Imports¶

In [ ]:

%load_ext autoreload
%autoreload 2

from __future__ import annotations

In [ ]:

# debug
import sys

print(sys.executable)
print(sys.version_info)

In [ ]:

# main imports
import holoviews as hv
import numpy as np
import pandas as pd
import xarray as xr

hv.extension("bokeh")

Open a dataset¶

Thalassa supports the output of several solvers (e.g. Schism, ADCIRC). In order to do so, it converts the output of these files to a common "schema". This process is called "normalization".

"normalization" == Renaming of dimensions and variables

The most convenient way to apply this normalization process is to use the api.open_dataset() function which is a wrapper around xr.open_dataset().

In [ ]:

from thalassa import api

api.open_dataset?

In [ ]:

filename = "../tests/data/fort.63.nc"
ds = api.open_dataset(filename)
ds

Main API¶

thalassa supports several types of graphs, including:

A visualization of the variables at various timestamps and/or layers.
A visualization of the mesh
Extraction of timeseries from specific points

The following cell shows the basic usage:

In [ ]:

#variable, layer, timestamp = "salt", 40, 
#variable, layer, timestamp = "depth", None, None
variable, layer, timestamp = "zeta", None, ds.time.values[4]

# The trimesh is the most basic object. This is what you need to create all the others graphs
# It is on this object that you specify the timestamp and/or the layer.
trimesh = api.create_trimesh(ds.sel(time=timestamp), variable=variable)

# The wireframe is the representation of the mesh
wireframe = api.get_wireframe(trimesh)

# The tiles is using the tiling service from Open Street maps
tiles =  api.get_tiles() 

# The raster object is the basic Map that visualizes the variable. 
# You can specify things like the colorbar limits and/or the extents
#raster = api.get_raster(trimesh, clim_min=0, clim_max=15)
raster = api.get_raster(trimesh)

# The pointer/tap timeseries extract the timeseries of a specific node from the xr.Dataset and visualize it.
pointer_dmap = api.get_pointer_timeseries(ds=ds, variable=variable, source_raster=raster)
tap_dmap = api.get_tap_timeseries(ds=ds, variable=variable, source_raster=raster)

Let's visualize!¶

After you render the layout, move the mouse over the map and click on it. This will fill in the pointer_dmap and the tap_dmap.

In [ ]:

raster_layout = tiles * raster.opts(width=600, cmap="viridis") 
raster_layout

In [ ]:

import warnings

with warnings.catch_warnings():
    warnings.simplefilter("ignore")
    
    tap_dmap

In [ ]:

mesh_layout = tiles * wireframe.opts(width=600)
mesh_layout