plot_rmse

plot_rmse(*args, ref=None, title_font_size=0.4, legend_font_size=0.35, y_max=None)

New in metview-python version 1.9.0

High level function with automatic styling to generate RMSE (root mean square error) curve plots from GRIB data.

Parameters

  • ref (Fieldset) – the reference GRIB data (typically it is an analysis)

  • title_font_size (number) – specifies the font size in cm for the plot title

  • legend_font_size (number) – specifies the font size in cm for the plot legend

  • y_max (number) – specifies the maximum value for the y axis in the RMSE plot. When it is not specified the value is automatically determined from the data.

plot_rmse() is a convenience function allowing to generate RMSE curve plots in a simple way using predefined settings.

The RMSE value for a given field f with respect to the corresponding field from ref can be written as follows (N is the number of gridpoints in the field):

\[rmse = \sqrt {\frac {1}{N} \sum_{i}^{N} (f_{i}-ref_{i})^{2}}\]

The averaging in the formula above is carried out with integrate().

Data

The Fieldset data to be compared to ref is defined by the positional arguments (*args). In *args each Fieldset must be defined on the same grid as ref and contain the same number of fields. It is also assumed that the dates/times/steps are properly matching that of ref.

Plot style

The curve style is automatically assigned to each input data item. The title and the legend is automatically generated using the label attribute in the Fieldset data in *args and ref. If a data item is part of a Dataset the label is automatically set, otherwise it can be defined like this:

an = mv.read("an.grib")
an.label = "an"

Limitations

plot_rmse() is a high-level function using pre-defined settings, therefore it comes with certain limitations:

  • the curve style cannot be customised

  • the title and the legend cannot be customised

  • the horizontal axis cannot be customised

  • on the y axis the maximum value can be set via y_max but everything else is automatically generated

Example with deterministic forecast

The following example shows how to create an RMSE curve plot with deterministic forecast data (geopotential on 500 hPa):

import metview as mv

# get data with 500 hPa geopotential
f = mv.gallery.load_dataset("z_rmse.grib", check_local=True)
an = f.select(type="an")
fc = f.select(type="fc")

# assign a label
an.label = "AN"
fc.label = "OPER"

# generate plot
mv.plot_rmse(fc, ref=an)
../../_images/plot_rmse_1.png

Example with ensemble forecast

The following example shows how to create an RMSE curve plot with ENS forecast data (geopotential on 500 hPa):

import metview as mv

# get data with 500 hPa geopotential
f = mv.gallery.load_dataset("ens_z_rmse.grib", check_local=True)
an = f.select(type="an")
fc = f.select(type="fc")
en = f.select(type=["cf","pf"])

# assign a label
an.label = "AN"
fc.label = "OPER"

# generate plot
mv.plot_rmse(en, fc, ref=an)
../../_images/plot_rmse_2.png