|
###################################################
# PYTHON / PLANETPLOT #
###################################################
# ------------------- XY plots ------------------ #
# Added a new category of plot to unidim, contour, etc... called "xy"
# - xy plots are plots that do not use time,vert,lat or lon as axis
# - variables to be plotted are stored in plot_x and plot_y, which is done in
# select_getfield. (there is no "what_I_plot" var for "wy" plots)
# - plot_x and plot_y are also subject to reduce field. A None value indicates
# the plot is not 'xy'
# - "xy" plots are used for a specific subset of plots : histograms, fourier
# transforms, hodographs
# - added option --analysis to perform certain kinds of analysis on the data
# (corresponding to xy plots).
# - One selects the fields he wants to plot (e.g. -v UV --lon 0 --lat 20) and
# chooses a kind of analysis :
--analysis fft # in the particular case given above, this mode corresponds
# to the mean of the ampitude spectrum of the vertical spatial
# fast fourier transform taken at each time index
# note that for now, fft are always done along the vertical
# axis. This could be made more flexible.
# not that the minimum wavelength you can plot depends on the
# vertical step of your simulation. THIS STEP HAS TO BE
# CONSTANT, hence you MUST use API or ZRECAST with constant
# spacing.
--analysis histo # histogram on the flattened data. If the user asks for --lon 0,20,
# the average is done before computing the histogram (contrary to the fft).
# However, if given a 2D array (with only --lon and --lat on a
# 4D field for example), the data is flattened before computation
# so that the result is still a 1D histogram.
--analysis histodensity # histogram with a kernel density estimate to a
# gaussian distribution, also giving the mean, variance,
# skewness and kurtosis. Other distributions are available in
# the scipy.stats package and could be implemented.
# - added variables "-v hodograph" and "-v hodograph_2". First one is a
# regular hodograph with "u" and "v" as axis, with labels of the local times
# (use --axtime lt). This is a "xy" plot (you must specify a vertical level
# as well, usually --vert 0 with an interpolation at 5m (-i 4 -l 0.005)).
# Second one is the variation with local time (for exemple) of the wind
# rotation (arctan(v/u)). This is not a "xy" plot but "unidim".
# For --ope plots in "xy" cases, only the operation plot is displayed.
# ------------------- Operations ------------------ #
# - For operations --ope -, the histogram (fourth plot) has been removed. To get it
# back, call --ope -_histo.
# - _only has been added to "+","-","-%" operations (eg "-_only")
# - For operations "-","+","-%","-_only","+_only","-%_only","-_histo", it is now
# possible to call multiple files (sill one variable) and compare each of them
# with the unique given reference file. Ex: -f file1,file2 --fref file3 --ope - will give 6 plots:
file1 file3 file1-file3
file2 file3 file2-file3
# In the case of "xy" plots, the multiple operation plots are regrouped on a
# single plot (using multiple lines). the title of this plot is not 'fig(2)-fig(1)'
# (default) but the argument of the --title command. Labels have to be given
# as following : --labels "dummy","dummy","file1-file3 label","dummy","dummy","file2-file3 label" (dummy can be anything. this is to be improved)
# To be able to run on multiple files and easily introduce the correct number and order of plots, these operations have been moved outside of the main loop
# on namefiles and variables. Operation of the type "add_var" or "cat" have
# been left inside the loop and are unchanged.
# ------------------- Localtime ------------------ #
# Changed the way localtime is computed. Reasons:
# - it was assuming one timestep per hour in computations mixing indices and
# actual times
# - to determine the starting date, it was using the name of the file (for Meso), instead of using the netcdf
# attribute (START_DATE)
# - it was using the computed mean longitude of the domain, which is not correct for
# hemispheric domain. => better to use the netcdf attribute CEN_LON
# - it was using this mean longitude even for profile plots at given
# longitude => better to get the local time at this given lon, especially
# important for large domains
# => new localtime() is in myplot (old one is commented). Interv is obsolete (but not removed yet). Case "Geo" has not been looked at.
# new localtime in myplot correctly account for starting date of the file in
# all cases
# accounts for local longitude of the plot
# accounts for files that do not have per hour outputs, but per timestep.
# specific cases can be added in myplot in localtime()
# ------------------ Misc ------------------ #
# - added option --xlog to get x logarithmic axis (--ylog already existed)
# - added the possibility to use 2 files of different gridding (-f
# file1,file2) although of the same type (meso for exemple). For that purpose,
# lon, lat, alt and vert arrays are now indexed with 'index_f', as for
# all_var. => all_lon, all_lat, all_lat, all_vert
# - added function teta_to_tk in myplot so that a call to pp.py on standard
# LMD mesoscale file with "-v tk" can be done without the need to call API.
# The temperature is computed from T and PTOT, knowing P0, T0 and R_CP.
# - bug correction in determineplot() that was causing wrong plot number and
# slices number when not calling averaged lon, lat, vert or times. (which is
# often !)
# - added new options to redope: --redope edge_x1, edge_x2, edge_y1, edge_y2
# which plots the boundaries of the domain. This is different compared to
# asking for a fixed longitude, because the domain boundary might not be at constant
# longitude (hemispheric domain for exemple). x1 is the western boundary, x2
# the eastern, y1 the southern and y2 the northern. x1 and x2 reduce the
# dimension along --lon, y1 and y2 reduce the dimension along --lat.
# - added control in windamplitude() to determine whether winds are staggered or
# unstaggered. This is usefull when dealing with non LMD_MMM files.
# - corrected a bug in reduce_field where the mean was computed on the wrong
# axis !!! (pretty serious bug)
# Exemple of plots you can do with these new options can be found in
$YOUR_SVN/trunk/UTIL/PYTHON/Intercomparison/Plots_MasterScript/bam.sh
# ------------------ API ------------------ #
# - changed maximum of levels from 299 to 1000 in API (interpolation on 1000 levels is
# usefull to get larger bandwidth in fourier transform)
# the following concerns users of MRAMS files.
# - API has not been modified for MRAMS files. Instead, a python script
# (ic.py) is run on MRAMS .ctl and .dat files, which automatically format those files
# to be API and pp.py compatible.
# - ic.py is in $YOUR_SVN/trunk/UTIL/PYTHON/Intercomparison/File_conversion
###################################################
# INTERCOMPARISON TOOLS: #
###################################################
#ic.py in
$YOUR_SVN/trunk/UTIL/PYTHON/Intercomparison/File_conversion
#CDO installer with import_binary in
$YOUR_SVN/trunk/UTIL/PYTHON/Intercomparison/CDO
#Plotting scripts in
$YOUR_SVN/trunk/UTIL/PYTHON/Intercomparison/Plots_MasterScript
#1D sensibility tool in
$YOUR_SVN/trunk/UTIL/PYTHON/Intercomparison/1D_sensibility_tool
# See README in each of these folders for details.
|
