Interactive Visualuzation using hvPlot

---

Overview

ERA-5 Dataset is available from NCAR RDA in netcdf format. A subset of this dataset is processed into Zarr format and available from NCAR RDA endpoints. To learn how you can create Zarr files from NCAR RDA netcdf files, please see this notebook.

By the end of this notebook, you should be able to:

• Understand the importance for interactive plots and the challenges associated with them

• Use hvPlot to generate basic interactive plots with Xarray

Prerequisites

Concepts	Importance	Notes
Intro to Xarray	Necessary

• Time to learn: 30 minutes

Imports

import holoviews as hv
import xarray as xr
from holoviews import opts

hv.extension("bokeh")

Data

As we mentioned above a subset of NCAR RDA data is available in Zarr format.

rda_url = "https://data.rda.ucar.edu/"
annual_means = rda_url + "pythia_era5_24/annual_means/"
xrds = xr.open_dataset(annual_means + "temp_2m_annual_1940_2023.zarr", engine="zarr")
xrds = xrds.isel(time=slice(0, 5))
xrds.load()

<xarray.Dataset> Size: 21MB
Dimensions:    (time: 5, latitude: 721, longitude: 1440)
Coordinates:
  * latitude   (latitude) float64 6kB 90.0 89.75 89.5 ... -89.5 -89.75 -90.0
  * longitude  (longitude) float64 12kB 0.0 0.25 0.5 0.75 ... 359.2 359.5 359.8
  * time       (time) datetime64[ns] 40B 1940-12-31 1941-12-31 ... 1944-12-31
Data variables:
    VAR_2T     (time, latitude, longitude) float32 21MB 258.2 258.2 ... 226.1

xarray.Dataset

• Dimensions:
  • time: 5
  • latitude: 721
  • longitude: 1440
• Coordinates: (3)
  • latitude
    (latitude)
    float64
    90.0 89.75 89.5 ... -89.75 -90.0

    long_name :

    latitude

    short_name :

    lat

    units :

    degrees_north

    array([ 90.  ,  89.75,  89.5 , ..., -89.5 , -89.75, -90.  ])

  • longitude
    (longitude)
    float64
    0.0 0.25 0.5 ... 359.2 359.5 359.8

    long_name :

    longitude

    short_name :

    lon

    units :

    degrees_east

    array([0.0000e+00, 2.5000e-01, 5.0000e-01, ..., 3.5925e+02, 3.5950e+02,
           3.5975e+02])

  • time
    (time)
    datetime64[ns]
    1940-12-31 ... 1944-12-31

    array(['1940-12-31T00:00:00.000000000', '1941-12-31T00:00:00.000000000',
           '1942-12-31T00:00:00.000000000', '1943-12-31T00:00:00.000000000',
           '1944-12-31T00:00:00.000000000'], dtype='datetime64[ns]')

• Data variables: (1)
  • VAR_2T
    (time, latitude, longitude)
    float32
    258.2 258.2 258.2 ... 226.1 226.1

    QuantizeGranularBitGroomNumberOfSignificantDigits :

    7

    ecmwf_local_table :

    128

    ecmwf_parameter :

    167

    grid_specification :

    0.25 degree x 0.25 degree from 90N to 90S and 0E to 359.75E (721 x 1440 Latitude/Longitude)

    long_name :

    2 metre temperature

    maximum_value :

    320.42938232421875

    minimum_value :

    216.13075256347656

    original_format :

    WMO GRIB 1 with ECMWF local table

    rda_dataset :

    ds633.0

    rda_dataset_doi :

    DOI: 10.5065/BH6N-5N20

    rda_dataset_group :

    ERA5 atmospheric surface analysis [netCDF4]

    rda_dataset_url :

    https:/rda.ucar.edu/datasets/ds633.0/

    short_name :

    2t

    units :

    K

    array([[[258.23004, 258.23004, 258.23004, ..., 258.23004, 258.23004,
             258.23004],
            [258.1854 , 258.18552, 258.1857 , ..., 258.18515, 258.1852 ,
             258.1853 ],
            [258.17493, 258.17517, 258.17554, ..., 258.1743 , 258.1744 ,
             258.17462],
            ...,
            [226.41481, 226.41528, 226.4167 , ..., 226.41333, 226.41449,
             226.41505],
            [226.33344, 226.3346 , 226.3358 , ..., 226.33185, 226.33238,
             226.33289],
            [225.71968, 225.71968, 225.71968, ..., 225.71968, 225.71968,
             225.71968]],
    
           [[257.09677, 257.09677, 257.09677, ..., 257.09677, 257.09677,
             257.09677],
            [257.13657, 257.1367 , 257.1368 , ..., 257.1363 , 257.13638,
             257.13647],
            [257.1755 , 257.1757 , 257.1759 , ..., 257.17484, 257.17502,
             257.17523],
    ...
            [227.85681, 227.85716, 227.85823, ..., 227.85608, 227.85698,
             227.8573 ],
            [227.76569, 227.76662, 227.76758, ..., 227.76445, 227.76485,
             227.76527],
            [227.14423, 227.14423, 227.14423, ..., 227.14423, 227.14423,
             227.14423]],
    
           [[257.65396, 257.65396, 257.65396, ..., 257.65396, 257.65396,
             257.65396],
            [257.73788, 257.73813, 257.73837, ..., 257.73724, 257.73746,
             257.73767],
            [257.81528, 257.81577, 257.81625, ..., 257.81384, 257.81424,
             257.81473],
            ...,
            [227.14717, 227.14827, 227.15022, ..., 227.14409, 227.14583,
             227.14693],
            [226.99713, 226.99864, 227.00008, ..., 226.9947 , 226.99556,
             226.99634],
            [226.14192, 226.14192, 226.14192, ..., 226.14192, 226.14192,
             226.14192]]], dtype=float32)

• Indexes: (3)
  • latitude
    PandasIndex

    PandasIndex(Index([  90.0,  89.75,   89.5,  89.25,   89.0,  88.75,   88.5,  88.25,   88.0,
            87.75,
           ...
           -87.75,  -88.0, -88.25,  -88.5, -88.75,  -89.0, -89.25,  -89.5, -89.75,
            -90.0],
          dtype='float64', name='latitude', length=721))

  • longitude
    PandasIndex

    PandasIndex(Index([   0.0,   0.25,    0.5,   0.75,    1.0,   1.25,    1.5,   1.75,    2.0,
             2.25,
           ...
            357.5, 357.75,  358.0, 358.25,  358.5, 358.75,  359.0, 359.25,  359.5,
           359.75],
          dtype='float64', name='longitude', length=1440))

  • time
    PandasIndex

    PandasIndex(DatetimeIndex(['1940-12-31', '1941-12-31', '1942-12-31', '1943-12-31',
                   '1944-12-31'],
                  dtype='datetime64[ns]', name='time', freq=None))

• Attributes: (0)

Considerations for Interactive Plots

Add some markdown text on some of the following ideas:

• What are some reasons we want to make data visualuzation interactive?

Baisc Interactivity using hvPlot

The hvPlot package is a familiar and high level API for data exploration and visualuzation.

One of the most powerfull features of hvPlot is that it provides an alternative plotting API that directly attaches to existing Python objects through the .hvplot() attribute. For the case of Xarray, importing hvplot.xarray adds a brand new set of plotting routines accessible either through xr.DataArray.hvplot() or xr.Dataset.hvplot()

import hvplot.xarray

Before using hvPlot, let’s take a look at the default Xarray plotting methods.

xrds["VAR_2T"].plot()

(array([ 117402.,  170562.,   92662.,  147326.,  602810.,  510824.,
         719488.,  723657., 1055503., 1050966.]),
 array([216.61584473, 225.55987549, 234.50390625, 243.44793701,
        252.39198303, 261.33599854, 270.2800293 , 279.22409058,
        288.16812134, 297.1121521 , 306.05618286]),
 <BarContainer object of 10 artists>)

We can replace the .plot() function call with .hvplot(). By default, hvPlot uses the Bokeh backend, which has naitive interactive tools, such as :

• Panning

• Box Select

• Scroll Zoom

• Saving

• Resetting

xrds["VAR_2T"].hvplot()

If we wanted to plot …

xrds["VAR_2T"].isel(time=0).plot()

<matplotlib.collections.QuadMesh at 0x7f3ee49bf450>

Switching

xrds["VAR_2T"].isel(time=0).hvplot()

Time Widget

Climate data typically comes with multiple timesteps. We can create a basic widget that allows us to seek through time by setting the groupby='time' parameter in our .hvplot() call.

xrds["VAR_2T"].hvplot(groupby="time", widget_location="bottom")

You may notice that our colorbar is dynamically changing as we change our time steps. We can fix the colorbar by setting a clim value, which is a tuple of the minimum and maximum desired colorbar range.

One suggestion is to use the minimum and maximum of the data variable you are visualuzing across time.

clim = (xrds["VAR_2T"].values.min(), xrds["VAR_2T"].values.max())

xrds["VAR_2T"].hvplot(clim=clim, groupby="time", widget_location="bottom")

You may have noticed that there is a slight lag when switching time steps. This is due to hvPlot plotting the full resolution of our dataset. We can instead rasterize the output by setting rasterize=True, which will significantly improve the perfromance of our interactive plot.

xrds["VAR_2T"].hvplot(
    rasterize=True, clim=clim, groupby="time", widget_location="bottom"
)

Animation Widget

Another usefull interactive feature is animations. Instead of manually scrolling through time, we can set up a widget that lets us animate our data across time. This can be achieved by adding a Scrubber widget to our plot by setting widget_type="scrubber"

xrds["VAR_2T"].hvplot(
    rasterize=True,
    groupby="time",
    widget_type="scrubber",
    widget_location="bottom",
)