Re: Joining Daymet Tiles with NCL

Ping,

Your information helped me find a new problem in the join script.
When two adjacent tiles are separated by water, the previous version
would output an invalid merged file with corrupted coordinates, and no
warning message. This could easily explain your original problem.

I added a consistency check for bad joins with no common grid point,
and prevented the script from making a bad output file. I have also
added notes to the web page about Daymet tiles with small data grids.
Thank you for your help in finding this problem.

Please download and try the latest join script from the web page:
http://www.esrl.noaa.gov/psd/people/dave.allured/data/daymet/daymet.html

--Dave

On Sat, Oct 27, 2012 at 2:56 PM, Dave Allured <dave.allured@noaa.gov> wrote:
> Ping,
>
> I looked at 11393 on the coast. Daymet apparently minimizes the
> included X/Y grid for each tile to cover points over land only. So
> the tile collection includes some micro grids. Church Lady says,
> isn't that special!
>
> Well, the join script is capable of joining files of differing grid
> sizes, because that is the default paradigm for Daymet, anyway. So
> you are safe to join the small tiles with the normal ones.
>
> The primary joining rule remains. Any two files (or "grids") to be
> joined *by this method* must have at least one grid point in common.
> So a subsize tile along the coastline should join with the next tile
> inland, with no problem.
>
> Things may break down if you try to join two tiles with no land along
> the common edge. Now I see this particular problem in your tile map
> in yesterday's post. The script will probably fail if you try to join
> 11936 and 11937 directly (east-west). So join 11937 (Yarmouth, Nova
> Scotia) to land to the north first, to make a "bridge" to the
> continent. Also, please do inspect the console very carefully for
> error messages, when joining many tiles under control of a driver
> script. HTH.
>
> --Dave
>
> On Fri, Oct 26, 2012 at 8:10 PM, Ping Yang <pyang@ccny.cuny.edu> wrote:
>> Hi,
>>
>> I just ran all the original files, there are several files(size are very
>> small, especially tiles at the east corner near the atlantic sea) are not
>> 100% percent for every tile, some of them like:
>> 11393 50%
>> 11756 20 percent.
>>
>> Should I just exclude them(but it will lose the data)?
>>
>> Is there a way to deal with those files?
>>
>> Looking forward to hearing from you.
>>
>> Thank you.
>>
>> Regards,
>>
>> Ping
>> On Fri, Oct 26, 2012 at 9:10 PM, Dave Allured <dave.allured@noaa.gov> wrote:
>>>
>>> One more thing. Would you please run the *other* plot script,
>>> plot.daymet-orig.1024.ncl, to quickly step through all 68 original
>>> times and just make sure they all "look" okay. That is, the square
>>> areas with data should appear to be be 100% present for every tile,
>>> and adjacent tiles should "look" like they align properly. This is
>>> not a rigorous test, just a quick visual check. Thank you.
>>>
>>> --Dave
>>>
>>> On Fri, Oct 26, 2012 at 7:02 PM, Dave Allured <dave.allured@noaa.gov>
>>> wrote:
>>> > Hi Ping.
>>> >
>>> > Your problem may be in the plot and not the merged data file. Note that
>>> > on
>>> > my web page I say: "Streaking may occur if any of the missing 2-D
>>> > coordinates are within the map boundaries". If you made that plot with
>>> > NCL,
>>> > then you are probably getting severe streaking from the missing 12 tiles
>>> > in
>>> > the southwest corner.
>>> >
>>> > An easy check would to be to reduce the map limits in your plotting
>>> > script,
>>> > and only plot rectangular areas that are completely covered by Daymet
>>> > tiles
>>> > with data.
>>> >
>>> > If this is not the problem, then you may need to check very carefully
>>> > for
>>> > unexpected gaps in the 2-D coordinates, or similar problems. You may
>>> > have
>>> > also found another bug in the join script. You are only the second
>>> > person
>>> > to ever try it. ;-)
>>> >
>>> > But please first try plotting some map subsets. If you can localize
>>> > where
>>> > the problem is in the merged data by this method, that would help the
>>> > diagnosis.
>>> >
>>> > --Dave
>>> >
>>> > On Fri, Oct 26, 2012 at 4:59 PM, Ping Yang <pyang@ccny.cuny.edu> wrote:
>>> >>
>>> >> Dear NCL,
>>> >>
>>> >> The Daymet
>>> >>
>>> >> http://www.esrl.noaa.gov/psd/people/dave.allured/data/daymet/daymet.html is
>>> >> great.
>>> >>
>>> >> Thank you so much.
>>> >>
>>> >> I got the script for join 68 tiles for the Northeast U.S. using the
>>> >> attached scripts(a version from the newest ncl code and a bash script I
>>> >> created for the joining)
>>> >>
>>> >> However, I found some thing wrong here, I am copying a picture here:
>>> >>
>>> >> the tiles that I am working on is:
>>> >>
>>> >> I think there is some problem with the joining of with the row 117 and
>>> >> row
>>> >> 119.
>>> >>
>>> >> I didn't figure out where is the problem.
>>> >>
>>> >> Could you please give me some help.
>>> >>
>>> >> Best Regards,
>>> >>
>>> >> Ping
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Sat Oct 27 18:05:18 2012