %

90.2%

9.8% Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Lignano signals: Total (up), Tidal (mid), Non-tidal (lo) with their contributions to the total variance.

100 %

90.17%

9.83% Conclusions (1st part)

• Grado and Lignano inlets have been monitored with bottom- mounted ADCPs • Period of analysis goes from July 2010 to August/ September 2011 • Water flow is prevalently barotropic in both inlets • There is a strong polarization along the channel axes (as expected). • Average values show a out-flowing regime (long term?) • Astronomical forcing is the strongest one (about 90%) • Seiche signals are found in the spectra Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Let’s focus on the nontidal part! Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Seiche signals appear in the Non-tidal part

SD - Total Adriatic seiches D - Tidal TD Higher Higher harmonic Harmonics Adriatic seiches LF Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Higher harmonic Seiches has been previously reported

- Total Adriatic seiches - Non tidal Higher harmonic Adriatic seiches

Seiche 10-11h 7h 5h Seiche 21-22h

Vilibic et al., 1998 (Middle Adriatic islands) Mancero-Mosquera et al., 2009 (Venice inlets) Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

Coherent or in-phase regime of the non-tidal parts! Wavelet Transform Analysis Tools

In a similar way to windowed Fourier transform, the Wavelet transform of a function is: +∞ Wf (τ )s, = ,f ψ = )t(f ψ * dt τ s, ∫−∞ τ s,

Where * denotes complex conjugation. ψ τ s, is a scaled and translated version of a so called ‘mother wavelet’ψ )t( : 1  t − τ  ψτ s, = ψ  s  s  ψ )t( is well localized in time and frequency and must to satisfy some properties like having zero mean. The set of scaled and translated versions of ψ , { ψ τ s, }, can be used to represent a function f(t) in a bi-dimensional domain called translation-scale plane (τ )s, Current Velocities in the Grado-Marano Lagoon system: State of the knowledge The Wavelet Transform enables a decomposition equivalent to a band-pass filter bank, the Fourier spectrum is partitioned in bands (“scales”) with power-of-two (“dyadic”) frequency limits, here indicated with dashed blue lines. Green arrows show scales enclosing Seiche variability. 2-4 h 2-4 4-8 h 4-8 8-16 h 8-16 16-32 h 16-32 Each scale has a corresponding time series GRADO Each scale has a corresponding time series LIGNANO A Wavelet spectrum is thus coarser than a Fourier spectrum

Variance of Non-tidal signals on a scale-by-scale basis

Seiche scales: 84.6% Grado 77.9% Lignano of non tidal variance We can fit the scales containing Seiche variability to the Non- tidal series, via least-squares, and obtain a Non-Seiche series, by using a linear superposition premise: De-Seiched = X -a× W -b × W -c × W nontidal 4-8h 8-16h 16-32h seiche signal via Least Squares fit

Grado Lignano a = 0.5129 0.5236 b = 0.2146 0.2055 c = 0.2095 0.2146 R2 = 0.9081 0.8472 Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

GRADO Non tidal: 9.8%

Seiches: 8.2%

De-seiched: 1.4%

2010 2011 Current Velocities in the Grado-Marano Lagoon system: State of the knowledge

LIGNANO Non tidal: 9.83%

Seiches: 7.66%

De-seiched: 2.17%

2010 2011 Now let’s focus on the Residual (Non-tidal and no seiches) part of the signal!

LIGNANO Non tidal: 9.83%

Seiches: 7.66%

De-seiched: 2.17%

2010 2011 Conclusions (2nd part)

• Non tidal flows in Grado and Lignano are mostly in-phase when compared. • Seiche higher harmonics have been detected, which are reported in the literature for other shallow environments in the Adriatic Sea. • Wavelet Transform allows for a separation of frequency bands containing the Adriatic Seiche variability. • Adriatic Seiche signal is an important contributor of the Non-tidal variability: 84.6% in Grado and 77.9% in Lignano. • Since Non-tidal variability is 9.8% of the Total , therefore Seiche signals are about 8.2% of the Total variability in Grado and 7.6% of Total in Lignano. • Thus, Residuals (non-tidal and no seiches) are of the order of 1-2% of Total variability. A connection is attempted between the Residual series (no tides and no Seiches) with local Wind data Residual portion Winds in the area of the GradoMarano Lagoon System

Fossalon di Grado

Lignano Grado Winds in the area of the GradoMarano Lagoon System

PC2 (+)

PC1 (+)

W E

(-)

S (-) Winds in the area of the GradoMarano Lagoon PC2 System N

PC1 Angle Exp.Var mean std W E (°) (%) (m/s) (m/s) PC1 21.70 77.44 1.48 3.31 PC2 111.70 22.56 0.18 1.79

S Winds and Residual Current velocities in the Grado-Marano Lagoon system A classification of wind is made in order to understand the currents under different wind velocity-classes, for PC1 and PC2

Wind: PC1 classification (illustration only) PC2

“CALM” N in PC1

PC1

W E

S A classification of wind is made in order to understand the currents under different wind velocity-classes, for PC1 and PC2 Wind: PC2 classification (illustration only) PC2 “CALM” in PC2 N

PC1

W E

S • Response to PC2 is evaluated along the “Calm” of PC1. Response to PC1 is evaluated along the “Calm” of PC2. • For each wind class, simple average and standard deviation of currents is computed. Wind classification (illustration of the method) PC2

“CALM” N

PC1

W E

S This current-velocity statistics are then plotted against the Wind velocity class marks:

Response to Wind PC1 (including Bora) with the proposed model and estimated parameters. PC2 is in “Calm” here.

e(− aW PC1 ) YGrad = C - b ×

e(− aWPC1 ) YLign = b ×

a b CR 2 Grado 0.1349 11.5418 30 0.9805 Bora Calm Lignano 0.1256 10.7598− 0.7905 This current-velocity statistics are then plotted against the Wind velocity class marks:

Response to Wind PC2 (including Scirocco) with the proposed model and estimated parameters. PC1 is in “Calm” here. (aW + b W ) Y = K ×e PC2 PC2 Calm scirocco Grad e(aWPC2 ) YLign = K ×

a b K R 2 Grado− 0.01165 0.18991 12.6316 0.9644 Lignano− 0.1927 − 11.5070 0.7905 Conclusions (3rd part)

• Residual currents amount to 1.4 % of Total variability in Grado, while 2.17 % of Total variability in Lignano. • However, a connection with the prevalent winds (Bora, Scirocco) blowing in the area is found to be statistically significant. • Winds are decomposed in Principal components, so that PC1 (where Bora is expected to appear) contributes 77.4% of total wind variance; while PC2 (Scirocco) with 22.6% • Classification of wind velocity is made in order to provide Control Groups (“Calm” intervals) to better evaluate the effect of PC1 and PC2 on the currents. • This response is modeled statistically with exponential funcions, which provide excellent fits with R2 over 90% for Grado and almost 80% in Lignano.