[Met_help] [rt.rap.ucar.edu #95578] History for Combining Multiple runs of Series_Analysis
John Halley Gotway via RT
met_help at ucar.edu
Fri Jun 12 10:00:39 MDT 2020
----------------------------------------------------------------
Initial Request
----------------------------------------------------------------
Hi,
I have been using series_analysis, which is a really nice tool for looking at spatial variation.
Whilst in the past this has been used for case studies, just putting long lists of files through,
I was hoping to use it in a more routine fashion,
i.e. running series after each forecast run / or each day and gradually accumulating stats over a longer period.
The main reason for doing this is so that I do not have to have all the input data available at run time, when assessing a long time period, but can instead accumulate in smaller runs, reducing the data volumes needed.
So essentially stat aggregation of series_analysis output (where appropriate, e.g. partial sums/ contingency counts).
Stat_analysis already does this sort of thing for text output, but I don't think it does it for NetCDF.
So question 1) Is there something already existing amongst the Met tools which I can use to aggregate the series_analysis output.
2) If not, is this something that can potentially be added to the MET package in future developments.
(I have a fallback of running series_analysis each run and combining the resulting NetCDF files via some bespoke code I can put together , but just wanted to check that it wasn't already possible before going down that route)
Cheers,
Ric
Ric Crocker Senior Scientist
Met Office FitzRoy Road Exeter Devon EX1 3PB United Kingdom
Tel: +44(0)330 135 2137
Email: ric.crocker at metoffice.gov.uk Website: www.metoffice.gov.uk
----------------------------------------------------------------
Complete Ticket History
----------------------------------------------------------------
Subject: Combining Multiple runs of Series_Analysis
From: Minna Win
Time: Thu Jun 11 09:03:59 2020
Hello Ric,
It looks like you have a few questions about using the MET
series_analysis
tool. I'm assigning this ticket to John Halley Gotway. Please allow
a few
business days for a full response.
Regards,
Minna
---------------
Minna Win
National Center for Atmospheric Research
Developmental Testbed Center
Phone: 303-497-8423
Fax: 303-497-8401
On Thu, Jun 11, 2020 at 7:50 AM Crocker, Richard via RT
<met_help at ucar.edu>
wrote:
>
> Thu Jun 11 07:50:39 2020: Request 95578 was acted upon.
> Transaction: Ticket created by ric.crocker at metoffice.gov.uk
> Queue: met_help
> Subject: Combining Multiple runs of Series_Analysis
> Owner: Nobody
> Requestors: ric.crocker at metoffice.gov.uk
> Status: new
> Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95578 >
>
>
> Hi,
> I have been using series_analysis, which is a really nice tool for
looking
> at spatial variation.
> Whilst in the past this has been used for case studies, just putting
long
> lists of files through,
> I was hoping to use it in a more routine fashion,
> i.e. running series after each forecast run / or each day and
gradually
> accumulating stats over a longer period.
> The main reason for doing this is so that I do not have to have all
the
> input data available at run time, when assessing a long time period,
but
> can instead accumulate in smaller runs, reducing the data volumes
needed.
>
> So essentially stat aggregation of series_analysis output (where
> appropriate, e.g. partial sums/ contingency counts).
> Stat_analysis already does this sort of thing for text output, but I
don't
> think it does it for NetCDF.
>
> So question 1) Is there something already existing amongst the Met
tools
> which I can use to aggregate the series_analysis output.
> 2) If not, is this something that can
potentially
> be added to the MET package in future developments.
>
> (I have a fallback of running series_analysis each run and combining
the
> resulting NetCDF files via some bespoke code I can put together ,
but just
> wanted to check that it wasn't already possible before going down
that
> route)
>
> Cheers,
> Ric
>
> Ric Crocker Senior Scientist
> Met Office FitzRoy Road Exeter Devon EX1 3PB United Kingdom
> Tel: +44(0)330 135 2137
> Email: ric.crocker at metoffice.gov.uk Website: www.metoffice.gov.uk
>
>
>
------------------------------------------------
Subject: Combining Multiple runs of Series_Analysis
From: John Halley Gotway
Time: Thu Jun 11 13:07:35 2020
Hi Ric,
You raise a very good question about Series-Analysis. There currently
is no
existing functionality in MET designed to accumulate statistics across
multiple runs of Series-Analysis. The closest existing functionality
is in
the pcp_combine tool, which can be run to compute the mean of multiple
fields. That might be useful for a field of mean error (ME)
statistics, but
would not properly handle other statistics like RMSE or any of the
categorical stats.
I can see how Series-Analysis could be enhanced to read it's own
output and
properly aggregate statistics across multiple runs. Similar to the
functionality of Stat-Analysis, the required inputs would be based on
the
requested outputs. For example, when contingency table statistics are
requested, the tool would require input fields for the 4 cells of a
2x2
contingency table. Similarly, when continuous statistics are requested
(e.g. RMSE), it'd need the columns of the SL1L2 line type as input.
For
probabilistic stats, we'd need all the cells of the Nx2 probabilistic
contingency table. From a design perspective, this could either be an
enhancement to the existing tool or a new tool entirely. For
convenience, I
think it makes sense to just enhance Series-Analysis to do this. It
should
be easy enough to check the format of the input files up front and
branch
the processing logic from there.
Another idea to consider is whether it'd be feasible to configure
Series-Analysis to enable MPR output. By that I mean, just write out
the
raw series of FCST and OBS pairs that define the time-series.
Subsequent
runs of Series-Analysis could just read in those FCST and OBS series,
concatenate them, and recompute statistics from them. That would
obviously
make for much larger output files, but would provide the most
flexibility
in aggregating across multiple runs. But perhaps that just shifts the
storage problem from one spot to another.
Do you have an opinion on the solution of aggregating partial sums and
contingency table counts versus concatenating and processing series of
matched pairs?
I would say that this is a non-trivial task. The first step is writing
up a
GitHub issue (or issues) to describe the feature request. And from
there,
we determine how to pay for and when to do the work.
Thanks,
John
On Thu, Jun 11, 2020 at 9:04 AM Minna Win via RT <met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95578 >
>
> Hello Ric,
>
> It looks like you have a few questions about using the MET
series_analysis
> tool. I'm assigning this ticket to John Halley Gotway. Please
allow a few
> business days for a full response.
>
> Regards,
> Minna
> ---------------
> Minna Win
> National Center for Atmospheric Research
> Developmental Testbed Center
> Phone: 303-497-8423
> Fax: 303-497-8401
>
>
>
> On Thu, Jun 11, 2020 at 7:50 AM Crocker, Richard via RT
<met_help at ucar.edu
> >
> wrote:
>
> >
> > Thu Jun 11 07:50:39 2020: Request 95578 was acted upon.
> > Transaction: Ticket created by ric.crocker at metoffice.gov.uk
> > Queue: met_help
> > Subject: Combining Multiple runs of Series_Analysis
> > Owner: Nobody
> > Requestors: ric.crocker at metoffice.gov.uk
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95578 >
> >
> >
> > Hi,
> > I have been using series_analysis, which is a really nice tool for
> looking
> > at spatial variation.
> > Whilst in the past this has been used for case studies, just
putting long
> > lists of files through,
> > I was hoping to use it in a more routine fashion,
> > i.e. running series after each forecast run / or each day and
gradually
> > accumulating stats over a longer period.
> > The main reason for doing this is so that I do not have to have
all the
> > input data available at run time, when assessing a long time
period, but
> > can instead accumulate in smaller runs, reducing the data volumes
needed.
> >
> > So essentially stat aggregation of series_analysis output (where
> > appropriate, e.g. partial sums/ contingency counts).
> > Stat_analysis already does this sort of thing for text output, but
I
> don't
> > think it does it for NetCDF.
> >
> > So question 1) Is there something already existing amongst the Met
tools
> > which I can use to aggregate the series_analysis output.
> > 2) If not, is this something that can
potentially
> > be added to the MET package in future developments.
> >
> > (I have a fallback of running series_analysis each run and
combining the
> > resulting NetCDF files via some bespoke code I can put together ,
but
> just
> > wanted to check that it wasn't already possible before going down
that
> > route)
> >
> > Cheers,
> > Ric
> >
> > Ric Crocker Senior Scientist
> > Met Office FitzRoy Road Exeter Devon EX1 3PB United Kingdom
> > Tel: +44(0)330 135 2137
> > Email: ric.crocker at metoffice.gov.uk Website: www.metoffice.gov.uk
> >
> >
> >
>
>
------------------------------------------------
Subject: Combining Multiple runs of Series_Analysis
From: John Halley Gotway
Time: Thu Jun 11 21:50:41 2020
Ric,
I captured the details of this feature request in this GitHub issue:
https://github.com/NCAR/MET/issues/1371
Thanks,
John
On Thu, Jun 11, 2020 at 1:07 PM John Halley Gotway <johnhg at ucar.edu>
wrote:
> Hi Ric,
>
> You raise a very good question about Series-Analysis. There
currently is
> no existing functionality in MET designed to accumulate statistics
across
> multiple runs of Series-Analysis. The closest existing functionality
is in
> the pcp_combine tool, which can be run to compute the mean of
multiple
> fields. That might be useful for a field of mean error (ME)
statistics, but
> would not properly handle other statistics like RMSE or any of the
> categorical stats.
>
> I can see how Series-Analysis could be enhanced to read it's own
output
> and properly aggregate statistics across multiple runs. Similar to
the
> functionality of Stat-Analysis, the required inputs would be based
on the
> requested outputs. For example, when contingency table statistics
are
> requested, the tool would require input fields for the 4 cells of a
2x2
> contingency table. Similarly, when continuous statistics are
requested
> (e.g. RMSE), it'd need the columns of the SL1L2 line type as input.
For
> probabilistic stats, we'd need all the cells of the Nx2
probabilistic
> contingency table. From a design perspective, this could either be
an
> enhancement to the existing tool or a new tool entirely. For
convenience, I
> think it makes sense to just enhance Series-Analysis to do this. It
should
> be easy enough to check the format of the input files up front and
branch
> the processing logic from there.
>
> Another idea to consider is whether it'd be feasible to configure
> Series-Analysis to enable MPR output. By that I mean, just write out
the
> raw series of FCST and OBS pairs that define the time-series.
Subsequent
> runs of Series-Analysis could just read in those FCST and OBS
series,
> concatenate them, and recompute statistics from them. That would
obviously
> make for much larger output files, but would provide the most
flexibility
> in aggregating across multiple runs. But perhaps that just shifts
the
> storage problem from one spot to another.
>
> Do you have an opinion on the solution of aggregating partial sums
and
> contingency table counts versus concatenating and processing series
of
> matched pairs?
>
> I would say that this is a non-trivial task. The first step is
writing up
> a GitHub issue (or issues) to describe the feature request. And from
there,
> we determine how to pay for and when to do the work.
>
> Thanks,
> John
>
>
>
> On Thu, Jun 11, 2020 at 9:04 AM Minna Win via RT <met_help at ucar.edu>
> wrote:
>
>>
>> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95578 >
>>
>> Hello Ric,
>>
>> It looks like you have a few questions about using the MET
series_analysis
>> tool. I'm assigning this ticket to John Halley Gotway. Please
allow a
>> few
>> business days for a full response.
>>
>> Regards,
>> Minna
>> ---------------
>> Minna Win
>> National Center for Atmospheric Research
>> Developmental Testbed Center
>> Phone: 303-497-8423
>> Fax: 303-497-8401
>>
>>
>>
>> On Thu, Jun 11, 2020 at 7:50 AM Crocker, Richard via RT <
>> met_help at ucar.edu>
>> wrote:
>>
>> >
>> > Thu Jun 11 07:50:39 2020: Request 95578 was acted upon.
>> > Transaction: Ticket created by ric.crocker at metoffice.gov.uk
>> > Queue: met_help
>> > Subject: Combining Multiple runs of Series_Analysis
>> > Owner: Nobody
>> > Requestors: ric.crocker at metoffice.gov.uk
>> > Status: new
>> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95578 >
>> >
>> >
>> > Hi,
>> > I have been using series_analysis, which is a really nice tool
for
>> looking
>> > at spatial variation.
>> > Whilst in the past this has been used for case studies, just
putting
>> long
>> > lists of files through,
>> > I was hoping to use it in a more routine fashion,
>> > i.e. running series after each forecast run / or each day and
gradually
>> > accumulating stats over a longer period.
>> > The main reason for doing this is so that I do not have to have
all the
>> > input data available at run time, when assessing a long time
period, but
>> > can instead accumulate in smaller runs, reducing the data volumes
>> needed.
>> >
>> > So essentially stat aggregation of series_analysis output (where
>> > appropriate, e.g. partial sums/ contingency counts).
>> > Stat_analysis already does this sort of thing for text output,
but I
>> don't
>> > think it does it for NetCDF.
>> >
>> > So question 1) Is there something already existing amongst the
Met tools
>> > which I can use to aggregate the series_analysis output.
>> > 2) If not, is this something that can
potentially
>> > be added to the MET package in future developments.
>> >
>> > (I have a fallback of running series_analysis each run and
combining the
>> > resulting NetCDF files via some bespoke code I can put together ,
but
>> just
>> > wanted to check that it wasn't already possible before going down
that
>> > route)
>> >
>> > Cheers,
>> > Ric
>> >
>> > Ric Crocker Senior Scientist
>> > Met Office FitzRoy Road Exeter Devon EX1 3PB United Kingdom
>> > Tel: +44(0)330 135 2137
>> > Email: ric.crocker at metoffice.gov.uk Website: www.metoffice.gov.uk
>> >
>> >
>> >
>>
>>
------------------------------------------------
Subject: RE: [rt.rap.ucar.edu #95578] Combining Multiple runs of Series_Analysis
From: Crocker, Richard
Time: Fri Jun 12 00:56:43 2020
Thanks John,
That’s a really good summary.
I think my preference would be to avoid the MPR files route, because I
can imagine them becoming very large very quickly
(though I have no data on hand to confirm the relative size of the 2
potential solutions).
I have half an eye on scalability as model resolution increases (and
even sub hourly sampling for that matter), so tend to err on the side
of caution.
As you say, the MPR files would make aggregating a lot easier and
flexible.
Thank you for opening the ticket,
I am assuming that any further questions/comments on this should go
via that ticket as the preferred route of communication?
Cheers,
Ric
-----Original Message-----
From: John Halley Gotway via RT <met_help at ucar.edu>
Sent: 11 June 2020 20:08
To: Crocker, Richard <ric.crocker at metoffice.gov.uk>
Subject: Re: [rt.rap.ucar.edu #95578] Combining Multiple runs of
Series_Analysis
Hi Ric,
You raise a very good question about Series-Analysis. There currently
is no existing functionality in MET designed to accumulate statistics
across multiple runs of Series-Analysis. The closest existing
functionality is in the pcp_combine tool, which can be run to compute
the mean of multiple fields. That might be useful for a field of mean
error (ME) statistics, but would not properly handle other statistics
like RMSE or any of the categorical stats.
I can see how Series-Analysis could be enhanced to read it's own
output and properly aggregate statistics across multiple runs. Similar
to the functionality of Stat-Analysis, the required inputs would be
based on the requested outputs. For example, when contingency table
statistics are requested, the tool would require input fields for the
4 cells of a 2x2 contingency table. Similarly, when continuous
statistics are requested (e.g. RMSE), it'd need the columns of the
SL1L2 line type as input. For probabilistic stats, we'd need all the
cells of the Nx2 probabilistic contingency table. From a design
perspective, this could either be an enhancement to the existing tool
or a new tool entirely. For convenience, I think it makes sense to
just enhance Series-Analysis to do this. It should be easy enough to
check the format of the input files up front and branch the processing
logic from there.
Another idea to consider is whether it'd be feasible to configure
Series-Analysis to enable MPR output. By that I mean, just write out
the raw series of FCST and OBS pairs that define the time-series.
Subsequent runs of Series-Analysis could just read in those FCST and
OBS series, concatenate them, and recompute statistics from them. That
would obviously make for much larger output files, but would provide
the most flexibility in aggregating across multiple runs. But perhaps
that just shifts the storage problem from one spot to another.
Do you have an opinion on the solution of aggregating partial sums and
contingency table counts versus concatenating and processing series of
matched pairs?
I would say that this is a non-trivial task. The first step is writing
up a GitHub issue (or issues) to describe the feature request. And
from there, we determine how to pay for and when to do the work.
Thanks,
John
On Thu, Jun 11, 2020 at 9:04 AM Minna Win via RT <met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95578 >
>
> Hello Ric,
>
> It looks like you have a few questions about using the MET
> series_analysis tool. I'm assigning this ticket to John Halley
> Gotway. Please allow a few business days for a full response.
>
> Regards,
> Minna
> ---------------
> Minna Win
> National Center for Atmospheric Research Developmental Testbed
Center
> Phone: 303-497-8423
> Fax: 303-497-8401
>
>
>
> On Thu, Jun 11, 2020 at 7:50 AM Crocker, Richard via RT
<met_help at ucar.edu
> >
> wrote:
>
> >
> > Thu Jun 11 07:50:39 2020: Request 95578 was acted upon.
> > Transaction: Ticket created by ric.crocker at metoffice.gov.uk
> > Queue: met_help
> > Subject: Combining Multiple runs of Series_Analysis
> > Owner: Nobody
> > Requestors: ric.crocker at metoffice.gov.uk
> > Status: new
> > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95578 >
> >
> >
> > Hi,
> > I have been using series_analysis, which is a really nice tool for
> looking
> > at spatial variation.
> > Whilst in the past this has been used for case studies, just
putting long
> > lists of files through,
> > I was hoping to use it in a more routine fashion,
> > i.e. running series after each forecast run / or each day and
gradually
> > accumulating stats over a longer period.
> > The main reason for doing this is so that I do not have to have
all the
> > input data available at run time, when assessing a long time
period, but
> > can instead accumulate in smaller runs, reducing the data volumes
needed.
> >
> > So essentially stat aggregation of series_analysis output (where
> > appropriate, e.g. partial sums/ contingency counts).
> > Stat_analysis already does this sort of thing for text output, but
I
> don't
> > think it does it for NetCDF.
> >
> > So question 1) Is there something already existing amongst the Met
tools
> > which I can use to aggregate the series_analysis output.
> > 2) If not, is this something that can
potentially
> > be added to the MET package in future developments.
> >
> > (I have a fallback of running series_analysis each run and
combining the
> > resulting NetCDF files via some bespoke code I can put together ,
but
> just
> > wanted to check that it wasn't already possible before going down
that
> > route)
> >
> > Cheers,
> > Ric
> >
> > Ric Crocker Senior Scientist
> > Met Office FitzRoy Road Exeter Devon EX1 3PB United Kingdom
> > Tel: +44(0)330 135 2137
> > Email: ric.crocker at metoffice.gov.uk Website: www.metoffice.gov.uk
> >
> >
> >
>
>
------------------------------------------------
Subject: Combining Multiple runs of Series_Analysis
From: John Halley Gotway
Time: Fri Jun 12 09:59:37 2020
Ric,
Yes, please feel free to add comments to that ticket that are relevant
specifically to that development. Please direct questions about other
aspects of series-analysis or any of the other MET tools to met_help
instead. We're trying to keep GitHub content focused on development
and
met_help focused on support issues. When the two get mixed, it's easy
to
lose track of things.
Since we've documented a path forward, I'll go ahead and resolve this
ticket.
Thanks,
John
On Fri, Jun 12, 2020 at 12:57 AM Crocker, Richard via RT
<met_help at ucar.edu>
wrote:
>
> <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95578 >
>
> Thanks John,
> That’s a really good summary.
>
> I think my preference would be to avoid the MPR files route, because
I can
> imagine them becoming very large very quickly
> (though I have no data on hand to confirm the relative size of the 2
> potential solutions).
> I have half an eye on scalability as model resolution increases (and
even
> sub hourly sampling for that matter), so tend to err on the side of
caution.
> As you say, the MPR files would make aggregating a lot easier and
flexible.
>
> Thank you for opening the ticket,
> I am assuming that any further questions/comments on this should go
via
> that ticket as the preferred route of communication?
>
> Cheers,
> Ric
>
>
> -----Original Message-----
> From: John Halley Gotway via RT <met_help at ucar.edu>
> Sent: 11 June 2020 20:08
> To: Crocker, Richard <ric.crocker at metoffice.gov.uk>
> Subject: Re: [rt.rap.ucar.edu #95578] Combining Multiple runs of
> Series_Analysis
>
> Hi Ric,
>
> You raise a very good question about Series-Analysis. There
currently is
> no existing functionality in MET designed to accumulate statistics
across
> multiple runs of Series-Analysis. The closest existing functionality
is in
> the pcp_combine tool, which can be run to compute the mean of
multiple
> fields. That might be useful for a field of mean error (ME)
statistics, but
> would not properly handle other statistics like RMSE or any of the
> categorical stats.
>
> I can see how Series-Analysis could be enhanced to read it's own
output
> and properly aggregate statistics across multiple runs. Similar to
the
> functionality of Stat-Analysis, the required inputs would be based
on the
> requested outputs. For example, when contingency table statistics
are
> requested, the tool would require input fields for the 4 cells of a
2x2
> contingency table. Similarly, when continuous statistics are
requested
> (e.g. RMSE), it'd need the columns of the SL1L2 line type as input.
For
> probabilistic stats, we'd need all the cells of the Nx2
probabilistic
> contingency table. From a design perspective, this could either be
an
> enhancement to the existing tool or a new tool entirely. For
convenience, I
> think it makes sense to just enhance Series-Analysis to do this. It
should
> be easy enough to check the format of the input files up front and
branch
> the processing logic from there.
>
> Another idea to consider is whether it'd be feasible to configure
> Series-Analysis to enable MPR output. By that I mean, just write out
the
> raw series of FCST and OBS pairs that define the time-series.
Subsequent
> runs of Series-Analysis could just read in those FCST and OBS
series,
> concatenate them, and recompute statistics from them. That would
obviously
> make for much larger output files, but would provide the most
flexibility
> in aggregating across multiple runs. But perhaps that just shifts
the
> storage problem from one spot to another.
>
> Do you have an opinion on the solution of aggregating partial sums
and
> contingency table counts versus concatenating and processing series
of
> matched pairs?
>
> I would say that this is a non-trivial task. The first step is
writing up
> a GitHub issue (or issues) to describe the feature request. And from
there,
> we determine how to pay for and when to do the work.
>
> Thanks,
> John
>
>
>
> On Thu, Jun 11, 2020 at 9:04 AM Minna Win via RT <met_help at ucar.edu>
> wrote:
>
> >
> > <URL: https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95578 >
> >
> > Hello Ric,
> >
> > It looks like you have a few questions about using the MET
> > series_analysis tool. I'm assigning this ticket to John Halley
> > Gotway. Please allow a few business days for a full response.
> >
> > Regards,
> > Minna
> > ---------------
> > Minna Win
> > National Center for Atmospheric Research Developmental Testbed
Center
> > Phone: 303-497-8423
> > Fax: 303-497-8401
> >
> >
> >
> > On Thu, Jun 11, 2020 at 7:50 AM Crocker, Richard via RT <
> met_help at ucar.edu
> > >
> > wrote:
> >
> > >
> > > Thu Jun 11 07:50:39 2020: Request 95578 was acted upon.
> > > Transaction: Ticket created by ric.crocker at metoffice.gov.uk
> > > Queue: met_help
> > > Subject: Combining Multiple runs of Series_Analysis
> > > Owner: Nobody
> > > Requestors: ric.crocker at metoffice.gov.uk
> > > Status: new
> > > Ticket <URL:
https://rt.rap.ucar.edu/rt/Ticket/Display.html?id=95578
> >
> > >
> > >
> > > Hi,
> > > I have been using series_analysis, which is a really nice tool
for
> > looking
> > > at spatial variation.
> > > Whilst in the past this has been used for case studies, just
putting
> long
> > > lists of files through,
> > > I was hoping to use it in a more routine fashion,
> > > i.e. running series after each forecast run / or each day and
gradually
> > > accumulating stats over a longer period.
> > > The main reason for doing this is so that I do not have to have
all the
> > > input data available at run time, when assessing a long time
period,
> but
> > > can instead accumulate in smaller runs, reducing the data
volumes
> needed.
> > >
> > > So essentially stat aggregation of series_analysis output
(where
> > > appropriate, e.g. partial sums/ contingency counts).
> > > Stat_analysis already does this sort of thing for text output,
but I
> > don't
> > > think it does it for NetCDF.
> > >
> > > So question 1) Is there something already existing amongst the
Met
> tools
> > > which I can use to aggregate the series_analysis output.
> > > 2) If not, is this something that can
> potentially
> > > be added to the MET package in future developments.
> > >
> > > (I have a fallback of running series_analysis each run and
combining
> the
> > > resulting NetCDF files via some bespoke code I can put together
, but
> > just
> > > wanted to check that it wasn't already possible before going
down that
> > > route)
> > >
> > > Cheers,
> > > Ric
> > >
> > > Ric Crocker Senior Scientist
> > > Met Office FitzRoy Road Exeter Devon EX1 3PB United Kingdom
> > > Tel: +44(0)330 135 2137
> > > Email: ric.crocker at metoffice.gov.uk Website:
www.metoffice.gov.uk
> > >
> > >
> > >
> >
> >
>
>
>
>
------------------------------------------------
More information about the Met_help
mailing list