Re: problem in running regresst

[Ansley Manke <Ansley.B.Manke@xxxxxxxx>]

Hi Dongxiao,
The script regresst.jnl defines a number of variables in terms of the
inputs p and q.  The calculation of the values of these variables
depends on the time region that is in effect when these variables are
evaluated; i.e. on the LIST or PLOT command that uses qhat.
When you specify a time in the command, Ferret uses that time to
go and calculate qhat. When you said

   list/x=180e/y=0/t=16-nov-1975 qhat

Ferret  evaluated qhat using only that single value of time, not the entire
range of time for the input data. To compute the regression for the entire
series, but list or plot the value for only a subset of time, you can define
the inputs to regresst with an explicit range on the time axis, specifying
the entire time range.  This could be done either with indices or with dates.
These commands give the correct results:

   let maxt = `sst,return=lsize`

   let p=t[g=sst,L=1:`maxt`]
   let q=sst[L=1:`maxt`]
   set grid q
   go regresst

   list/x=180e/y=0/t=16-aug-1975 qhat

The LIST command computes qhat for all time in the range L=1:`maxt`
and at the specified X and Y, and then lists only the value at the
specified T.


Ansley

Dongxiao Zhang wrote:

> Dear all,
>
> Try this script and please let me know what goes run. Thanks.
>
> !get COADS sst anomaly
> use
> "http://iridl.ldeo.columbia.edu/SOURCES/.DASILVA/.SMD94/.anomalies/.sst/dods";
>
> let p=t[g=sst]
> let q=sst
> set grid q
> go regresst
> plot/x=180e/y=0 qhat !shows a nice trend and a good value on 16-AUG-1975
> list/x=180e/y=0/t=16-aug-1975 qhat !but gives a missing value
> cont/t=16-aug-1975 q-qhat !shows bad value everywhere
>
> --------------------------------------------------------------------
> Dr. Dongxiao Zhang
> NOAA/PMEL Bld#3
> 7600 Sand Point Way NE
> Seattle, WA 98115-0070
> Phone  : (206)526-4184
> --------------------------------------------------------------------

--
Ansley Manke  Pacific Marine Environmental Laboratory  Seattle WA  (206)526-6246