[ncl-talk] calculating annual max values with varying amount of yearly values

Mon Sep 15 10:32:44 MDT 2014

Hi Alan,

Great, thanks that did work!  I went about doing it the following way.  Not as elegant, but I checked both and they yield the same results:

do i = 0,tot_events-1
  if(i .ne. 0)then
     beg_vals(i) = ind(year(i) .gt. year(i-1))
     if(i .ne. tot_events-1) then
                 end_vals(i) = ind(year(i) .lt. year(i+1))
     else
               end_vals(tot_events-1)=tot_events-1
     end if
  else
     beg_vals(0) = 1
     end_vals(0) = ind(year(i) .lt. year(i+1))
  end if
end do

beg_index=ind(.not.ismissing(beg_vals))
end_index=ind(.not.ismissing(end_vals))

max_year_temp=new(dimsizes(end_index),float)
max_year=new(10000,float,0)

do i = 0,dimsizes(end_index)-1

               max_year(i)=max(city_list(beg_index(i):end_index(i)))
end do

From: Alan Brammer [mailto:abrammer at albany.edu]
Sent: Friday, September 12, 2014 3:24 PM
To: Marcella, Marc
Cc: ncl-talk at ucar.edu
Subject: Re: [ncl-talk] calculating annual max values with varying amount of yearly values

Does this do what you originally questioned?
I may have interpreted the year file incorrectly.

Good luck,

years = asciiread("cittagong_years.txt",-1, "float")
wind = asciiread("cittagong_wind_all.txt",-1, "float")

year_min = min(years)
year_max = max(years)
year_span = fspan( year_min, year_max, toint((year_max-year_min)+1) )

output = new( dimsizes(year_span), typeof(wind) )
output!0 = "year"
output&year = year_span

do y=year_min, year_max
  year_ind = ind(years .eq. y)
  if(all(ismissing(year_ind)))
    delete(year_ind)
    continue            ;;;  if there are no matching years then move onto the next year
  end if
  output({y}) = max( wind(year_ind) )  ; will return FillValue if all values == FillValue.
  delete(year_ind)
end do

On Fri, Sep 12, 2014 at 2:56 PM, Marcella, Marc <MMarcella at air-worldwide.com<mailto:MMarcella at air-worldwide.com>> wrote:
Hi Alan,

Thanks, Ive attached the file for the wind values and the corresponding file for the years corresponding to each event.

Thanks!
Marc

From: Alan Brammer [mailto:abrammer at albany.edu<mailto:abrammer at albany.edu>]
Sent: Friday, September 12, 2014 2:40 PM

To: Marcella, Marc
Cc: ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
Subject: Re: [ncl-talk] calculating annual max values with varying amount of yearly values

It might be worth attaching an example of both input files (year and value).  I'm struggling to work out your file structure.  It seems like it would be possible to create a single array of all your values and then use ind() to select the relevant values for each year.

Good luck,
Alan.

On Fri, Sep 12, 2014 at 1:39 PM, Marcella, Marc <MMarcella at air-worldwide.com<mailto:MMarcella at air-worldwide.com>> wrote:
Yeah, im really trying to streamline this and do it all in one language/program. I have to do it multiple times so it would be much better just to it all in NCL.  But thank you for your help/suggestions. Hopefully someone out there will have a solution!!

From: jbuzan [mailto:jbuzan at purdue.edu<mailto:jbuzan at purdue.edu>]
Sent: Friday, September 12, 2014 1:17 PM

To: Marcella, Marc
Cc: ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
Subject: Re: [ncl-talk] calculating annual max values with varying amount of yearly values

hmmmm…
You could use a csh script (or whatever shell you are comfortable with) to delimitate the text file year, start, and end values.  You feed this into environment variables and import into ncl.  ncl would then use ascii read, and the getenv to set the ranges.  After that, run stat_dispersion, and export to netcdf.

http://www.ncl.ucar.edu/Document/Functions/Built-in/asciiread.shtml
https://www.ncl.ucar.edu/Document/Functions/Built-in/fspan.shtml
http://www.ncl.ucar.edu/Document/Functions/Built-in/getenv.shtml

-Jonathan

On Sep 12, 2014, at 1:05 PM, Marcella, Marc <MMarcella at AIR-WORLDWIDE.COM<mailto:MMarcella at AIR-WORLDWIDE.COM>> wrote:

Hmm thanks, but therein lies the problem, the start and end or increment is variable from year to year (and the year values are in a separate text file than the wind values which are in a text file) So start end for year one may be 1 4 but for year 2 it may be 5 5 and year 3 6-12, etc.

From: jbuzan [mailto:jbuzan at purdue.edu]
Sent: Friday, September 12, 2014 1:04 PM
To: Marcella, Marc
Cc: ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
Subject: Re: [ncl-talk] calculating annual max values with varying amount of yearly values

ah.
Well, a quick way to deal with this is to parse the data into years using nco commands.  (ncks -d time start,end,increment data_in.nc<http://data_in.nc> data_out.nc<http://data_out.nc>).  Then run stat_dispersion.

-Jonathan

On Sep 12, 2014, at 1:01 PM, Marcella, Marc <MMarcella at AIR-WORLDWIDE.COM<mailto:MMarcella at AIR-WORLDWIDE.COM>> wrote:

This is 1D data. It’s an array for the wind value for each event for a given year at a specific lat/lon location

From: jbuzan [mailto:jbuzan at purdue.edu]
Sent: Friday, September 12, 2014 1:01 PM
To: Marcella, Marc
Cc: ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
Subject: Re: [ncl-talk] calculating annual max values with varying amount of yearly values

stat_dispersion only works on 1D data.  So you have to loop through each grid cell.  However, it will provide you with the local statistics of each grid cell.

-Jonathan

On Sep 12, 2014, at 12:55 PM, Marcella, Marc <MMarcella at AIR-WORLDWIDE.COM<mailto:MMarcella at AIR-WORLDWIDE.COM>> wrote:

Yes, I know of stat dispersion, but wont this look at the statistics/distribution of the entire dataset? I was looking to figure out only for each year. I believe I could use the max command provided I can get the beginning and final index value for each year easily enough.

From: jbuzan [mailto:jbuzan at purdue.edu]
Sent: Friday, September 12, 2014 12:54 PM
To: Marcella, Marc
Cc: ncl-talk at ucar.edu<mailto:ncl-talk at ucar.edu>
Subject: Re: [ncl-talk] calculating annual max values with varying amount of yearly values

I recommend using stat_dispersion:
http://www.ncl.ucar.edu/Document/Functions/Contributed/stat_dispersion.shtml

-Jonathan

On Sep 12, 2014, at 12:49 PM, Marcella, Marc <MMarcella at AIR-WORLDWIDE.COM<mailto:MMarcella at AIR-WORLDWIDE.COM>> wrote:

Hi all,

Im having a bit of a problem trying to elegantly script calculating annual max values when I have varying amount of values for each year. For example, I have one text file with the year stamp for each event, and in another I have the corresponding wind value for that event.  Some years there are four measurements, some years one, other years ten. Im trying to use a do loop that goes through each year in the text file to find the beginning and end index (i.e. what index refers to the beginning of that year’s values in the wind file and the index that refers to the last value of that year).  I was then going to use that specific year’s indices with the max command on the wind value file.  But I’m having not having too much luck and it’s getting messy with all the do and if statements.  Any ideas?  Are there simpler functions that Im not aware of built into NCL that can help with this?  Any help would be greatly appreciated.

Thanks,
Marc
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20140915/1de3a6cc/attachment.html 