Steve,
Is the total dimension of the data in temp.cdf the same as the 12x2
dimensions of the grid you define? That is, does tmax have 24 points?
or perhaps does it start with some missing values, which are getting
mapped
onto the new grid?
RESHAPE takes the entire variable tmax (unless you specify a subset of
it in the call to RESHAPE) and unravels it in standard Fortran order,
then wraps that list of data values onto the output grid.
Ansley
Steve Knox wrote:
Has any one out there used the "reshape" function?
I seem to be still missing something re the "reshape" function.
Here's what I'm doing but I get and empty dataset returned, although I
expect
some data.
use temp.cdf
DEFINE AXIS/T=1951:1952:1 tyear
DEFINE AXIS/Z=1:12:1 zmonth
LET out_grid = Z[GZ=zmonth]+T[GT=tyear]
LET my_reshaped = RESHAPE(tmax, out_grid)
SHOW GRID my_reshaped
GRID (G002)
name axis # pts start end
normal X
normal Y
ZMONTH Z 12 r 1 12
TYEAR T 2 r 1951 1952
(the grid seems ok by when I try to list the data, I get an empty set)
list my_reshaped
VARIABLE : RESHAPE(TMAX, OUT_GRID)
FILENAME : temp.cdf
SUBSET : 12 by 2 points (Z-T)
1 2 3 4 5 6 7 8 9 10 11 12
1 2 3 4 5 6 7 8 9 10 11 12
1951 / 1:................................................
1952 / 2:................................................
Thanks,
Ansley Manke wrote:
Steve,
The message you're seeing is just a warning that "MONTHS" is an
ambiguous
unit; it depends on the length of the year (models might use
even-spaced,
30-
day months, or you might have in mind unequally-spaced months based on
the
actual number of days in each calendar month). This warning is to make
sure
you're aware of this.
It looks like the axis tcal you've created doesn't quite match the
range
of your
data -- your data starts in 1901 if its time origin is in January of
1900
and the
first time step is 12. In fact, I don't think there's any reason to
regrid
your time
series at all. The example in the Users Guide is just creating an
example
time
series to use. If you just create the T (year) and Z (month) axes to
correspond
to
the length of your time series and do the RESHAPE I think it'll work
fine.
Ansley
Steve Knox wrote:
Thanks for pointing me to the reshape command.
I seem to be doing something wrong, however, when I try to follow the
example.
Why do I get an error from UNITS=MON. (If I omit "UNITS=MON" I get no
warning message but the results are the same).
The reshaped grid seems ok but when I list my_reshaped I get no data.
The time variable in my file tmp.cdf is :
float time(time) ;
time:units = "months since 1900-01-15 00:00:00" ;
time:title = "Time" ;
time:long_name = "Time axis" ;
time:time_origin = " 1900-01-15 00:00:00" ;
The variable time has values:
time = 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27,
28,
...
...
When I follow the example this is what I get:
use tmp.cdf
DEFINE AXIS/t=1-JAN-1900:31-DEC-1903/NPOINTS=48/UNITS=MON tcal
*** NOTE: /UNIT=MONTHS is ambiguous ... using 1/12 of 365.2425 days
Replacing definition of axis TCAL
LET my_time_series = tmp[gt=tcal] ! tmp is my average temp variable
DEFINE AXIS/t=1900:1903:1 tyear
DEFINE AXIS/Z=1:12:1 zmonth
LET out_grid = Z[GZ=zmonth]+T[GT=tyear]
LET my_reshaped = RESHAPE(my_time_series, out_grid)
SHOW GRID my_reshaped
GRID (G002)
name axis # pts start end
normal X
normal Y
ZMONTH Z 12 r 1 12
TYEAR T 4 r 1900 1903
list my_reshaped
VARIABLE : RESHAPE(MY_TIME_SERIES, OUT_GRID)
FILENAME : tmp.cdf
SUBSET : 12 by 4 points (Z-T)
1 2 3 4 5 6 7 8 9 10 11 12
1 2 3 4 5 6 7 8 9 10 11 12
1900 / 1:................................................
1901 / 2:................................................
1902 / 3:................................................
1903 / 4:................................................
Ansley Manke wrote:
Hi
Steve,
The question of how to represent data as year vs month-of-year happens
to be explained as an example of the RESHAPE function. Take a look in
the Users Guide under reshape.
Ansley Manke
Steve Knox wrote:
I have a netcdf file with 1 variable, tmp for latitudes -90 to 90 (by 1/2 deg) and longitudes -180 to 180 (by 1/2 degree)
The time unit is "months since 1900-01-15 00:00:00". There are 1200 months (100 years).
I would like to plot year along the x-axis and month (1-12) on the y-axis for a paricular spatial aggregation.
The "value" plotted should be a shaded area, representing the value at each month. I want to look for monthly
or seasonal variations over time.
12 B A
...
3 A
2 B A
1 B B
1901 1902 ...
Where the letters" images would actually be shaded
areas.
Is this possible with Ferret?
yes? use tmp
yes? show data
currently SET data sets:
1> ./tmp.cdf (default)
name title I J
K L
TMP Surface air temperature 1:720 1:360
...
1:1200
--
_________________________
Steve Knox
NREL
Colorado State University
(970)491-1972
--
_________________________
Steve Knox
NREL
Colorado State University
(970)491-1972
--
_________________________
Steve Knox
NREL
Colorado State University
(970)491-1972
|