BinWaves & SWAN#

import warnings
warnings.filterwarnings('ignore')

# common
import os
import os.path as op
import sys
import pickle as pk
import shutil

# pip
import numpy as np
import pandas as pd
import xarray as xr

# DEV: bluemath
sys.path.insert(0, op.join(op.abspath(''), '..', '..', '..', '..'))

# bluemath modules 
from bluemath.common import store_xarray_max_compression

from bluemath.binwaves.plotting.grid import Plot_grid_cells, Plot_grid_cases
from bluemath.binwaves.plotting.swan_case import Plot_swan_case

from bluemath.wswan.wrap import SwanProject, SwanMesh, SwanWrap_STAT
from bluemath.wswan.plots.stationary import scatter_maps

Methodology#

BinWaves

A new methodology for efficiently propagating multivariate wave systems to nearshore locations is propossed.
The methodology, which is named BinWaves, consists on disagregating the full wave spectra into its different energy “bins” which are characterized by monochromatic wave trains represented by a frequency and a direction.
Based on the linearity of the waves in deep waters, every directional spectra can be later on aggregated by supperposing all its different constituents at each location and time.

Methodology

These are the steps followed in the BinWaves methodology

Divide the full energy spectra into its individual energy “bins”. The spectra, characterized by 29 frequencies and 24 directions, results in 696 monochromatic wave trains.
Numerically run the 696 monochromatic cases with unitary significant wave height.
Post process the 696 simulations to obtain the propagation coefficients (Kp) of each energy bin at every location.

The propagation coefficients nearshore, as exemplified in the sketch below at each point of the grid, can be characterized by more than 1 different directional component due to the interaction with the bathymetry along the propagation.

SWAN

Individual “bins” have been numerically simulated using SWAN (Simulating WAves Nearshore), which is a third-generation open source wave model developed at Delft University of Technology.

TITLE

Reconstruction methodology

Once the outputs of the BinWaves methodology have been obtained at each point of the grid, for every generic wave spectra, the propagated spectra can be reconstructed by supperposing all of its components following the equation below.

TITLE

Database and site parameters#

# database
p_data = '/media/administrador/HD2/SamoaTonga/data'
site = 'Samoa'
p_site = op.join(p_data, site)

# site bathymetry
p_bathy = op.join(p_site, 'bathymetry')

# deliverable folder
p_deliv = op.join(p_site, 'd05_swell_inundation_forecast')

p_spec = op.join(p_deliv, 'spec')
p_swan = op.join(p_deliv, 'swan')

# SWAN grid depth and parameters
p_grid_depth = op.join(p_bathy, 'grid300_depth.pkl')
p_grid_param = op.join(p_bathy, 'grid300_params.pkl')

# launch SWAN
launch_swan = False

# superpoint
deg_sup = 15
p_superpoint = op.join(p_spec, 'super_point_superposition_{0}_.nc'.format(deg_sup))


# output files
name = 'binwaves'
p_swan_subset = op.join(p_swan, name, 'subset.nc')
p_swan_output = op.join(p_swan, name, 'out_main_binwaves.nc')
p_swan_input_spec = op.join(p_swan, name, 'input_spec.nc')

# Load superpoint
sp = xr.open_dataset(p_superpoint)

dirs = sp.dir.values
freqs = sp.freq.values

A number of 696 cases need to be run, corresponding to all the combinations of the spectra disctretized in 24 directions and 29 frequencies.
Below is an example of one of the monochromatic cases, with a freq ~0.3 (T = 3.3s) and Dir = 75

Plot_grid_cells(sp, pos=[25, 12]);

Obtain the parameters of the monochromatic cases

# each SWAN case is defined by 5 wave parameters: hs, tp, dir, spr, gamma
vns = ['hs','per','dir','spr','gamma']

gamma = 50  # waves gamma 
spr = 3     # waves directional spread

# initialize case variables
cases = np.full([len(sp.freq) * len(sp.dir), len(vns)], np.nan)
cases_id = np.full([len(sp.freq), len(sp.dir)], np.nan)

c = 0  # case counter

# generate one case for each superpoint direction and frequency values
for ix_d, d in enumerate(sp.dir.values):
    for ix_f, f in enumerate(sp.freq.values):
        
        # hs switch
        if 1 / f > 5:
            hs = 1
        else:
            hs = 0.1
       
        # set case parameters
        cases[c, :] = [hs, np.round(1/f,4), d, spr, gamma] 
        
        # set case ID
        cases_id[ix_f, ix_d] = c
        
        c+=1

# cases DataFrame
df = pd.DataFrame(cases, columns = vns)

# plot cases ID
Plot_grid_cases(sp, cases_id.astype('int'));

Plotting#

Individual cases#

Note

Once the cases have been run and outputs have been extracted, here we plot some results.
First, some individual cases have been plot. From left to right the different variables are: Hs (which corresponds to the propagation coefficient Kp, since we are running unitary wave heights), Tp, Dir and Dspr for the bulk parameters of the simulation, although partitions will be used to determine the propagation coefficients for each case.

case = 0 
lon_s = 184
lat_s = -21

Plot_swan_case(out_sea_mm_sim, subset, case);

case = 500
lon_s = 184
lat_s = -21

Plot_swan_case(out_sea_mm_sim, subset, case);

case = 690
lon_s = 184
lat_s = -21

Plot_swan_case(out_sea_mm_sim, subset, case);

All cases#

Note

The grids bellow contain the same results but for all of the 696 cases ordered by Dir in rows and Tp in columns.

Hsig

nc = 29
nr = 24
n_cases = nc * nr
vars_ = ['Hsig']

scatter_maps(
    out_sea_mm_sim, 
    n_cases = n_cases,
    n_cols = nc, n_rows = nr,
    var_list = vars_, 
    figsize = [30, 0.6 * nr],
    var_limits = {'Hsig': (0, 2)},
);

TPsmoo

nc = 29
nr = 24
n_cases = nc * nr
vars_ = ['Tp']

scatter_maps(
    out_sea_mm_sim,
    n_cases = n_cases, 
    n_cols = nc, n_rows = nr, 
    var_list = vars_,
    figsize = [30, 0.6 * nr],
);

Dir

nc = 29
nr = 24
n_cases = nc * nr
vars_ = ['Dir']

scatter_maps(
    out_sea_mm_sim,
    n_cases = n_cases,
    n_cols = nc, n_rows = nr,
    var_list = vars_,
    figsize = [30, 0.6 * nr],
);

Dspr

nc = 29
nr = 24
n_cases = nc * nr
vars_ = ['Dspr']

scatter_maps(
    out_sea_mm_sim,
    n_cases = n_cases,
    n_cols = nc, n_rows = nr,
    var_list = vars_,
    figsize  = [30, 0.6 * nr],
);

BinWaves & SWAN

Contents

BinWaves & SWAN#

Methodology#

Database and site parameters#

SWAN Project#

Data extraction#

Plotting#

Individual cases#

All cases#