SIGMET errors

[Mikkel]

Hi Hardy and everyone.

Sometimes when using the SIGMET weather download option in PSX i get a few excessively large SIGMET's.
Mainly from Canadian FIR's tho not exclusively.

Here are those that at the time writing is giving me problems.

RCAA SIGMET 11 sould be a circle covering all of Taiwan. But covers a large part of the world.
The Canadian ones are covering from south Canada to Argentina.
MMEX SIGMET 1 is covering all of Mexico, should be a small circle off the coast in the Pacific Ocean

OBS: Just noted that if i put one of the SIGMET's in they show up correctly,
But if i start putting all/some of them in things start to go wrong?

----------------------
HAZARD: TC
WCCI31 RCTP 241849
RCAA SIGMET 11 VALID 241900/242200 RCTP-
RCAA TAIPEI FIR TC GAEMI PSN N2436 E12130 CB OBS AT 1800Z WI 190NM
OF TC CENTRE TOP ABV FL500 WKN FCST AT 2100Z TC CENTRE PSN N2451
E12100=
----------------------
HAZARD: TURB
WSCN05 CWAO 241845
CZUL SIGMET E4 VALID 241845/242245 CWEG-
CZUL MONTREAL FIR SEV TURB FCST WI 60NM WID LINE BTN N6205 W07355 - N6059 W07023
SFC/FL030 STNR NC=
----------------------
HAZARD: TS
WSCN02 CWAO 241939
CZEG SIGMET Q1 VALID 241935/242335 CWEG-
CZEG EDMONTON FIR FRQ TS OBS WI 30NM WID LINE BTN N5441 W11427 - N5549 W11216
TOP FL340 MOV NNE 15KT NC=
----------------------
HAZARD: TS
WSCN04 CWAO 242038
CZYZ SIGMET R1 VALID 242035/250035 CWUL-
CZYZ TORONTO FIR FRQ TS OBS WI 20NM WID LINE BTN N4346 W07920 - N4510 W07756 TOP
FL380 MOV ENE 25KT NC=
----------------------
HAZARD: TS
WSCN05 CWAO 242044
CZUL SIGMET N3 VALID 242040/250040 CWUL-
CZUL MONTREAL FIR FRQ TSGR OBS WI N4622 W07315 - N4602 W07310 - N4528 W07503 -
N4452 W07608 - N4558 W07546 - N4622 W07315 TOP FL450 MOV ENE 30KT NC=
----------------------
HAZARD: TC
WCMX31 MMMX 242113
MMEX SIGMET 1 VALID 242107/250307 MMMX-
MMFR MEXICO FIR TC BUD OBS
N1706W11212 AT 2100Z
FRQ TS FL530 WI 180NM OF CENTRE
MOV WNW AT 14KT .
FCST 250600Z N1736W11354=
----------------------

Best Regards Mikkel

[Hardy Heinlin]

Hi Mikkel,

these 3 work OK together:

HAZARD: TS
WSCN05 CWAO 242044
CZUL SIGMET N3 VALID 242040/250040 CWUL-
CZUL MONTREAL FIR FRQ TSGR OBS WI N4622 W07315 - N4602 W07310 - N4528 W07503 -
N4452 W07608 - N4558 W07546 - N4622 W07315 TOP FL450 MOV ENE 30KT NC=
----------------------
HAZARD: TC
WCMX31 MMMX 242113
MMEX SIGMET 1 VALID 242107/250307 MMMX-
MMFR MEXICO FIR TC BUD OBS
N1706W11212 AT 2100Z
FRQ TS FL530 WI 180NM OF CENTRE
MOV WNW AT 14KT .
FCST 250600Z N1736W11354=
----------------------
HAZARD: TC
WCCI31 RCTP 241849
RCAA SIGMET 11 VALID 241900/242200 RCTP-
RCAA TAIPEI FIR TC GAEMI PSN N2436 E12130 CB OBS AT 1800Z WI 190NM
OF TC CENTRE TOP ABV FL500 WKN FCST AT 2100Z TC CENTRE PSN N2451
E12100=


The following 3 don't work, neither together nor as singles:

HAZARD: TURB
WSCN05 CWAO 241845
CZUL SIGMET E4 VALID 241845/242245 CWEG-
CZUL MONTREAL FIR SEV TURB FCST WI 60NM WID LINE BTN N6205 W07355 - N6059 W07023
SFC/FL030 STNR NC=
----------------------
HAZARD: TS
WSCN02 CWAO 241939
CZEG SIGMET Q1 VALID 241935/242335 CWEG-
CZEG EDMONTON FIR FRQ TS OBS WI 30NM WID LINE BTN N5441 W11427 - N5549 W11216
TOP FL340 MOV NNE 15KT NC=
----------------------
HAZARD: TS
WSCN04 CWAO 242038
CZYZ SIGMET R1 VALID 242035/250035 CWUL-
CZYZ TORONTO FIR FRQ TS OBS WI 20NM WID LINE BTN N4346 W07920 - N4510 W07756 TOP
FL380 MOV ENE 25KT NC=


As far as I recall, it's because PSX doesn't really understand the phrase "NM WID LINE BTN". In such cases it just makes a very small circle or a very large rectangle in the area of the coords included. During the development of my SIGMET interpreter that phrase never occured so I wasn't able to test it in practice along with automatic typo correction and certain omissions etc. Did you just see this effect for the first time?


Regards,

|-|ardy

[simonijs]

Morning,

https://aerowinx.com/board/index.php/topic,7232.msg77782.html#msg77782

Regards,
Simon

[Mikkel]

Hi again.

No it's not the first. I have usually switched them off if i get big rectangles while flying in their area. I think Canada uses the "NM WID LINE BTN" often because they usually show up exactly like in post linked by Simon

BTW i am on Windows 10 X64 with Danish language if that helps.


Mikkel

[evaamo]

Thanks for reporting this. I've been hit by this issue lately. I even had a situ file with all cached weather files ready to be sent to Hardy. After reading this report, I checked that my cached SIGMET had several "WID LINE BTN". After deleting these manually, the problem went away.

Thanks
-E

[Bluestar]

This is a link from ICAO that explains WID LINE BTN.

https://www.icao.int/MID/MIDANPIRG/Documents/eDocuments/MID%20Doc%20009%20-%20MID%20Region%20SIGMET%20Guidance.pdf

[Hardy Heinlin]

Thanks. I have this documentation already. That's not the problem :-)

[Mikkel]

Hi Hardy, did you find what is causing the bugs with the SIGMETs?

Regards Mikkel

[Hardy Heinlin]

I'll work on it in September ...

[Jeroen Hoppenbrouwers]

His very first reply #1 already contains 99% of the answer. "Now just fix it," which will happen in September...

[Hardy Heinlin]

Gentlemen,

I've checked the issue. Unfortunately, I can't fix it within the scope of my time management. The SIGMET texts are just too chaotic; any fix may cause an unwanted side effect somewhere else. I would have to hire an engineering team for that stuff. I think the issue is too unimportant in relation to the work required. Maybe I'll retry next year or later ...


Sorry!

|-|ardy


P.S.: They are not bugs. I know what's going on. The trick with the circle instead of using a rectangle is an intentional simplification. The problem with the oversize is something else; it only occurs in certain combinations of typos or non-standard expressions in the text.

[Jeroen Hoppenbrouwers]

Can you hack it by, for example, disregarding any rectangle above a certain size?  Or are these the unintended side effects you mention, as there _are_ occasionally such large rectangles?

[Hardy Heinlin]

These large rectangles appear when multiple certain unusual SIGMETs occur in a row, i.e. when the parser finds any reasonable end point only in the next or ensuing SIGMET in the list. It's a very complex monster code with zillions of "if" statements going forward and backward, up and down. A cup of worms. I haven't touched my parser code here in the past 5 years; it would take months to restudy every existing safety link and to test new modifications.

[Jeroen Hoppenbrouwers]

Time to insert AI into the mix    :-D