Hi Upendra,
Thank you for the report, and I understand your concern. When LAS
reads this time series (via Ferret), the software is not returning
any error code, so LAS doesn't have any way to report that anything
is wrong. In general, I would say that if you see errors such as
Malformed or inaccessible DAP DATADDS
with any slice of a dataset, it's an indication of problems with the
dataset or aggregation or the thredds server (as you know).
However I can do a few things to check into this.
Following the Info link from this dataset in your LAS, the URL for
the data is shown to be
http://data.nodc.noaa.gov/thredds/dodsC/pathfinder/testdata/crw/oa_data/aggregations/yearly_aggregate_2000.ncml
If I look at that URL directly in Ferret, using the latest official
release of Ferret, I see the kind of thing that's showing up in
your LAS. This version of Ferret is linked with NetCDF4.1.1 for
reading NetCDF files and OPeDNAP url's. Here are the Ferret
commands to list the NetCDF version, and list a time series from the
data. The netCDF library is making a data request, and data is read
from the URL without any error flags.
yes? show symbol netcdf*
NETCDF_VERSION = "4.1.1 of Sep 16 2010 10:42:11 $"
yes? use
"http://data.nodc.noaa.gov/thredds/dodsC/pathfinder/testdata/crw/oa_data/aggregations/yearly_aggregate_2000.ncml"
yes? list/x=60w/y=22 pH
VARIABLE : NOAA Coral Reef Watch Experimental
Ocean Acidification Product Suite - pH 25km (Total Scale)
DATA SET : Archived NOAA Coral Reef Watch 25km
Ocean Acidification Product Suite for the Greater Caribbean
Region -monthly, 2000
FILENAME : yearly_aggregate_2000.ncml
FILEPATH :
http://data.nodc.noaa.gov/thredds/dodsC/pathfinder/testdata/crw/oa_data/aggregations/
SUBSET : 12 points (TIME)
LONGITUDE: 60.1W
LATITUDE : 21.9N
60.13W
121
15-JAN-2000 / 1: 82.2
14-FEB-2000 / 2: 257.0
16-MAR-2000 / 3: 82.4
15-APR-2000 / 4: 151.1
16-MAY-2000 / 5: 82.2
15-JUN-2000 / 6: 295.5
...
When the same Ferret code is linked with the most-recent beta NetCDF
library, I see what looks like to be the correct result
yes? show symbol netcdf*
NETCDF_VERSION = "4.1.2-beta2-snapshot2011010522 of Jan 6 2011
10:14:32 $"
yes? use
"http://data.nodc.noaa.gov/thredds/dodsC/pathfinder/testdata/crw/oa_data/aggregations/yearly_aggregate_2000.ncml"
yes? list/x=60w/y=22 pH
...
SUBSET : 12 points (TIME)
LONGITUDE: 60.1W
LATITUDE : 21.9N
60.13W
121
15-JAN-2000 / 1: 8.100
14-FEB-2000 / 2: 8.100
16-MAR-2000 / 3: 8.100
15-APR-2000 / 4: 8.100
16-MAY-2000 / 5: 8.100
15-JUN-2000 / 6: 8.100
...
So it seems that the upcoming netCDF4.1.2 release does something
that is more robust with respect to whatever is wrong with the
dataset. Unidata is planning a release of netCDF4.1.2 very soon, and
we will be releasing a new version of Ferret which will be linked
with that.
Ansley
On 1/7/2011 11:02 AM, Upendra Dadi wrote:
Hi,
On 1/7/2011 11:02 AM, Upendra Dadi wrote:
Hi,
I have a dataset which is not being read properly by THREDDS due
to some issues with the file holding the dataset. But when
plotting a time-series graph involving a data point from this
file, it still generates a graph. But the plot is wrong. I am not
sure where LAS is getting data for this plot. As an example,
choose the following dataset from
http://data.nodc.noaa.gov/las/getUI.do :
NODE > Archived NOAA Coral Reef Watch 25km Ocean Acidification
Product Suite.... > Aggregation by year > 2000 > pH 25km
When plotting the latitude-longitude map for April with this
dataset, I get an error (though the error message itself doesn't
seem to be representative of the problem). But when plotting time
series plot, it doesn't give any error message. But, the values in
the plot are wrong. The pH values as far as I know should be
between 0-14, but the plot shows that the values are between 80
and 300. You will also see that the latitude-longitude maps for
all other months is fine. Apparently, the problem is occurring due
to data from the month of April. LAS is not catching the problem.
The solution for me would be to replace the dataset on THREDDS
with correct datasets. I just wanted to let you know anyway since
LAS is not catching the problem.
Upendra
|