Page 9 of 10

Re: NEMS NMMB topic

Posted: Fri Mar 10, 2017 2:11 pm
by oib
Hi guys,

are you still running NEMS-NMMB? I do it :-)

I read on NCEP site that on March 15, 2017 they will implement newer Version North American Mesoscale (NAM), version 4.

Significative NMMB model physics improvements will be:
  1. Advect specific humidity every dynamics time step
  2. Changes to Betts-Miller-Janjic convective scheme to improve 12 km parent QPF bias, especially during the cool season
  3. Updated Ferrier-Aligo microphysics to improve stratiform precipitation, better anvil reflectivity, reduce areas of light/noisy reflectivity over oceans
  4. Improved effect of frozen soil on transpiration and soil evaporation, leads to reduced cold/moist bias during cool season
  5. Radiation/microphysics changes to reduce incoming surface shortwave radiation; reduced warm-season 2-m temperature bias
I know DTC lack of funds to further develop the source code for the community, but I still hope they could provide us an updated source code.

Here are some problems I faced during the past months, by running the last available NEMS-NMMB model provided by DTC:
+ too high diurnal T2m on summer days
+ model restart capability
+ occasionally ghost low-level clouds near the shoreline that affect the prognostic cloud microphysics schemes
Did you encountered these or other problems?


Re: NEMS NMMB topic

Posted: Fri Mar 10, 2017 2:48 pm
by meteoadriatic
I gave up. I had model crashing that I could not fix and then I replaced all projects with ARW.

I will try again only if new public release comes out. I don't think NAM code will be easy to compile on linux.

Re: NEMS NMMB topic

Posted: Fri Mar 10, 2017 6:28 pm
by oib
Ivan, I was lucky.
By now I was able to fix ALL the errors I have seen when running NEMS.x simply reducing the time step.
At we worked hard on NMMB and have promoted NMMB as our operational forecast model.
Hope to see some news about NEMS... sorry you gave up and put everything on WRF... now feel like I'm alone in the universe! :?

Re: NEMS NMMB topic

Posted: Mon Mar 13, 2017 12:26 pm
by Ivana
I also gave up, for now :) Maybe in a future I'll come back and try again.
Good luck. And I hope to hear some good news from you or from someone else who is using NMMB as operational weather forecast model.

Re: NEMS NMMB topic

Posted: Tue Mar 14, 2017 9:42 am
by oib
Ivana wrote:
Mon Mar 13, 2017 12:26 pm
I also gave up, for now :) Maybe in a future I'll come back and try again.
Good luck. And I hope to hear some good news from you or from someone else who is using NMMB as operational weather forecast model.
Thanks Ivana,
I hope source code update will be released by DTC.
If not, I will wait till NAM version 4 source code will be published to the NCEP source code repository, then I will try to adapt to Linux some of the fixes/updates announced in the newer version.

Re: NEMS NMMB topic

Posted: Fri Mar 24, 2017 5:01 pm
by oib
Starting from March 22 the new NEMS-NAM v4.0.0 source code is available and public on NCEP site.

I have successfully build it on Linux.

Building on Linux is quite trivial if you successfully did it with the DTC package. In fact it takes exactly the same steps you already did when you built the source code distributed by DTC.

All the requested libraries are the same of those you already compiled for the DTC version.

The pre-processor NEMS-NPS has not received any upgrade, so you can still use the one from the DTC version.

Starting from 12Z run of today, March 24, all forecast products on will come from the v.4.0.0.

Hope to see any forecast accuracy enanchment :geek:

By now I can see that the new version increases integration time of about 9%... :roll:

Re: NEMS NMMB topic

Posted: Fri Mar 24, 2017 5:08 pm
by meteoadriatic
Great! Let us know how it is going :)

Re: NEMS NMMB topic

Posted: Sat Mar 25, 2017 12:10 pm
by meteoadriatic
Are there some new configuration entries like new physics?

Re: NEMS NMMB topic

Posted: Sat Mar 25, 2017 2:28 pm
by oib
meteoadriatic wrote:
Sat Mar 25, 2017 12:10 pm
Are there some new configuration entries like new physics?
No new physic scheme.
I don't know for sure if there are new configuration options, maybe one or two... but maybe they are not interesting.

After a superficial source code comparison between DTC vs NAM versions, it seems to me that most of the changes have affected the pattern of microphysics. But need further investigation...

- New things in Ferrier-Aligo MP -

-- Further modifications starting on 28 July 2015
! (15) Added calculations for radar reflectivity and number concentrations of
! rain (Nrain) and precipitating ice (Nsnow).
! (16) Removed double counting of air resistance term for riming onto ice (PIACW)
! (17) The maximum rime factor (RFmx) is now a function of MASSI(INDEXS), accounting
! for the increase in unrimed ice particle densities as values of INDEXS
! decrease from the maximum upper limit of 1000 microns to the lower limit of
! 50 microns, coinciding with the assumed size of cloud ice; see lines 1128-1134.
! (18) A new closure is used for updating the rime factor, which is described in
! detail near lines 1643-1682. The revised code is near lines 1683-1718.
! (19) Restructured the two-pass algorithm to be more robust, removed the HAIL
! & LARGE_RF logical variables so that NLICE>NLImax can occur.
! (20) Increased nsimax (see !aug27 below)
! (21) Modified the rain sedimentation (see two !aug27 blocks below)
! (22) NInuclei is the lower of Fletcher (1962), Cooper (1986), or NSImax.
! (23) NLImax is no longer used or enforced. Instead, INDEXS=MDImax when RF>20,
! else INDEXS is a function of temperature. Look for !sep10 comment.
! (24) An override was inserted for (18), such that the rime density is not diluted
! diluted when RF>20. Look for !sep10 comment.
! (25) Radar reflectivity calculations were changes to reduce radar bright bands,
! limit enhanced, mixed-phase reflectivity to RF>=20. Look for !sep10 comments.
! (26) NLICE is not to exceed NSI_max (250 L^-1) when RF<20. Look for !sep16 comments.
! Commented out! (28) Increase hail fall speeds using Thompson et al. (2008). Look for !sep22 comments.
! (29) Modify NLImax, INDEXS for RF>=20. Look for !sep22 comments.
! (30) Check on NSmICE, Vci based on whether FLIMASS<1. Look for !sep22a comments.
! Revised in (34)! (31) Introduced RFlag logical, which if =T enforces a lower limit of drop sizes not
! to go below INDEXRmin and N0r is adjusted. Look for !nov25 comments (corrections,
! refinements to sep25 & nov18 versions, includes an additional fix in nov25-fix).
! Also set INDEXRmin=500 rather than 250 microns.
!--- The following changes now refer to dates when those were made in 2016.
! (32) Convective (RF>=20, Ng~10 L^-1, RHOg~500 kg m^-3), transition (RF=10, Ng~25 L^-1,
! RHOg~300 kg m^-3), & stratiform (RF<2) profiles are blended based on RF. !mar08
! (33) Fixed bug in Biggs' freezing, put back in collisional drop freezing. !mar03
! (34) Changes in (31) are revised so that INDEXRmin at and below 0C level is
! based on a rain rate equal to the snowfall rate above the 0C level. !mar03
! (35) Increase radar reflectivity when RF>10 and RQSnew > 2.5 g m^-3. !mar12
! (36) !mar10 combines all elements of (32)-(35) together.
! (37) Bug fixes for the changes in (34) and the RFLAG variable !apr18
! (38) Revised Schumann-Ludlam limit. !apr18
! (39) Simplified PCOND (cloud cond/evap) calculation !apr21
! (40) Slight change in calculating RF. !apr22
! (41) Reduce RF values for calculating mean sizes of snow, graupel, sleet/hail !apr22a
! (42) Increase reflectivity from large, wet, high rime factor ice (graupel) by
! assuming |Kw|**2/|Ki|**2 = 0.224 (Smith, 1984, JCAM).
! (43) Major restructuring of code to allow N0r to vary from N0r0 !may11
! (44) More major restructuring of code to use fixed XLS, XLV, XLF !may12
! (45) Increased VEL_INC ~ VrimeF**2, put the enhanced graupel/hail fall speeds
! from Thompson into the code but only in limited circumstances, restructured
! and streamlined the INDEXS calculation, removed the upper limit for
! for the vapor mixing ratio is at water saturation when calculating ice
! deposition, and N0r is gradually increased for conditions supporting
! drizzle when rain contents decrease below 0.25 g/m**3. !may17
! (46) The may11 code changes that increase N0r0 when rain contents exceed 1 g m^-3
! have been removed, limit the number of iterations calculating final rain
! parameters, remove the revised N0r calculation for reflectivity. All of
! the changes following those made in the may10 code. !may20
! (47) Reduce the assumed # concentration of hail/sleet when RF>10 from 5 L^-1 to
! 1 L^-1, and also reduce it for graupel when RF>5 from 10 L^-1 to 5 L^-1.
! This is being done to try and make greater use of the Thompson graupel/hail
! fallspeeds by having INDEXS==MDImax.
! (48) Increased NCW from 200e6 to 300e6 for a more delayed onset of drizzle,
! simplified drizzle algorithm to reduce/eliminate N0r bulls eyes and to allow
! for supercooled drizzle, and set limits for 8.e6 <= N0r (m^-4) <= 1.e9 !may31
! (49) Further restructuring of code to better define STRAT, DRZL logicals,
! add these rain flags to mprates arrays !jun01
! (50) Increase in reflectivity due to wet ice was commented out.
! (51) Fixed minor bug to update INDEXR2 in the "rain_pass: do" loop. !jun13
! (52) Final changes to Nsnow for boosting reflectivities from ice for
! mass contents exceeding 5 g m^-3. !jun16
! (53) Cosmetic changes only that do not affect the calculations. Removed old, unused
! diagnostic arrays. Updated comments.

Re: NEMS NMMB topic

Posted: Sat Mar 25, 2017 3:43 pm
by meteoadriatic
Interesting. What about those missing fields from Thompson MP, did you tried it? Like radar reflectivity, cape, cin, lightning, those were not available with Thompson before. I know I ask too much probably and could just try myself but I'm pretty sceptical at this point and don't want to waste my time trying again and possibly giving up again after few months of testing :/