Hi all,
I think Jaison has the right idea for why Sindu was getting unexpected
results. The samplexy function assumes that arguments 2 and 3 are
simple 1-Dimensional lists of values, independent of time.
yes? show func/details samplexy
SAMPLEXY(DAT_TO_SAMPLE,XPTS,YPTS)
Returns data sampled at a set of (X,Y) points, using linear
interpolation
Axes of result:
X: ABSTRACT (result will occupy indices 1...N)
Y: NORMAL (no axis)
Z: inherited from argument(s)
T: inherited from argument(s)
DAT_TO_SAMPLE: variable (x,y,z,t) to sample
Influence on output axes:
X: no influence (indicate argument limits with "[]")
Y: no influence (indicate argument limits with "[]")
Z: passed to result grid
T: passed to result grid
XPTS: X values of sample points
Influence on output axes:
X: no influence (indicate argument limits with "[]")
Y: no influence (indicate argument limits with "[]")
Z: no influence (indicate argument limits with "[]")
T: no influence (indicate argument limits with "[]")
YPTS: Y values of sample points
Influence on output axes:
X: no influence (indicate argument limits with "[]")
Y: no influence (indicate argument limits with "[]")
Z: no influence (indicate argument limits with "[]")
T: no influence (indicate argument limits with "[]")
And Jaison's solution with a REPEAT loop looks like a good one; you
could
alternatively write a netcdf file, defining the variables on the time
axis of the variable long, and appending in time within the repeat
loop.
Ansley
jaison@xxxxxxxxxxxxxx wrote:
Hi,
The reason for the difference between two sampling is the nature in which
SAMPLEXY works.
- SAMPLEXY sample values from the source data, for all given pairs of
xpt and ypt, for each time and depth points of the given time-depth
region.
When the xpts and ypts are 1-D vectors, there is no way to control SAMPLEXY
such that, it samples at first xpt,ypt pair from first time-point of source
data, and at second xpt,ypt pair from second time-point etc.
A simple solution is to do the sampling within a REPEAT loop for every day,
and write the results immediately to a text file, and read it back. An example
using a monthly data (monthly_navy_winds) is shown below, which has been reported
to work well for Sindu's question.
See User Manual Ch3 Sec2.3.34. SAMPLEXY for more details. It should be
noted that the ideal form of xpts and ypts is variables defined on X or Y
axes, instead of Z or T.
Thanks and Regards,
Jaison
!---------example starts here-----------------------------------------------
\ cancel mode verify
use monthly_navy_winds
let long = 80 + SIN(t[gt=uwnd])
let latd = 15 + SIN(t[gt=uwnd])
! get starting and ending L-values for two periods, based on wind data
! (there is no need to give the hh:mm:ss for daily data, unless the
! hr value is 12).
define symbol ts1 = 15-APR-1988 ; let ls1 = `t[gt=uwnd,t=($ts1)],r=lstart`
define symbol te1 = 15-JUN-1988 ; let le1 = `t[gt=uwnd,t=($te1)],r=lstart`
define symbol ts2 = 15-MAY-1988 ; let ls2 = `t[gt=uwnd,t=($ts2)],r=lstart`
define symbol te2 = 15-JUN-1988 ; let le2 = `t[gt=uwnd,t=($te2)],r=lstart`
let lsz1 = le1 - ls1 + 1
let lsz2 = le2 - ls2 + 1
! Use the L-values to constrain sampling along XY, and write results to
! intermediate text files. Use the date-string without hh:mm:ss, as
! the wind data can be daily mean in practice
!
! Add a list or save command to save the wind speed and/or lon, lat variables
! to a text file
set list/precision=7
sp rm -f wind_per1.txt
say
sp echo " Date Lon Lat Wind (m/s)" > wind_per1.txt
sp echo " -------------------------------------------" >> wind_per1.txt
REPEAT/RANGE=`ls1`:`le1`:1/NAME=tt (;\
define symbol ll = `tt` ;\
let date = SUBSTRING("`t[gt=uwnd,d=1,l=($ll)],r=tstart`",1,11) ;\
let xpt = long[t="`date`"] ;\
let ypt = latd[t="`date`"] ;\
let wspd = SAMPLEXY(uwnd[d=1,t="`date`"],`xpt`,`ypt`) ;\
say " Period 1 : L-point = `($ll)-ls1+1`/`lsz1` : Lon = `xpt,p=6` Lat =
`ypt,p=6` Wind = `wspd,p=6`" ;\
sp echo " `date` `xpt,p=7` `ypt,p=7` `wspd,p=7`" >> wind_per1.txt ;\
)
sp rm -f wind_per2.txt
say
sp echo " Date Lon Lat Wind (m/s)" > wind_per2.txt
sp echo " -------------------------------------------" >> wind_per2.txt
REPEAT/RANGE=`ls2`:`le2`:1/NAME=tt (;\
define symbol ll = `tt` ;\
let date = SUBSTRING("`t[gt=uwnd,d=1,l=($ll)],r=tstart`",1,11) ;\
let xpt = long[t="`date`"] ;\
let ypt = latd[t="`date`"] ;\
let wspd = SAMPLEXY(uwnd[d=1,t="`date`"],`xpt`,`ypt`) ;\
say " Period 2 : L-point = `($ll)-ls2+1`/`lsz2` : Lon = `xpt,p=6` Lat =
`ypt,p=6` Wind = `wspd,p=6`" ;\
sp echo " `date` `xpt,p=7` `ypt,p=7` `wspd,p=7`" >> wind_per2.txt ;\
)
say
! how to read these ascii files? here is the best way to do that (for daily
! data)
! (time axis calendar should match with that of wind and lon/lat)
!
! define axis/t="($ts1)":($te1)":1/units=days/T0=31-DEC-1989 tper1
! define grid/t=tper1 gfile
! FILE/grid=gfile/skip=2/format=delimited/delimiter="\b"/\
! type="text,num,num,num"/var="-,-,-,wind" wind_per1.txt
! list wind
!---------example ends here-------------------------------------------------
Thank you Jaison,
Please see the script I used to pick out winds from QSCAT 2006. I've
copied it from the original; it doesnt give any errors while
running in ferret. The version I use is Ferret 6.00
The time axis for Latitude & Longitude is 21-JUL-2005 to 10-AUG-2007, out
which I select my region of interest, 2006. Wind has values from Jan-Dec
2006. Latitude, Longitude & wind are daily data.
-----
let xpts_jandec = long[d=1,t="01-jan-2006:00:00":"31-dec-2006:00:00"]
let ypts_jandec = latd[d=1,t="01-jan-2006:00:00":"31-dec-2006:00:00"]
let winds_jandec = SAMPLEXY(ws,xpts_jandec,ypts_jandec)
let xpts_mayoct = long[d=1,t="01-may-2006:00:00":"31-oct-2006:00:00"]
let ypts_mayoct = latd[d=1,t="01-may-2006:00:00":"31-oct-2006:00:00"]
let winds_mayoct = SAMPLEXY(ws,xpts_mayoct,ypts_mayoct)
----
2. The variables winds_jandec & winds_jandec, have the following grids:
yes? sh da
name title I J K L
WINDS_JANDEC
SAMPLEXY(WS,XPTS_JANDEC,YPTS_JA 1:365 ... ... 1:365
(T=31-DEC-2005 12:00:31-DEC-2006 12:00)
WINDS_MAYOCT
SAMPLEXY(WS,XPTS_MAYOCT,YPTS_MA 1:184 ... ... 1:365
(T=31-DEC-2005 12:00:31-DEC-2006 12:00)
3. The difference in winds_mayoct & winds_jandec (when i=1 is chosen) is
shown below.
yes? li/t="01-may-2006:00:00":"08-may-2006:00:00" winds_jandec[i=1],winds_mayoct[i=1]
Column 1: WINDS_JANDEC is SAMPLEXY(WS,XPTS_JANDEC,YPTS_JANDEC)
Column 2: WINDS_MAYOCT is SAMPLEXY(WS,XPTS_MAYOCT,YPTS_MAYOCT)
WINDS_JANDEC WINDS_MAYOCT
01-MAY-2006 00 / 121: 2.159 4.868
02-MAY-2006 00 / 122: 2.909 4.673
03-MAY-2006 00 / 123: 3.565 4.913
04-MAY-2006 00 / 124: 4.141 5.081
05-MAY-2006 00 / 125: 4.918 5.536
06-MAY-2006 00 / 126: 5.265 5.638
07-MAY-2006 00 / 127: 5.223 5.834
08-MAY-2006 00 / 128: 6.139 6.394
Additionally, the longitude & latitude values for the corresponding days are:
01-MAY-2006 00 / 285: 88.79 15.40
02-MAY-2006 00 / 286: 88.74 15.42
03-MAY-2006 00 / 287: 88.71 15.45
04-MAY-2006 00 / 288: 88.67 15.47
05-MAY-2006 00 / 289: 88.64 15.50
06-MAY-2006 00 / 290: 88.60 15.53
07-MAY-2006 00 / 291: 88.56 15.55
08-MAY-2006 00 / 292: 88.52 15.57
Both xpts_jandec & xpts_mayoct (ypts_jandec & ypts_mayoct) list the same values.
Hope the problem is better conveyed this time.
thanks
Sindu
On Fri, 4 Jul 2008, jaison@xxxxxxxxxxxxxx wrote:
Hi Sindu,
Setting region/selecting xpts and ypts along time for SAMPLEXY
input should work just fine, unless and otherwise there is something wrong
with your time axis (or it can be a bug with SAMPLEXY...but hold on...).
However, I will be really surprised if the given script has worked without
any error. Because :
1. Ferret (or any similar tools) will not allow minus sign (-) in
variable name, since it is a "special operator/symbol". Ferret
refers such names as "illegal" (please see user manual,
Ch3 Sec1.2. File variables, for example), as shown below
yes? let xpts_jan-dec = 100
**ERROR: command syntax: illegal name: XPTS_JAN-DEC
yes?
2. The time region syntax is wrong (double quotes are not closed).
So, please do the following :
- make sure the time axes of lon, lat and wind variables are ok, and
matches well (number of points, dates etc.)
- define variables without minus sign (xpts_dec2jan, ypts_dec2jan.....)
- demonstrate the error with listing few values
- send a mail back with all details, demonstrating the error.
If you are providing lines from your script, please make sure that you
are making an exact copying. With these details, we can have a better
idea about the problem....and a better solution!
Thanks and Regards,
Jaison
Dear Ferreters,
I'm facing a problem with the function SAMPLEXY as I use it to pick out
values from a gridded dataset (eg wind speed).
I have xpts and ypts from Jan-Dec and I select the lat,lon points
in 2 ways: 1) Jan-Dec and 2) May-Oct.
The wind values I obtain when I select xpts & ypts from Jan-Dec are
different from when they are selected only for May-Oct.
Please see my script below.
1. Why isnt the wind during May-Oct in both cases identical? &
2. Why does the May-Oct selection give me wind data from Jan-Dec?
Am I missing something important while selecting the lat and
lon points in time and using it in combination with SAMPLEXY ?
Thanks
Sindu
------ script-----------------------------------------------
let xpts_jan-dec = longitude[t="01-jan-2006:00:00":"31-dec-2006:00:00]
let ypts_jan-dec = latitude[t="01-jan-2006:00:00":"31-dec-2006:00:00]
let winds_jan-dec = SAMPLEXY(wind,xpts_jan-dec,ypts_jan-dec)
let xpts_may-oct = longitude[t="01-may-2006:00:00":"31-oct-2006:00:00]
let ypts_may-oct = latitude[t="01-may-2006:00:00":"31-oct-2006:00:00]
let winds_may-oct = SAMPLEXY(wind,xpts_may-oct,ypts_may-oct)
------ script-----------------------------------------------
Sindu Raj Parampil
Ph.D Student
Centre for Atmospheric and Oceanic Sciences
Indian Institute of Science
Bangalore-12
INDIA
=========================================================================
"You don't understand anything until you learn it more than one way."
Marvin Minsky, 'The Society of Mind'
=========================================================================
--
Sindu Raj Parampil
Ph.D Student
Centre for Atmospheric and Oceanic Sciences
Indian Institute of Science
Bangalore-12
INDIA
=========================================================================
"You don't understand anything until you learn it more than one way."
Marvin Minsky, 'The Society of Mind'
=========================================================================
--
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
|