.. _user.faq.graphics: Graphiques ========== How to modify the plot parameters? ---------------------------------- Edit the Matplotlib configuration file. It is typically located here: :file:`~/.matplotlib/matplotlibrc`. For more information: http://matplotlib.sourceforge.net/users/customizing.html My map takes time to draw ------------------------- The map layouts that take time are those of a reduced area which use a high resolution coastline ("f" from GSHHS or "s" for Histolitt from SHOM). In the case of GSHHS traîts, a card caching system plotted with mod:`~mpl_toolkits.basemap` is used (see :mod:`vacumm.misc.grid.basemap`): if a map using the same zoom and resolution must be traced, it is loaded from the cache. For your information, cached maps are stored in the directory given by the function :func:`~vacumm.misc.grid.basemap.get_map_dir` (a priori :file:`~/.matplotlib/basemap/cached_maps`). .. note:: This cache system can generate errors during use of a saved card. See:`user.faq.graphics.err_basemap`. In addition, a map may want to appear on the screen while you're doing a remote plot, which can take time. If you want to disable this display, read : :ref:`user.faq.graphics.window`. How to tuner a graph to plot? ----------------------------- VACUMM's graphical routines transfer keywords to most of the matplotlib functions used. We specify a keyword to transfer a prefixing this one of the name of the function suffixed with a"_" (use: func:`~vacumm.misc.misc.kwfilter` to filter keywords). For example, it is possible to change the size of the title of a graph, and the thickness of the contours as follows:: >>> map2(sst, title_size=20, contour_linewidths=3) In this specific case, to have the list of the other possible options we refer to the documentation of functions :func:`matplotlib.pyplot.title` and func:`matplotlib.pyplot.contour`. In addition, Matplotlib plotted objects are often accessible using the :meth:`~vacumm.misc.core_plot.Plot.get_obj` method that returns a list:: >>> m = map2((u, v), show=False) >>> myquiver = m.get_obj('quiver')[0] >>> mycontour = m.get_obj('contour')[0] >>> myquiver.set_zorder(10) >>> m.show() Note you must set the ``show`` parameter to ``False`` if your want to tune your plot (write mode), which is not necessary if you just want to gather come parameters. How to place a colorbar yourself? --------------------------------- Turn off the automatic colorbar plot and figure display, then draw it by specifying its position (cf. : func:`matplotlib.pyplot.axes`) : m = map2(data, colorbar=False, right=.9, show=False) m.colorbar(cax=[.91, .2, .3, .6]) m.show() or directly map2(data, colorbar_cax==[.91, .2, .3, .6], right=.9, show=False) See: func:`matplotlib.pyplot.colorbar` for plotting options. .. _user.faq.graphics.window: My figure is not displayed / turn off the display of figures like in batch mode ------------------------------------------------------------------------------- The creation of figures requires the use of a graphical software layer. These software layers are called backends. The minimum for the saving the figures is to use the backend ``Agg``, this allows also to bypass errors when using batch scripts when the display is not available. The other backends require a display, the ``TkAgg`` backend is almost systematically available. The qt4agg backend is also available with UVCDAT. If you want to make a display of your figures, use ``qt4agg``, if you use ``Agg``, no display will be made. There are two ways to modify the backend: 1) The permanent one consists in modifying the corresponding line in the Matplotlib configuration file:: backend : Agg 2) The second way is to do it on the fly, at the beginning of a script (before any loading of :mod:`matplotlib`) and :mod:`vacumm`):: from matplotlib import use use('Agg') How to disable automatic zoom? ------------------------------ For example when you draw a map with func:`~vacumm.misc.plot.map2`, if the unmasked data does not cover the entire domain, automatic zoom will avoid drawing areas that are not covered. You can avoid this by using the ``xmasked=False`` options, symasked=False`, see ``xymasked=False`` to disable zoom on both axes (in the case of a 2D plot):: map(sst, xymasked=False) How to add shadow or glow effects? -------------------------------------------------- Most of the time, keywords are provided for this purpose for the plotting functions. For example, if you draw contours, you can do this to add a shadow to these and an effect glow to labels:: map2(data, contour_shadow=True, clabel_glow=True) You can pass keywords to modify effects (see :func:`~vacumm.misc.core_plot.add_shadow` and :func:`~vacumm.misc.core_plot.add_glow` to know the options):: map2(data, contour_shadow_width=4, contour_shadow_xoffset=3) You can also proceed manually for an external trace:: m = map2(data, show=False) m.add_shadow(m.axes.plot(x,y)) # m.axes.plot or pylab.plot or:: from vacumm.misc.core_plot import add_shadow add_shadow(P.plot(x,y)) .. _user.faq.graphics.err_basemap: I have an error with class :class:`~mpl_toolkits.basemap.Basemap` ----------------------------------------------------------------- It may be that the use of a map (typically when plotting with :func:`~vacumm.misc.plot.map2`) produces an error (for example a missing attribute, such as :attr:`celestial` or :attr:`_mapboundarydrawn`). This is probably related to the automatic caching of already drawn maps. This process saves time when plotting a map with exactly the same characteristics (domain, projection, coastline). However, the cached card may not be compatible with the current version of :mod:`~mpl_toolkits.basemap`, following a python package update. To solve the problem radically, you can delete all cached cards. They are found by default in the directory given by the function :func:`~vacumm.misc.grid.basemap.get_map_dir`. There is a function to perform this operation:: >>> from vacumm.misc.grid.basemap import reset_cache >>> reset_cache()