Author Topic: Atmosphere Engine (Read 5954 times)

atomic7732 · « **on:** July 13, 2017, 02:49:19 AM »

instead of spamming the coding thread when i'm working on it i'll use this thread

this is a project to simulate reasonable tropical cyclone track and intensity changes which will eventually be developed into an app and a video game for pc

here is the most recent season run (June-November) of the atmosphere engine (the part that will make the cyclones for forecasting or spawning or whatever depending on the game that is played)

currently only EPAC and ATL can be used

Darvince · « **Reply #1 on:** July 13, 2017, 03:51:51 AM »

no
Bret no
Bret no
Bret no

atomic7732 · « **Reply #2 on:** July 14, 2017, 01:53:14 AM »

- troughs are now tilted
- epac disturbances start at -70% formation instead of -30% so they don't form southeast of guatemala
- i fixed a problem with MPI where storms were actually decaying at a slower rate than they used to (which is why epac looked like 2015 literally every year)
- working on a new way to do shear... it seems to hold the atlantic back a lot, see this run with no shear:

[08:46]   Kalassak   epac gets like 3 more storms than it usually does
[08:46]   Kalassak   and atl gets like 20 more than it usually does
[08:47]   Kalassak   so yeah i think revamping shear would be a good idea
[08:48]   Kalassak   maybe attach shear to all the lows
[08:48]   Kalassak   and then spawn like blobs of shear in the tropics
[08:49]   Kalassak   i also have to figure out how to move epac north like 3 degrees

tuto99 · « **Reply #3 on:** July 14, 2017, 02:47:30 AM »

that's really cool kalassak
what if you do this for pavala

FiahOwl · « **Reply #4 on:** July 14, 2017, 09:05:10 AM »

This message is only viewable with Universe Sandbox Galaxy Edition. Access it and much more with promo-code '181708'.

atomic7732 · « **Reply #5 on:** July 14, 2017, 11:38:56 AM »

yes the plan is
- redo shear so that atl can be alive and make epac storms not all majors
- tweak it so epac formation region shifts north a bit
- make wpac/nio view and make sure program can switch back and forth easily with a simple option
- make a script to generate a pavalan coastline map and pavalan sst mask
- copy wpac/nio view to make an uqe/hac-ganuo region thing and tweak it

beyond that idk

atomic7732 · « **Reply #6 on:** July 17, 2017, 02:52:29 AM »

the shear update is basically complete
i'm just trying to determine if i should code in #epacyear and #atlyear such that in certain years (like 1992, 2005) where one basin is supposed to do well, shear will be significantly decreased so that conditions will be extremely favorable and the basin will approach greeks (and the other basin will be impeded a bit)

(red contours are shear)

https://pastebin.com/HyMUnR56 - some output data from the pictured simulation

format: name, wind, pressure/formation chance, gale radius, size factor, sea surface temperature, shear, rapid intensification chance/status, eyewall replacement cycle chance/status

atomic7732 · « **Reply #7 on:** July 18, 2017, 01:11:59 AM »

added epac and atl years and stuff so they don't hang near average all the time

and got rid of the contours on the last frame of the simulation so it will be nice and pretty and uncluttered

Darvince · « **Reply #8 on:** July 18, 2017, 07:17:02 AM »

That's actually much easier to look at because I couldn't really understand what was going on in the picture when there were two continuous functions laid on the same image.

FiahOwl · « **Reply #9 on:** July 18, 2017, 09:52:22 AM »

This message is only viewable with Universe Sandbox Galaxy Edition. Access it and much more with promo-code '181811'.

Darvince · « **Reply #10 on:** July 18, 2017, 09:54:56 AM »

yes, but the eyes cannot focus on two different fields at once.

FiahOwl · « **Reply #11 on:** July 18, 2017, 10:43:57 AM »

This message is only viewable with Universe Sandbox Galaxy Edition. Access it and much more with promo-code '181817'.

atomic7732 · « **Reply #12 on:** July 19, 2017, 01:58:31 AM »

slight improvements + i made it output some file i can parse easily to analyze peak intensity distributions and stuff

just a few more tweaks then i can work on wpac/nio

atomic7732 · « **Reply #13 on:** July 20, 2017, 02:57:55 AM »

updated the way landfalls work so now strong storms weaken really quick but weaker storms don't weaken super rapidly

also (i think) i fixed a bug where the program would crash infrequently when trying to plot a cyclone that had no track

have an interesting season in which both basins did well somehow (i didn't code it to do that, 2006 is a neutral year so the atmosphere engine did this all on its own)

atomic7732 · « **Reply #14 on:** July 21, 2017, 03:14:01 AM »

unfortunately despite explictly programming RI into the atmosphere engine, it suffers from a near single mode of maximum winds near the observed local minimum, with the exception of some cat 4's and a cat 5

i'm not sure i'm going to do anything about it, because i'm not sure i can

although i do wonder if maybe restricting RI to a certain intensity (maybe 50-75 or 50-100 kts) might result in a better distribution... cause what happens is is not that few storms reach cat 1, it's that a lot do, but like... half of them don't peak there as you would expect, they peak near cat 3 or higher cause of rapid intensification
so maybe i'll tweak RI to be powerful enough to force that, and then limit the intensities at which cyclones can undergo RI (maybe with probabilities

)

data from this run (which is the first run with the new shear which got to greeks in either basin)

Darvince · « **Reply #15 on:** July 22, 2017, 10:27:00 AM »

lmao the gulf coast in your latest one

atomic7732 · « **Reply #16 on:** July 24, 2017, 04:07:48 AM »

apparently i uncovered a bug when tweaking the RI formulae because my changes caused some weird things to happen in epac (storms making it to the west coast as storms and only weakening about 1-2 kts per 6 hours)

apparently when a storm would undergo an ERC it wouldn't reset the RI rate...

here's a run with that bug fixed:

i still have to do a few things, like...
sometimes a storm thinks it's under negative shear, i have no idea why, but it can't be good
get rid of gulf high i added a few versions ago, it doesn't really curve the epac storms, i need more nepac lows instead
add extratrop/subtrop transition/natures
#epacyears seem really ineffective other than for producing a bunch of major hurricanes, but not many more storms (in fact there tends to be less storms than in neutral or atlyears for some reason)
and continuing to tweak RI and ERC rates and occurences cause this is the new histogram:

i can hardly tell a difference from looking at the track map, but maybe you guys can? i'm amazed the histogram is so different like wth

maybe i should raise the base RI rate and lower the peak of the new strength multiplier?

blotz · « **Reply #17 on:** July 24, 2017, 04:10:39 PM »

use MatLa

atomic7732 · « **Reply #18 on:** August 08, 2017, 03:28:15 AM »

hi wpac (and nio)

although there is a surprising lack of cat 5's

atomic7732 · « **Reply #19 on:** December 04, 2017, 12:53:34 AM »

just a little showcase of some seasons

some wpacnios

and our most recent addition... swioaus!

atomic7732 · « **Reply #20 on:** February 04, 2018, 04:11:34 AM »

some updates:

i came up with an idea for phase transitions (tropical, subtropical, extratropical)
- phase transitions come in 3 different types
- currently tropical disturbances can become tropical cyclones by randomly generating numbers above a threshold that is decided based on the conditions of how conducive they are for cyclone formation, and if they generate a number above the threshold, their formation chance increases 10% per iteration, starting from -30% (that's the most complex "phase" transition)
- tropical cyclones can become extratropical cyclones just by meeting the correct criteria and when they do they count up a certain amount toward 1 from 0, when they hit 1 they become extratropical
- tropical disturbances can't do this yet, but i think they will be able to, because of the following part
- i plan to make extratropical cyclones have a subtropical and tropical chance, both with different criteria, and both concurrently calculated
- they will basically just be a certain directly calculated chance per turn to become subtropical or tropical, like, if waters are cool but not cold and the cyclone isn't moving much, it'll immediately become a subtropical cyclone
- if waters are really warm it'll immediately become a tropical disturbance and then follow normal procedure
- a subtropical cyclone can become tropical in much the same way that tropical cyclones become extratropical, there will be a counter from 0 to 1 and when certain criteria are met it counts up toward becoming tropical when it reaches 1
- obviously, tropical disturbances need to be able to become extratropical or else they wouldn't be able to become subtropical cyclones

also i know why the southern hemisphere was acting weird... i didn't fix the south motion penalty for forming cyclones, which is why there was so much spam, because north-moving disturbances were favored over south-moving ones

the southwest indian is so much better now, but now western australia forms everything so we'll have to figure that out

and

there's hope for spac now but... it needs some work

atomic7732 · « **Reply #21 on:** May 28, 2018, 03:12:19 AM »

new update

v0.13

i weakened the default environmental flow and made the beta effect more realistic. instead of just nudging the storm north at a rate of .2 deg/6 hr/50 kts i made it more scientifically accurate and made it .3 deg/6 hr north and west and dependent upon the size of the storm (the larger it is, the stronger the beta effect, intensity is not really important)

i think the tracks are much more interesting now, though there is still room for improvement. i probably need to adjust the high and low locations/intensities/frequencies