Comparing Leeway and ShipDrift model

The Leeway model contain many object categories from persons-in-water to various types of boats and ships. Here we compare the Leeway model with a fishing vessel to the ShipDrift model using the same ship dimensions.

from datetime import datetime, timedelta
from opendrift.models.leeway import Leeway
from opendrift.models.shipdrift import ShipDrift

We use a simple case with constant wind northwards, and no current. Wave height and period is calculated automatically from wind, and wave direction is the same as wind direction

environment = {
    'land_binary_mask': 0,
    'x_sea_water_velocity': 0,
    'y_sea_water_velocity': 0,
    'x_wind': 0,
    'y_wind': 18}

For the Leeway model we use “FISHING-VESSEL-3” with a length of 62m For the ShipDrift model we define a ship with the same dimensions, as well as a “large ship” scaled up by a factor of 8

ship = {'length': 62, 'beam': 8, 'height': 10, 'draft': 5}
large_ship = {k: v*8 for k, v in ship.items()}

cases = {
    'Leeway, FISHING-VESSEL-3': {
        'model': Leeway, 'kwargs': {'object_type': 52}},
    'Shipdrift, large ship': {
        'model': ShipDrift, 'kwargs': large_ship},
    'Shipdrift, small ship': {
        'model': ShipDrift, 'kwargs': ship}
    }

lon=3.5
lat=60
time = datetime.now()
duration = timedelta(hours=24)
simulations = []
for cname, case in cases.items():
    o = case['model'](loglevel=50)
    for var, value in environment.items():
        o.set_config('environment:constant:' + var, value)
    o.set_config('general:use_auto_landmask', False)
    o.seed_elements(lon=lon, lat=lat, time=time, number=1000,
                    **case['kwargs'])
    o.run(duration=duration)
    simulations.append(o)

We see that the overall drift speed of the ShipDrift model is comparable to the Leeway model. However, the Leeway model yields a much larger directional spread for ships orienters left and right respectively. For the scaled up ship, the drift speed is slightly larger, and the directional spread is slightly larger, but still much smaller than with the Leeway model.

simulations[0].plot(compare=simulations[1:], legend=list(cases), fast=False)
simulations[0].animation(compare=simulations[1:], legend=list(cases), fast=False)

/opt/conda/envs/opendrift/lib/python3.11/site-packages/cartopy/mpl/geoaxes.py:1692: UserWarning: No data for colormapping provided via 'c'. Parameters 'cmap' will be ignored
  result = super().scatter(*args, **kwargs)