Document Actions

CDAT Newsletter, November 2007, Tech Tips

by Renata McCoy — last modified 2009-05-28 13:17

	CDAT Newsletter
	Vol4, November, 2007

Home

Tech Tips for beginners

Understanding the time axis.

There is always a lot of questions about the time axis and how to properly understand it and convert between different time representations and different time units, in order to, for example, analyze two datasets with different time representations.

We are providing here the tutorials on the time axis and different ways of working with it.

In this tutorial you will learn how to read tje time axis, change time from relative to component time, or change units of a time axis.,

#############################################
#
# Time Concepts:
#
#############################################

#############################################
#
# Qn : How do I check what times the data is defined for?
# Ans : You can retrieve the time axis using the getTime() method
#       and print out the time axis units and values.
#       For example:
#
#############################################
f = cdms.open(TESTDIR + 'ta_1994-1998.nc')
ta = f('ta')
time_axis = ta.getTime()
print '\nTime units are : \n', time_axis.units
print '\nTime axis values are : \n', time_axis[:]

#############################################
#
# Qn : How do I see what the start time in the data set is?
#
# Ans : The first value in the time axis corresponds to the first time point.
#
##############################################
print '\nFirst time point is ', time_axis[0]

##############################################
#
# Qn : What does this number signify?
# Ans : The number signifies a "relative" time i.e "value since basetime"
#
# Qn : How do I know what "calendar time" this corresponds to?
# Ans : The "calendar time" is referred to as "component time" in CDAT.
#       To get the time in "component" time format:
#
###############################################
import cdtime
print '\nStart time = ', cdtime.reltime(time_axis[0], time_axis.units).tocomp()

# Similarly you can print the last or end time....

print 'End time = ', cdtime.reltime(time_axis[-1], time_axis.units).tocomp()

###############################################
#
# Extra! : You can also just use a function asComponentTime()
# that converts all the time points in the time axis to
# "component time" and then you can just print the start and end times
#
###############################################
all_component_times = time_axis.asComponentTime()
print '\nStart time = ', all_component_times[0],
print 'End time   = ', all_component_times[-1]

###############################################
#
# Qn : How do I extract the data based on time axis values?
# Ans : Since you know the first and last time point you are
#       interested in, you can specify those in the data extraction stage.
#
###############################################
ta_slice = f('ta', time=(0., 15.))
print '\nTime values in ta_slice = ', ta_slice.getTime()[:]

###############################################
#
# Qn : How do I extract a time slice with specific
#       start and end times? Say jan-dec 1996.
# Ans : You can actually specify time = (start, end) in the following way.
#
###############################################
ta_1996_only = f('ta', time=('1996-1-1', '1996-12-1'))

###############################################
#
# A more precise alternative would be to create the
# component times for the start and end time points
# of interest and select the variable at
# 'time = (start_component_time, end_component_time)'
# This would look like:
#
###############################################
st = cdtime.componenttime(1996, 1, 1)
en = cdtime.componenttime(1996, 12, 1)
ta_1996_only = f('ta', time=(st, en))

###############################################
#
# Extra! : We can specify "time selectors" just like we specified
# spatial selectors for re-use. The hypothetical case of the time
# period of interest between May 1996 and Feb 1997 would look like:
#
##############################################
my_time = cdms.selectors.Selector(time =('1995-5-1', '1997-2-1'))

##############################################
#
# Note that we do not need the cdutil.region specification
# because unlike lat and lon dimensions, we do not readjust
# the bounds on time axes.
#
##############################################
ta_time = f('ta', my_time)
print '\nThe data ta_time extracted using my_time starts at ',
print cdtime.reltime(ta_time.getTime()[0], ta_time.getTime().units).tocomp()

----------------------------

# Now creating an axis
## First let's create the value
vals=Numeric.arange(3,24,6,'d')
tim=cdms.createAxis(vals)
tim.units='hours since 2004-6-16'
# Make sure it is a time axis nomater its name
tim.designateTime()
tim.id='time'

## or if we want to change the units to something else
## say some standard units: days since 1900
## The following takes care of converting the values and bounds
## appropriately
tim.toRelativeTime('days since 1900')
## Now let's apply it to data MV array
data.setAxis(0,tim)

Advanced Tech Tips

Spinning the Earth - learn how to make a movie of spinning Earth together with the overplotted cloud fields and wind vectors.

We will show you how to create a very cool presentation of your 3D earth data.

To see the simulation click on the icon below:

Here is a script to produce the simulation above.

#!/usr/bin/env python
"""
This exmaple shows how to draw a spininng animated earth (gif)
TECHNICS DEMONSTRATED: ANIMATED GIF, PROJECTIONS
MODULE USED: cdms,vcs,cdutil
"""

import cdms,vcs, os, sys

f=cdms.open(sys.prefix+'/sample_data/tas_mo.nc')
f1=cdms.open(sys.prefix+'/sample_data/clt.nc')

## Retrieve 2 years
s=f('tas',time=('1980','1982','co'),longitude=(-180,180))

##Retrieve wind data
u=f1('u',plev=200,longitude=(-180,180))
v=f1('v',plev1=200,longitude=(-180,180))

## Reset the time bounds
import cdutil
cdutil.times.setTimeBoundsMonthly(s)

## Computes departures from annual cycle
s=cdutil.times.ANNUALCYCLE.departures(s)

## Initializes VCS
x=vcs.init()
y=vcs.init()

## Create a new template and scale it
t=x.createtemplate('new')
t.scale(.56,axis='x',font=1)

## Create isofill graphic method
iso=x.createisofill('new')

## Create vector graphic method
vc=y.createvector('new')
vc=x.createvector('vector')

# do not plot min, max, mean annotations,
# skip most comments
t.mean.priority = 0
t.min.priority = 0
t.max.priority = 0
t.crtime.priority = 0
t.crdate.priority = 0
t.zvalue.priority = 0
t.tvalue.priority = 0
t.yvalue.priority = 0
t.units.priority = 0
t.tunits.priority = 0
t.zunits.priority = 0
t.yunits.priority = 0
t.xunits.priority = 0
t.comment4.priority = 0
t.comment3.priority = 0
t.comment2.priority = 0
t.comment1.priority = 0
t.dataname.priority = 0
t.title.priority =0
t.source.priority = 1

for i in range(12):
    print 'Slice : ',i,'of 12'
    # Sets the longitude to watch
    lon1=(lon1+180)%360-180
    print lon1
    p.centerlongitude=lon1
    ## retrieve the correct slice
    s2=s(time=slice(i,i+1))
    u2=u()
    v2=v()
    x.plot(s2,t,iso,bg=1)
    u2.long_name = 'deg C, Air Temp & Wind Speed'
    x.plot(u2,v2,t,vc,bg=1)
    t.crdate.priority = 1
    lon1-=30
    a='a'
    if i==0: a='r' # First time create a new gif
    x.gif('spinning_earth',merge=a,
geometry='450x400')
    x.clear()

Home

PCMDI Software Portal

Sections

Personal tools

Document Actions

CDAT Newsletter, November 2007, Tech Tips

Tech Tips for beginners

Advanced Tech Tips