<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p>Hi Jacob.</p>
<p>Many thanks for letting me know the reference values for the DFI
filter parameters you use operationally with you NAM CONUS. <br>
</p>
<p>Definitely, we will do some verification, some parallel runs with
different filter parameters. In our case, so far, we are not doing
analysis, so interpolating from global model (10 or more km
horizontal resolution) to NMMB 2.5 km resolution. It can be that
the error introduced by the resolution change jumping needs
another kind of filter parameters, no clear idea, I am not an
expert in balancing stuff. In any case, we now have all the
ingredients to tune our DFI and to have the surface parameters at
t=0! <br>
</p>
<p>Many thank to all.</p>
<p>Pau<br>
</p>
<br>
<div class="moz-cite-prefix">El 20/04/18 a les 14:11, Jacob Carley -
NOAA Federal ha escrit:<br>
</div>
<blockquote type="cite"
cite="mid:CADWMng1=n8oFMXj_ad1-E3PTxEGW2xUbfW+tDxt99bCdBKx_tA@mail.gmail.com">
<div dir="ltr">Hi Pau,
<div><br>
</div>
<div>If you run one of the DFI variants that involves a
backward, adiabatic integration at such a high resolution you
can run into instability issues. I suggest keeping the filter
window length short and making the filter timestep short as
well. For reference, here is what we use for the operational
3km NAM CONUS Nest at NOAA/NCEP/EMC:</div>
<div><br>
</div>
<blockquote style="margin:0 0 0 40px;border:none;padding:0px">
<div>
<div>##########################################</div>
</div>
<div>
<div>##### Fundamental Timestep (seconds) #####</div>
</div>
<div>
<div>##########################################</div>
</div>
<div>
<div><br>
</div>
</div>
<div>
<div>dt_int: 6 # Integer seconds</div>
</div>
<div>
<div>dt_num: 1 # Numerator of fractional second</div>
</div>
<div>
<div>dt_den: 4 # Denominator of fractional second</div>
</div>
<div>
<div><br>
</div>
</div>
<div>
<div>filt_dt_int: 2 # Integer seconds</div>
</div>
<div>
<div>filt_dt_num: 0 # Numerator of fractional
second</div>
</div>
<div>
<div>filt_dt_den: 4 # Denominator of fractional
second</div>
</div>
<div><br>
</div>
<div><br>
</div>
<div>
<div>###################################</div>
</div>
<div>
<div>##### Digital Filtering options ###</div>
</div>
<div>
<div>###################################</div>
</div>
<div>
<div><br>
</div>
</div>
<div>
<div>filter_method: 3 # 0--->none ;
1--->DFL ; 2---> DDFI ; 3--->TDFI</div>
</div>
<div>
<div><br>
</div>
</div>
<div>
<div>nsecs_dfl: 1800 # HALF foward filter window
(s) (DFL)</div>
</div>
<div>
<div>nsecs_bckddfi: 1800 # HALF backward integration
duration (s) (DDFI)</div>
</div>
<div>
<div>nsecs_fwdddfi: 3600 # HALF foward filter window
(s) (DDFI)</div>
</div>
<div>
<div>nsecs_bcktdfi: 600 # HALF backward filter
window (s) (TDFI)</div>
</div>
<div>
<div>nsecs_fwdtdfi: 600 # HALF forward filter window
(s) (TDFI)</div>
</div>
</blockquote>
<div><br>
</div>
Hope this helps!
<div>Jacob<br>
<div><br>
</div>
</div>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">On Fri, Apr 20, 2018 at 6:30 AM, Marco
Biordi <span dir="ltr"><<a
href="mailto:marco.biordi@gmail.com" target="_blank"
moz-do-not-send="true">marco.biordi@gmail.com</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div dir="ltr">
<div>
<div>We're using 3600.<br>
<br>
</div>
Anyway keep this value to be an integer multiple of the
time step, elsewhere give error.<span class="HOEnZb"><font
color="#888888"><br>
<br>
</font></span></div>
<span class="HOEnZb"><font color="#888888">Marco<br>
</font></span></div>
<div class="HOEnZb">
<div class="h5">
<div class="gmail_extra"><br>
<div class="gmail_quote">2018-04-20 9:50 GMT+00:00 Pau
<span dir="ltr"><<a
href="mailto:pescribaa@aemet.es" target="_blank"
moz-do-not-send="true">pescribaa@aemet.es</a>></span>:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div text="#000000" bgcolor="#FFFFFF">
<p>Hi Marco,</p>
<p>Thanks very much for the answer, we are going
to test it right now.</p>
<p>We should see if DFI changes our verification
scores but I would expect it doesn't, taking
into account we are not doing analysis cycling
with NMMB. <br>
</p>
<p>Just a small question, apart from
filter_method=3, are using the regular values
for filtering parameters:</p>
<p>nsecs_bcktdfi: 3600 # HALF backward
filter window (s) (TDFI)<br>
nsecs_fwdtdfi: 3600 # HALF forward
filter window (s) (TDFI)<br>
</p>
<p>?</p>
<p>In our case we are running with around 2.5 km
horizontal resolution and 5.625 seconds of
time step. Probably we should make some
tunning of this parameters...<br>
</p>
<p>Many thanks.</p>
<p>Pau<br>
</p>
<br>
<div
class="m_-6799693077398366421m_4924499790318523189moz-cite-prefix">El
20/04/18 a les 10:37, Marco Biordi ha escrit:<br>
</div>
<div>
<div class="m_-6799693077398366421h5">
<blockquote type="cite">
<div dir="ltr">
<div>
<div>Hello Pau,<br>
<br>
</div>
you can get t=0 missing parameters
when you enable the Digital Filter
Initialization.<br>
<br>
Do that by setting configuration
option "filter_method" to value 1 or 2
or 3 (usual one).<br>
<br>
</div>
<div>Regards<br>
</div>
<div><br>
Marco Biordi<br>
<br>
</div>
<div>San Marino Republic Weather
Forecast Center<br>
</div>
<div><a href="http://meteo.sm"
target="_blank"
moz-do-not-send="true">meteo.sm</a><br>
</div>
<br>
</div>
<div class="gmail_extra"><br>
<div class="gmail_quote">2018-04-20 8:26
GMT+00:00 Pau <span dir="ltr"><<a
href="mailto:pescribaa@aemet.es"
target="_blank"
moz-do-not-send="true">pescribaa@aemet.es</a>></span>:<br>
<blockquote class="gmail_quote"
style="margin:0 0 0
.8ex;border-left:1px #ccc
solid;padding-left:1ex">Hello all,<br>
<br>
I am a new user of NMMB model. I
work at the Spanish Met Agency with
<br>
several NWP-LAM models. We are
running daily NMMB at home and we
see <br>
that, after UPP processing, we don't
get some surface parameters like <br>
t2m, rh2m, u10m, v10m or visibility,
at t=0, but we do for any lead time
<br>
t>0. After dedicating some time
to look to UPP and NMMB user's
guides <br>
and namelists, I am quite sure this
variables cannot be got in an easy <br>
manner for t=0. Could anybody
confirm me this? In such a case, we
have <br>
to do some further final work with
vertical interpolations and so.<br>
<br>
Thanks in advance!<br>
<br>
Pau Escriba<br>
<br>
Spanish Met Agency (AEMET)<br>
<br>
______________________________<wbr>_________________<br>
Ncep.list.emc.nmmb-questions-a<wbr>nd-answers
mailing list<br>
<a
href="mailto:Ncep.list.emc.nmmb-questions-and-answers@lstsrv.ncep.noaa.gov"
target="_blank"
moz-do-not-send="true">Ncep.list.emc.nmmb-questions-a<wbr>nd-answers@lstsrv.ncep.noaa.go<wbr>v</a><br>
<a
href="https://www.lstsrv.ncep.noaa.gov/mailman/listinfo/ncep.list.emc.nmmb-questions-and-answers"
rel="noreferrer" target="_blank"
moz-do-not-send="true">https://www.lstsrv.ncep.noaa.g<wbr>ov/mailman/listinfo/ncep.list.<wbr>emc.nmmb-questions-and-answers</a><br>
</blockquote>
</div>
<br>
</div>
</blockquote>
<br>
</div>
</div>
</div>
</blockquote>
</div>
<br>
</div>
</div>
</div>
<br>
______________________________<wbr>_________________<br>
Ncep.list.emc.nmmb-questions-<wbr>and-answers mailing list<br>
<a
href="mailto:Ncep.list.emc.nmmb-questions-and-answers@lstsrv.ncep.noaa.gov"
moz-do-not-send="true">Ncep.list.emc.nmmb-questions-<wbr>and-answers@lstsrv.ncep.noaa.<wbr>gov</a><br>
<a
href="https://www.lstsrv.ncep.noaa.gov/mailman/listinfo/ncep.list.emc.nmmb-questions-and-answers"
rel="noreferrer" target="_blank" moz-do-not-send="true">https://www.lstsrv.ncep.noaa.<wbr>gov/mailman/listinfo/ncep.<wbr>list.emc.nmmb-questions-and-<wbr>answers</a><br>
<br>
</blockquote>
</div>
<br>
</div>
</blockquote>
<br>
</body>
</html>