Welcome to the Cumulus Support forum.

Latest Cumulus MX release v3.2.0 build 3056 - 19 November 2019.
Legacy Cumulus 1 release v1.9.4 (build 1099) - 28 November 2014

See the Wiki for downloading either version.

Fire Weather, a new approach

Talk about anything that doesn't fit elsewhere - PLEASE don't put Cumulus queries in here!
Phil23
Posts: 307
Joined: Sat 16 Jul 2016 11:59 pm
Weather Station: Davis Vantage Pro2 Plus
Operating System: Win7 Pro
Location: Australia

Re: Fire Weather, a new approach

Post by Phil23 » Fri 13 Sep 2019 10:01 pm

HansR wrote:
Fri 13 Sep 2019 5:58 am
Presume I'll need the SS from the Linux version;
What do you mean by that?
I mean the Stylesheet file created by the Linux Version; but thinking from memory it may have related to the top 10.
So probably irrelevant.
Personally I run it from CumulusMX from the external program section, daily frequency.....
That's where I run my first daily script from.

Basically all it does is merge my old dayfile (2011-2017) with the current one & then FTP transfer the merged dayfile to weather.inverellit.com/historic/dayfile.txt so I have a bigger data set at http://weather.inverellit.com/historic/datasummary.php

2011-2017 Data is from my old Fine Offset & Cumulus 1
2017-2019 Data is from my Davis VP/2 & MX.

I just need to add a line to run you exe & update my FTP input file.
(Using DOS's FTP to upload the 2nd dayfile as the merged one needs to be updated AFTER the end day roll over.


Cheers.
Image

Phil23
Posts: 307
Joined: Sat 16 Jul 2016 11:59 pm
Weather Station: Davis Vantage Pro2 Plus
Operating System: Win7 Pro
Location: Australia

Re: Fire Weather, a new approach

Post by Phil23 » Fri 13 Sep 2019 11:14 pm

Very slight changes & contains latest data (1 day),

http://weather.inverellit.com/fwi.php

& added a <tr> & </tr> for the beetle... :D
Image

User avatar
HansR
Posts: 374
Joined: Sat 20 Oct 2012 6:53 am
Weather Station: Davis Vantage Pro 2+
Operating System: Raspbian GNU/Linux 9 (stretch)
Location: Wagenborgen (NL)
Contact:

Re: Fire Weather, a new approach

Post by HansR » Sat 14 Sep 2019 4:44 am

Phil23 wrote:
Fri 13 Sep 2019 10:01 pm
I mean the Stylesheet file created by the Linux Version; but thinking from memory it may have related to the top 10.
So probably irrelevant.
I do not use Style Sheets.
Phil23 wrote:
Fri 13 Sep 2019 11:14 pm
& added a <tr> & </tr> for the beetle...
I still have no idea what the beetle is actually referring to. I did fix some HTML error but I cannot find the omission of <tr> he is referring to. Did you have such a problem in your site and is he referring to that, or is the problem in my tables?

For the rest I think you have it running well.
Please be aware, that my current published release here at the forum apparently has some problem with the numbers. Probably better not use it. It must be a build problem, I am looking into it.
https://meteo-wagenborgen.nl / Cumulus (3.1.2-b3055)
Davis Vantage Pro 2+ / Raspberry Pi 3 Model B Plus Rev 1.3
Raspbian GNU/Linux 10 (buster) / Linux 4.19.75-v7+ armv7l
Mono JIT compiler version 6.4.0.198 (tarball Tue Sep 24 01:45:47 UTC 2019)

User avatar
HansR
Posts: 374
Joined: Sat 20 Oct 2012 6:53 am
Weather Station: Davis Vantage Pro 2+
Operating System: Raspbian GNU/Linux 9 (stretch)
Location: Wagenborgen (NL)
Contact:

Re: Fire Weather, a new approach

Post by HansR » Sat 14 Sep 2019 8:21 am

Phil23 wrote:
Fri 13 Sep 2019 9:47 pm
Just not sure what's going on here with the bars etc.
This is totally as I intended it to be. The bar above shows the colouring relative to each other: green has 30%, the red has 10%, all from the max value (which is set to 700 now). These represent the warning levels as interpreted from all data I have analysed so far. The values can go up even to 1300, but anything above 650 must be seen as a red situation.

The second bar shows the actual situation and the little arrow (>) in the sliding grey bar embedded in the colours show where you stand on the scale. It is implemented as pure HTML after trying the <meter> and <progress> tags which did not give enough liberty.

Phil23 wrote:
Fri 13 Sep 2019 9:47 pm
Yes I've taken not of your comments on that. It's inherited from the original distributions, which don't have it either.
So I assume beteljuice is talking about the Cumulus web distribution and not my tables.
Phil23 wrote:
Fri 13 Sep 2019 9:47 pm
Plan to make it look like my other pages & also change the credits.
Hans can suggest the most appropriate place in his Blog to point the credit link to.
In the new version, I made a text which shows with the pwsFWI. If you disagree, would like it smaller or placed otherwise let me know.
So far you have it running perfectly. I am happy to see the software on the other side of the world in an area for which it was created. Curious how it will behave over the season. Thanks!
https://meteo-wagenborgen.nl / Cumulus (3.1.2-b3055)
Davis Vantage Pro 2+ / Raspberry Pi 3 Model B Plus Rev 1.3
Raspbian GNU/Linux 10 (buster) / Linux 4.19.75-v7+ armv7l
Mono JIT compiler version 6.4.0.198 (tarball Tue Sep 24 01:45:47 UTC 2019)

Phil23
Posts: 307
Joined: Sat 16 Jul 2016 11:59 pm
Weather Station: Davis Vantage Pro2 Plus
Operating System: Win7 Pro
Location: Australia

Re: Fire Weather, a new approach

Post by Phil23 » Sat 14 Sep 2019 9:39 am

Hi Hans,

Can you define the range for each colour?

For example today FWI showed High risk & local weather showed Very high.

But 1 or 2 points on your scale could have moved it into the next zone, if you get what I mean.

https://www.weatherzone.com.au/nsw/nort ... s/inverell
Image

User avatar
HansR
Posts: 374
Joined: Sat 20 Oct 2012 6:53 am
Weather Station: Davis Vantage Pro 2+
Operating System: Raspbian GNU/Linux 9 (stretch)
Location: Wagenborgen (NL)
Contact:

Re: Fire Weather, a new approach

Post by HansR » Sat 14 Sep 2019 10:29 am

Hi Phil,

No, you can not define the danger level colours and I will not implement that. Those colours are actually more or less standardized. The FWI (used by Australia, Canada, France and Europe as a whole) use the same colours, sometimes expanded with a purple zone above the red with a name Very Extreme or Catastrophic or something. I notice from your link that The Weatherzone moved to yellow with the qualification Very High.

Very High is not the usual qualification for yellow if colour indicators are used and if we are using a five step indicator (apparently the McArthur index thinks differently about this). I looked on the weatherzone site for an explanation of the index but I can not find it. I assume, they use the Canadian FWI (which has been adopted by Australia as well as far as I know) or the McArthur FFDI. Both indices are very complex (as I tried to describe on my blog for the FWI). A complexity which triggered me to develop a simpler index, which can be used on a PWS.

From what I see and the qualification used, I assume the Weatherzone uses the McArthur FFDI which is a six-step index with a final step Catastrophic.

A FWI is not about adjusting colours and limits to other indices and that I will not do, it is about a calculation and fitting the concluding number to a danger level. That is what I am currently doing: functional behavioural testing. So I will not try to fit an existing index on a day to day basis (although pwsFWI must have similar results), but I will think about the limits I use (for crossing danger levels) and I will think (am already thinking) of adding a Catastrophic level. Simply because I already noticed that in your area (NSW) the values over the season may go sky high, beyond what can be seen elsewhere. But I need to think about the limit for that.

My interpretation of my pwsFWI is described in this blog. Note that the current version uses a max-value of 700 and a five step index. If there is reason, I will change this. But not from day to day. Hope this answer satisfies you.

Btw: your pwsFWI is now in yellow as well, it's only the words which are different!

[edit:] Apparently Australia has adopted a new FWI in 2017. Still under investigation I assume.

[edit 2:] I now see that you did not ask for changing the colours but the (value)ranges. I could delete all text above :) but I'll let it be. And Yes, the goal of our exercise is to get good ranges (and maybe make a catastrophic level). We'll look and evaluate.

Best,
Last edited by HansR on Sat 14 Sep 2019 3:55 pm, edited 2 times in total.
https://meteo-wagenborgen.nl / Cumulus (3.1.2-b3055)
Davis Vantage Pro 2+ / Raspberry Pi 3 Model B Plus Rev 1.3
Raspbian GNU/Linux 10 (buster) / Linux 4.19.75-v7+ armv7l
Mono JIT compiler version 6.4.0.198 (tarball Tue Sep 24 01:45:47 UTC 2019)

User avatar
beteljuice
Posts: 2774
Joined: Tue 09 Dec 2008 1:37 pm
Weather Station: None !
Operating System: ex XP SP3 - now W10 laptop grr
Location: Dudley, West Midlands, UK

Re: Fire Weather, a new approach

Post by beteljuice » Sat 14 Sep 2019 3:48 pm

@Hans ...
I still have no idea what the beetle is actually referring to. ....
the beteljuice wrote:.... although (navigation) you still insist on ...
Nothing you've done Hans ... Phils approach to html is to let the browser work it out :groan:
Image
......................Imagine, what you will KNOW tomorrow !

Phil23
Posts: 307
Joined: Sat 16 Jul 2016 11:59 pm
Weather Station: Davis Vantage Pro2 Plus
Operating System: Win7 Pro
Location: Australia

Re: Fire Weather, a new approach

Post by Phil23 » Sat 14 Sep 2019 8:29 pm

Phil23 wrote:
Sat 14 Sep 2019 9:39 am
Can you define the range for each colour?

For example today FWI showed High risk & local weather showed Very high.

But 1 or 2 points on your scale could have moved it into the next zone, if you get what I mean.
Sorry Hans,

Maybe my question was lost in translation.

Was asking this to explain better.

Green = aa to bb
Blue = bb to cc
Yellow = dd to dd

and so on.

What are the numbers for each range?


Phil.
Image

User avatar
HansR
Posts: 374
Joined: Sat 20 Oct 2012 6:53 am
Weather Station: Davis Vantage Pro 2+
Operating System: Raspbian GNU/Linux 9 (stretch)
Location: Wagenborgen (NL)
Contact:

Re: Fire Weather, a new approach

Post by HansR » Sun 15 Sep 2019 6:14 am

Phil23 wrote:
Sat 14 Sep 2019 8:29 pm
Maybe my question was lost in translation.

Was asking this to explain better.

Green = aa to bb
Blue = bb to cc
Yellow = dd to dd

and so on.

What are the numbers for each range?
Yes, I understood later what you meant (see last remark on previous answer) so indeed something lost in translation. I have to be careful about this when talking to native English speakers. Will work on that :)

Currently the values are as follows:

green: 0% <= value <= 30% [that is absolute values: 0 - 210]
blue: 30% < value <= 50% [that is absolute values: 210 - 350]
yellow: 50% < value <= 70% [that is absolute values: 250 - 490]
orange: 70% < value <= 90% [that is absolute values: 490 - 630]
red: everything higher than value > 90% [that is absolute values: > 630]

Currently I am topping of the value at 700 (although in the table you will see the actual calculated values). And yes, these limits are under discussion, and yes, I am thinking of creating a Catastrophic class. I did not expect the weather in Australia to be so extreme :o

For the moment we will check the behaviour against your opinion and gut feeling and the "weatherzone index". Then after some months I will propose an adjustment. An adjustment not only means changing the above values but may also be a change to the algorithm. E.g. currently I am not happy about my implementation of the duration of the draught-period. But let's have a look at the behaviour for a period until the value goes really off the scale.

NB: I made a new version (051) of cumulusutils with some basic technical and textual changes and from now on you can find new versions here.
https://meteo-wagenborgen.nl / Cumulus (3.1.2-b3055)
Davis Vantage Pro 2+ / Raspberry Pi 3 Model B Plus Rev 1.3
Raspbian GNU/Linux 10 (buster) / Linux 4.19.75-v7+ armv7l
Mono JIT compiler version 6.4.0.198 (tarball Tue Sep 24 01:45:47 UTC 2019)

Phil23
Posts: 307
Joined: Sat 16 Jul 2016 11:59 pm
Weather Station: Davis Vantage Pro2 Plus
Operating System: Win7 Pro
Location: Australia

Re: Fire Weather, a new approach

Post by Phil23 » Sun 15 Sep 2019 10:17 am

Hi Hans,

I am not that familiar with your calcs, so would need another good read & digest of them.

But curious as to whether you have browsed the EVT figures in my data on beteljuice's dayfile page.
http://weather.inverellit.com/datasummary.php

And of course, these numbers need to be compared with the same months rainfall.

EVT for September is very extreme & relates to our dry conditions.

Phil.
Image

User avatar
HansR
Posts: 374
Joined: Sat 20 Oct 2012 6:53 am
Weather Station: Davis Vantage Pro 2+
Operating System: Raspbian GNU/Linux 9 (stretch)
Location: Wagenborgen (NL)
Contact:

Re: Fire Weather, a new approach

Post by HansR » Sun 15 Sep 2019 1:32 pm

Hi Phil,

Yes, I did browse those figures on http://weather.inverellit.com/datasummary.php. Nice page, isn't it ;)

With respect to evapotraspiration figures I am very reluctant to interpretation. In my blog I write:
The whole process of evaporation and drying is implicit. I will not try to quantify the water balance exactly. Evaporation is an extremely complex process, which is far beyond the scope of a fire weather index.
Any calculation on evapotranspiration must be looked at with caution. The number in your list must come from the Davis station but I do not know (and cannot find it) what calculation is behind it. I have been studying EVT an it leads to my quote above. Above that: if all water in an environment has evaporated, you can calculate what you want, but you can not get dryer than it is so at a certain moment EVT should become 0. I do agree that those figures for september are high and definitely relate to a high FWI. So maybe there is some thruth in EVT ;)

I use basic measurements in my calculations and I will not use a derivative like that evapotranspiration because it is a far too complex process and because it is unknown how it is calculated.
https://meteo-wagenborgen.nl / Cumulus (3.1.2-b3055)
Davis Vantage Pro 2+ / Raspberry Pi 3 Model B Plus Rev 1.3
Raspbian GNU/Linux 10 (buster) / Linux 4.19.75-v7+ armv7l
Mono JIT compiler version 6.4.0.198 (tarball Tue Sep 24 01:45:47 UTC 2019)

User avatar
HansR
Posts: 374
Joined: Sat 20 Oct 2012 6:53 am
Weather Station: Davis Vantage Pro 2+
Operating System: Raspbian GNU/Linux 9 (stretch)
Location: Wagenborgen (NL)
Contact:

Re: Fire Weather, a new approach

Post by HansR » Mon 16 Sep 2019 4:13 pm

List of sites:

1) De Wilgen - Groningen NL. My own weathersite, base of pwsFWI.
2) Phil's Backyard (Australia, very high FWI, high fire risk area from spring to autumn)
3) Pogoda, Niesiołowice (Poland, seasonal risk in summer)
4) Weatherstation 't Zandt (20 km north of my station)
5) Meteo Sangonera (10 km SW of Murcia in Spain)
6) San Sebastian - Arriola Meteo - San Sebastian in Spain, low risk area with 1300 mm rainfall/year;
7) Ségur-Le-Château - a small 70 km SE of Limoges. Low to medium risk area.
Last edited by HansR on Fri 04 Oct 2019 6:29 am, edited 2 times in total.
https://meteo-wagenborgen.nl / Cumulus (3.1.2-b3055)
Davis Vantage Pro 2+ / Raspberry Pi 3 Model B Plus Rev 1.3
Raspbian GNU/Linux 10 (buster) / Linux 4.19.75-v7+ armv7l
Mono JIT compiler version 6.4.0.198 (tarball Tue Sep 24 01:45:47 UTC 2019)

User avatar
HansR
Posts: 374
Joined: Sat 20 Oct 2012 6:53 am
Weather Station: Davis Vantage Pro 2+
Operating System: Raspbian GNU/Linux 9 (stretch)
Location: Wagenborgen (NL)
Contact:

Re: Fire Weather, a new approach

Post by HansR » Fri 20 Sep 2019 7:01 pm

Having seen the first results from pwsFWI and the stories of what is physically going on, I think we are on the good track.

One major change as per version 0.8.5 is the analysis mode for pwsFWI.

If in cumulusutils.ini the value of Analysis is set to a number (> 30) you will find that number of days in your HTML table. So if it is 365, you will find a full year of data on the pwsFWI page. If the number is larger than the number of records you have, all records are shown. If the number is smaller than 30, it will always show 30 records (minimum) to give pwsFWI the context you need. Also, the data is written to pwsFWIanalyse.csv for later use in Excel (or whatever tool).

Be careful, if you have 10 years of data (or more) and you use it all, processing time (also loading the webpage) becomes an issue.

If you use the analysis mode and have remark, I would appreciate feedback on your studies via the forum or via the mail!

Thanks!
https://meteo-wagenborgen.nl / Cumulus (3.1.2-b3055)
Davis Vantage Pro 2+ / Raspberry Pi 3 Model B Plus Rev 1.3
Raspbian GNU/Linux 10 (buster) / Linux 4.19.75-v7+ armv7l
Mono JIT compiler version 6.4.0.198 (tarball Tue Sep 24 01:45:47 UTC 2019)

User avatar
HansR
Posts: 374
Joined: Sat 20 Oct 2012 6:53 am
Weather Station: Davis Vantage Pro 2+
Operating System: Raspbian GNU/Linux 9 (stretch)
Location: Wagenborgen (NL)
Contact:

Re: Fire Weather, a new approach

Post by HansR » Sat 21 Sep 2019 5:41 am

meteosangonera wrote:
Fri 23 Aug 2019 5:55 am
Hi, I can test, or I can borrow my dayfile.txt
Hi, you offered to test my Fire Weather Index. If you are still interested, a distribution is now available here. I understand your meteorological attention is now elsewhere because of the torrential rains in your region, but still I would appreciate an installation thinking about the new season. Please, spread the word as well.

Best,
https://meteo-wagenborgen.nl / Cumulus (3.1.2-b3055)
Davis Vantage Pro 2+ / Raspberry Pi 3 Model B Plus Rev 1.3
Raspbian GNU/Linux 10 (buster) / Linux 4.19.75-v7+ armv7l
Mono JIT compiler version 6.4.0.198 (tarball Tue Sep 24 01:45:47 UTC 2019)

User avatar
HansR
Posts: 374
Joined: Sat 20 Oct 2012 6:53 am
Weather Station: Davis Vantage Pro 2+
Operating System: Raspbian GNU/Linux 9 (stretch)
Location: Wagenborgen (NL)
Contact:

Re: Fire Weather, a new approach

Post by HansR » Sat 21 Sep 2019 8:58 am

A blog dedicated to the theory behind pwsFWI.
More or less a specialist blog on Altitude correction on the August-Roche-Magnus equation. For the meteorologists on the forum.
https://meteo-wagenborgen.nl / Cumulus (3.1.2-b3055)
Davis Vantage Pro 2+ / Raspberry Pi 3 Model B Plus Rev 1.3
Raspbian GNU/Linux 10 (buster) / Linux 4.19.75-v7+ armv7l
Mono JIT compiler version 6.4.0.198 (tarball Tue Sep 24 01:45:47 UTC 2019)

Post Reply