Forecast plotting

From gfi

Both the forecast plotting script and the website creation script is based on python. The plotting script uses dynlib, the website creating script uses django the same library that also drives


Prepare dynlib

To get started, make sure you have followed the procedures to setup dynlib. Use the Guide to use the centrally installed dynlib. If you want to start developing own diagnostics, follow also the Quick start to developing with dynlib.

Obtain the example scripts

Example scripts as well as the actual forecast scripts currently used are available in a tar archive: /Data/gfi/users/csp001/prj/forecast_plotting.tar.bz2. In that archive there are three folders called examples, production and website.

  1. examples contains a thoroughly documented plot script and the necessary settings file that sets up the right paths and file name structure used for the forecasts.
  2. production contains the currently used plot script. The settings file also contains the colormaps developed by Franziska Menzel.
  3. website contains the script and the template file used to create web pages fitting to the plots.

The workflow

  1. Create a folder in which the plots should be created, and set conf.opath in the
  2. Create a folder which contains the contents of the folder website. The website will be created in this folder. The website expects an folder plots in that folder, so either create the folder in step 1 as a subfolder of this folder, or symlink the plot folder here.
  3. Follow the examples in to develop a new plot.
  4. Contain the plot in the structure
  5. When the forecasts are created, adapt the script to include your new variables/plots in the list starting around line 30.
  6. Create the website and open it in a browser!

Plotconf options

Many properties of the plots can be configured. The following list is a (hopefully complete) list of what is currently available for the dynlib plotting routines. Contact Clemens Spensberger in case you're missing an option. The implementation will be much easier and quicker if you check first such an option is available in the underlying plotting library matplotlib.

Name Line Fill Type Default Description
cb_orientation string 'vertical' Orientation of the color bar.
cmap mpl-colormap Color map for the contours.
cmap mpl-colormap cm.gist_ncar Color map.
coastcolor mpl-color black Color of the coastlines.
colors list of mpl-color black Presribe color of the contours, instead of using color map.
colors list of mpl-color Prescribe colors directly instead of using color map.
contour_labels bool False Label the contours within the plot.
contour_labels_fontsize int/string 12 Font size in point, or a string like 'smaller'.
contour_labels_inline bool True Remove the contour under the label?
contour_labels_inline_spacing int 2 Space in pixels around the label where the contour is removed as well.
contour_labels_format string/list of string '%1.1f' How to format the numbers, or list of strings used as labels.
disable_cb bool False Do not plot add a colorbar.
extend string 'both' Extend the color bar on the upper and lower boundaries? Possible values are: 'neither', 'upper', 'lower' and 'both'.
gridcolor mpl-color black Color lat/lon grid lines.
hook callable Function to apply to the data before plotting.
linestyles string Line styles for contours.
linewidths int/list Line widths for contours.
m Basemap worldmap Map projection.
maskcolor mpl-color light grey Color of the parts below orography.
mark tuple of list Tuple containing list of x-coordinates and y-coordinates to be marked on the map with circles.
oroalpha float 0.4 Transparency of the orography isolines, which 0 being entirely translucent.
orocolor mpl-color black Color of the orography isolines.
oroscale int/list Δ=1000 m Anything matplotlib accepts as a scale.
overlays list of overlay List of overlays (being fronts or countours) to plot on top.
plev string/int Vertical level of the plot. Used to auto-mask intersections with orography using the ERA-Interim average height of the level.
scale list/int/string 'auto' If 'auto', use the configurable autoscaling (see scale*-properties), otherwise anything matplotlib accepts.
scale_exceed_percentiles tuple of float (0.02, 0.98) Percentiles giving the minimal displayed data coverage of the colorbar. For the default values, the lowermost and uppermost inverval cuts of maximally 2% of the data values.
scale_intervals list of int [1,2,3,5,10] Which intervals are considered to be "round"? By default it is this list for any order of magnitude.
scale_intervals_periodic bool True Which intervals are considered to be "round"? If set to False the above list applies only for the given order of magnitude.
scale_target_steps int 7 If scale='auto', how many intervals are desired? The actual intervals might be a few more/less, to allow for a "round" interval value.
scale_symmetric_zero bool False If scale='auto', force the scale to be symmetric around zero.
save string If not None, save plot as the given file name with full path.
show bool True Show the plot in a window.
ticks list Where to put the ticks on the colorbar.
ticklabels list How to label the ticks.
title string Title for the plot.
Zdata numpy.ndarray 2D geopotential height field, used for masking intersections with orography.