Welcome, Guest

Author Topic: Kalassak runs WRF  (Read 33224 times)

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #30 on: March 19, 2018, 02:04:40 AM »
note to self: the dx and dy values in namelist.input are very, very important



(stronger jets = fixed dx & dy)

and a comparison with gfs at the same time



so it's pretty good



here's the track for marcus on the global grid



and also the vortex-following grid (smoothed by default)



and not smoothed



and finally... the moment you've all been waiting for...

« Last Edit: March 19, 2018, 04:16:00 AM by atomic7732 »

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #31 on: March 20, 2018, 03:46:09 AM »
today's report:

i decided to run a test to see if:
- my simple temporary sst fix works
- tc grid restarts weaken storms erroneously (since it will restart from the coarse grid every 24 hours in the official sim)
- we can observe some true pavalan cyclones
- multiple tc grids can be active at once

so i ran deland.py and reland.py to remove all land, replace the sst's with interpolated data (so that you couldn't have cold tibets and hot australias imposed on the ocean) and then add pavalan land over it, but i'm fairly certain that didn't work because in the run australia starts blowing up a lot of convection probably in preparation to make a hypercane :/

the sim broke about 28 hours in, and i restarted it assuming it was a timestep issue (it was)

based on the cyclone (marcus!pavala) in the sim's intensity, it seems to have lost only about 1 kt or less of intensity, so it's probably not that big of a deal

then the sim broke again at 47.5 hours from the original start time... and it threw an error i have never seen before, so i really have no idea what the problem is... the (potential) good news is that since the run was technically not intending what i wanted to do originally i plan on restarting the run and will hopefully never have to deal with that error ever again, but i figure that's probably not going to happen

the fourth point wasn't tested because we didn't get far enough for another cyclone to actually form but on the third point...

i hereby present to you...











the honorary first pavalan cyclone...













Severe Tropical Storm Apeana...











i decided to name it because it was so amazing to have simulated it and different enough from marcus!earth that i felt it was okay in this situation, being the first true cyclone simulated on pavala (also that land is the hac peninsula thing)

you can see in the animation several shockwaves passing through which are the result of the elevation changes, and this presumably also causes a lot of noisiness in the simulation as compared to the earth version (i don't have the graphics uploaded for that but trust me... it's really weird) so that's something else that needs to be fixed

you can also see apeana being sheared by the forming australian hypercane and weakening

in other news, since i ran this at the official resolution we have some pretty global imagery as well



tomorrow:
- fix deland.py so the sst's work
- see if another SIGABRT happens and if so, attempt to fix it
- email wrfhelp or sth to get tips on editing the wrfbdy_input to make another planet, especially with respect to sea surface temps and elevation related problems
« Last Edit: March 20, 2018, 03:54:05 AM by atomic7732 »

Darvince

  • *****
  • Posts: 1842
  • 差不多
Re: Kalassak runs WRF
« Reply #32 on: March 20, 2018, 09:39:56 AM »
holy crap for some reason i never spotted north america last night but here it's extremely obvious

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #33 on: March 21, 2018, 02:44:08 AM »
tomorrow:
- fix deland.py so the sst's work ✓
- see if another SIGABRT happens and if so, attempt to fix it ✓
- email wrfhelp or sth to get tips on editing the wrfbdy_input to make another planet, especially with respect to sea surface temps and elevation related problems

i got deland.py to fix the most problematic sst's so now we can officially unofficially run wrf

also figured out the sigabrt problem: the program was terminated aborted because it was writing out onto a full volume

didn't do the last thing yet cause eh

so here's the outgoing longwave radiation for this run, it's so pretty ^-^
also note that australia isn't making hypercanes and you can't really see earth land in the ocean (you can on land sometimes)
you can also see the clouds created along the shockwaves from changing elevation



here's the vortex-following domain for Tropical Cyclone Apeana
it's about to make landfall in ganuo (most likely), and at the end of the run it's 84 kts, 944 hPa



and its track



and here's the weird chaotic ripples or w/e, presumably from the elevation shockwaves, visible in these graphics for 250 hPa geopotential height & wind



compared with the smooth earth run at the same level



tomorrow i'd like to just continue this run for sexy imagery and also to get a new pavalan cyclone and run two vortex-following grids at the same time

maybe other things too idk
« Last Edit: March 21, 2018, 12:37:50 PM by atomic7732 »

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #34 on: March 23, 2018, 11:57:10 PM »
[11:45 PM] just ridin my moa: now
[11:45 PM] just ridin my moa: it does appear that wrfrst_d01 is basically wrfinput_d01 except with a bunch more fields
[11:45 PM] just ridin my moa: so this means to me
[11:46 PM] just ridin my moa: i might just be able to rename it and with a few tweaks have it work
[11:46 PM] just ridin my moa: it might just ignore useless variables
[11:46 PM] just ridin my moa: or else just delete them if it complains
[11:47 PM] just ridin my moa: the reason this might work as opposed to interpolating the output to create a restart file is the restart files have a lot more variable fields
[11:48 PM] just ridin my moa: so obviously figuring out how to make those is much much harder than just deleting them and tweaking a few things (mostly renaming variables and maybe changing some global attributes)
[11:49 PM] just ridin my moa: i can probably fool wrf into taking the TC domains that already exist restart files while generating new ones for new TCs by just telling it there's a file input on that domain

on doing a restart as a new run so that wrf will generate new domains for new TCs that i want to have vortex-following grids

because i tried continuing the current run with a new domain around a disturbance near 12S 145E (east of apeana) but wrf couldn't find the restart file for domain 3 (the new domain), which makes sense because it didn't exist, but to start the run normally you don't need any input data for the TC domain because it will do the interpolation/calculation/creation of the domain itself at the start... so i just have to fool wrf into thinking the restart data is completely new data and tell it to use pre-existing data for TCs that were active before the start

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #35 on: March 24, 2018, 11:37:48 PM »
i wrote a script that just renamed a few variables and it seems to have worked

now i need to work on automatic restarting every 24 hours with new domains... that's going to be fun

also i should upgrade my track data outputter and maybe even the graphics with fancy cyclonephase data

ran the sim for 24 hours to check if the south uqe disturbance was ready for a grid but didn't generate the apeana graphics

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #36 on: March 27, 2018, 01:57:45 AM »
apeana has become extratropical so we can have a full video!!!!

https://www.youtube.com/watch?v=hWTfygHz_oQ


atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #37 on: March 31, 2018, 03:12:35 AM »
living, breathing pavala





also there will likely be a south uqe cyclone tomorrow (i'll run it with a tc domain if possible), it got up to 33 kts (1 below tropical storm) just before the sim ended

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #38 on: April 01, 2018, 03:18:45 AM »
well, several new cyclones have managed to form and achieve wind speeds >=34 kt which has earned them the names Bikol and Dika

Tropical Storm Bikol







Tropical Storm Dika (on the Bikol grid... )





but you'll notice that they're not centered... and that's because the grid doesn't seem to properly update to follow the storm

which is weird because it will print out the correct ATCF location for Bikol, even after it leaves the grid (although idk it acts a little weird i think after a while the longitude stopped moving and sometimes the wind wouldn't update, presumably because the low actually leaves the grid eventually), but yet it won't move the grid properly (it will move the grid a bit)

i don't know why this would be the case, apeana was tracked at 500 hPa (default setting), but bikol and dika are too weak to be able to do that, so i moved it down to 900 hPa and it seems to track it but again... it doesn't really work

maybe there's some weird setting in the reset files that is saying the grid should be somewhere else (where apeana used to be) and that's messing it up or something? idk

edit:  corral_dist(max_dom)                = 8        ; how many coarse grid cells the moving nest is allowed to get
                                                  near the mother domain boundary

alternatively it might be that the grid starts inside the corral distance and that's messing with everything
« Last Edit: April 01, 2018, 03:34:00 AM by atomic7732 »

Austritistanian

  • *****
  • Posts: 121
  • Amateur shitposter
Re: Kalassak runs WRF
« Reply #39 on: April 01, 2018, 03:50:49 AM »
Absolutely loved this, keep it up

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #40 on: April 01, 2018, 11:30:09 PM »
Absolutely loved this, keep it up
thanks :)

since bikol and dika were such disasters i couldn't really get grids to follow them (i partly fixed the problem maybe but i don't know, the domains were acting weird around the dateline still) i don't have videos for them, but i do have some updated tracks (actually i don't have bikol's oops)



next sim i make i'm going to move the grid so the boundary isn't at 180E but is instead at 0E, where it'll be over land instead so we won't have these problems

but i do have a video of a new storm that now that i think about it probably isn't actually a storm but whatever... also it's in the north uqe, just south of bongatar

https://www.youtube.com/watch?v=YXLFf2B6u7Q



what this shows me is that maybe letting the sim generate coarse grid storms with winds up to near 35 kts and 900-300 hPa thicknesses of 935 dam might be a good threshold for giving them TC grids otherwise they are too weak and unconsolidated to really deal with (they caused me to start and restart a bunch of sims to check if they worked and all i got from them was... nothing really), so in the future i'll be more scrutinizing before i try to give a storm a grid

on the other hand, Bikol got up to 50 kts and Dika was a tropical storm on the coarse grid i believe

tomorrow:
- more running of the sim until we get a well-defined TC
- respond to wrfhelp about what i've found out so far (fool restarts, dateline problems)
« Last Edit: April 01, 2018, 11:37:42 PM by atomic7732 »

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #41 on: April 10, 2018, 03:03:36 AM »
oh you know why there might have been no decent TCs for an entire month (!)

i was running the sim on the conus physics... which i figured wouldn't be much of a problem considering marcus/apeana was simulated just fine

it's possible that the conus physics are not conducive to cyclogenesis, but will support an already existing cyclone relatively well (although apeana/marcus wasn't actually all that strong at initialization it might be "stronger"/structured differently than what wrf can generate

contrary to this point: i had hyper cyclogenesis when running wrf last time (old version earth sim) with an unknown set of physics which was probably default (and likely to be conus and not friendly to tropical weather) so i'm not entirely confident this is the case

but i will be re-running the sim, probably from the same starting point with tropical physics instead to see if:
- apeana strengthens more (it only got up to like 90 kts which is only a few kts stronger than the low res earth sim i did last year, disappointing for such higher resolution, also considering that irl it was a cat 5 and models generally had that down anyway, the physics might be the problem here)
- new cyclones form
- extratropical weather is handled differently

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #42 on: April 15, 2018, 01:05:33 AM »






well it looks different

it's also stronger than with the conus physics (98 kts vs 91 kts)

more later (like how the rest of the global weather is reacting and whether new TCs can form)

also idk how to shift the grid around 180 degrees

atomic7732

  • Global Moderator
  • *****
  • Posts: 3848
  • caught in the river turning blue
    • Paladin of Storms
Re: Kalassak runs WRF
« Reply #43 on: April 22, 2018, 11:12:58 PM »
well Apeana is extratropical now





but we got a new cyclone that wasn't a disaster! say hello to the edeto region TC PśÃ­lsḱete: