Original release topic

[HansR]

@all:
I just found out, that the repair of the XML did not work. Apparently the addition of a new station (Crafers) triggered a corruption of the XML write back. I will look into it and report back here. Until then consider the maps module as broken (existing maps remain functional of course).

[pernaczy]

@pernaczy: your maps has not reported in since 13 December! Could you check the running of the maps module for your station? Thanks.

Done. I just decided to tun the maps from time to time.
BTW. The "Crafers" station is on my map. (https://www.pogoda-niesiolowice.kaszuby ... isknew.php )
Piotr

[HansR]

@All: OK. Maps seems to be OK again. Hope it won't fail again, but unknown/not sure what caused it.
@pernaczy: OK, thnx.

[sjmcinness]

Hi Hans,
Thanks for work on writing this. I had been running FWIcalc on my previous PC up until a few days ago and had been meaning to add the output to my Cumulus v1 site but hadn't ever actually done it. With a bit of trial and error I have your version working, using the same base template as the other html files and have added the FWI link to each of the default pages.

Please add me to the map at your convenience; I'd encourage you to have a look at the current data, as well - particularly mid-late December - it's a tinder box!
http://www.users.on.net/~sjmcinness/weather/pwsfwi.php
Northgate Weather is probably the best name for it. I'll add the maps page at a later date.

Also, unless my maths is bad, with Analyse set to 90, it's only producing 80 days of output - likewise for 365 days, it's only producing 355 days (as mine is currently configured).

[HansR]

@sjmcinness:
Thanks for using pwsFWI in the middle of Adelaide! Though you have extreme values, I hope you stay away from the flames. Urban stations always are biased upward somewhat, but it definitely is an indication of the status. Great to have another Australian showing the extremes of what is going on!
NOTE: Interesting difference with Crafers, distance only 17 km.
NOTE2: a 1% RH is extreme. Almost un-liveable. (check your device?)

It would be nice if you indicate on the page (e.g. just above the legend) the fact that it is an urban station with some upward bias. Context is always good!

You ask me to put you on the map but the map is not my call, it is a user choice. See the README on how to use it. Run it manually or once a week or fortnight should be enough to remain up to date. It would be nice if you add the map to it (see eg. how pernaczy did it below the pwsFWI).

I will add you to the list I maintain.

Cheers, Hans

NOTE: indeed there is a difference in numbers with the analysis. It should differ 5 if I count well and probably came into existence when implementing the prediction. I may correct it sometime, but it does not really seem important to me. Unless you can convince me otherwise

[HansR]

@all:
I would like to make a note on FWIcalc, which @sjmcinness refers to. And I myself refer to it on my page (see bottom) as a source of inspiration. That is OK.

But please remember, that the indices are really very different. FWIcalc calculates the FWI as used in New Zealand/Australia/Canada and yes, also in Europe by the official instances. It is a very complex FWI (described here). Although moisture content attributes to the index, it differs in complexity and tools used - and therefore in cost - highly from what I do (described in these blogs). The science behind Vapour Pressure Deficiency (VPD) as an indicator of fuel aridity is tested and described here.

Anyway, just be conscience of the differences.

[HansR]

And while I am at it, I notice @sjmcinness has values of pwsFWI up to almost 1900 on 20 December 2019. As the catastrophic warning level starts at 800, 1900 seems a bit over the top. Although we have seen these values in Spain as well (during the analysis of last year by meteosangonera), they must be regarded as not so very useful. There are arguments to top of the value and I may do that sometime.

1. The equations seem to become unstable and meaningless with high VPD combined with high winds;
2. There is no such thing as a catastrophic catastrophe;
3. Instruments may become inaccurate/unstable in the extreme regions. Notably RH become inaccurate at high and low region;
4. I already mentioned the urban positioning of Chaffey Common, this will influence the value of pwsFWI upward, (temperature will be higher and RH will be lower).

Nevertheless, it is clear that a value of far beyond 800 may have an unnatural upward bias, there is very much a reason to use the warning level associated with it because without the bias, it would very likely still be extremely high.

I may start a blog sometime incorporating all my remarks on the functioning and use of pwsFWI. I probably will after the fire season in Europe has finished (so one year of functioning of pwsFWI). Meanwhile you'll have to do with this

[hills]

NOTE: Interesting difference with Crafers, distance only 17 km.

We're 17km apart horizontally but over 400m apart vertically,

I use youweather's prediction for Crafers which is usually about 3 or 4 degrees cooler than Adelaide's. that will account for some of the difference, but that said, they are quite extreme, given the values are not too far apart.

[hills]

Thanks Hans, my maps are working again as well after disabling and re-enabling them. Strange this is my maps worked initially, then broke when everyone else's did. I hope it wasn't something I did wrong, as the corruption was in my line.

[HansR]

@hills:
Concerning the maps, no I don't think it was anything you did but neither do I have a clue of what went wrong. Anyway, we'll see how it behaves and investigate before changing anything.

And indeed, huge difference between Crafers and Adelaide where altitude can't explain everything. Really interesting to see this. 100m altitude difference should give appr. a bit less than 1 degree so your 3-4 degrees sounds correct. But then again, the pwsFWI really explodes. Have to think about this.

[sjmcinness]

Thanks for using pwsFWI in the middle of Adelaide! Though you have extreme values, I hope you stay away from the flames. Urban stations always are biased upward somewhat, but it definitely is an indication of the status. Great to have another Australian showing the extremes of what is going on!
NOTE: Interesting difference with Crafers, distance only 17 km.
NOTE2: a 1% RH is extreme. Almost un-liveable. (check your device?)

The values are extreme but are pretty much what I expected. It gets hot here and when it's hot, it's almost always dry heat - so the <5% minimum RH is also not uncommon. Our mean yearly rainfall for the period 1929-2019 is 448.7mm.

I will run a comparison of my RH data against the data I can obtain from the Australian Bureau of Meteorology for the closest station (Parafield Airport, daily 09:00 and 15:00 readings) and see how it looks; I replaced the temperature/humidity sensor in early 2018 with the updated and more accurate Davis 7346.070 VP2 Digital Temperature Humidity Sensor (based on a Sensiron SHT31) used in the current VP2 ISS, which improves humidity accuracy to a flat ±2% from 0%-100%.

It would be nice if you indicate on the page (e.g. just above the legend) the fact that it is an urban station with some upward bias. Context is always good!

Done.

You ask me to put you on the map but the map is not my call, it is a user choice. See the README on how to use it. Run it manually or once a week or fortnight should be enough to remain up to date. It would be nice if you add the map to it (see eg. how pernaczy did it below the pwsFWI).

Ahh, I misunderstood; I've registered using MapsOn. I've generated the Maps page and have added it to the web server, along with Graphs and YADR, but haven't added them to the links on the bottom of the pages yet. The map kind of looks lonely on its own page, so I may end up moving it as pernaczy did.

NOTE: indeed there is a difference in numbers with the analysis. It should differ 5 if I count well and probably came into existence when implementing the prediction. I may correct it sometime, but it does not really seem important to me. Unless you can convince me otherwise

Fair enough. It's not terribly important and I can always adjust the number to compensate anyway I have just over six years worth of data - how much is a sensible amount of data to include in the FWI analysis?

And while I am at it, I notice @sjmcinness has values of pwsFWI up to almost 1900 on 20 December 2019. As the catastrophic warning level starts at 800, 1900 seems a bit over the top. Although we have seen these values in Spain as well (during the analysis of last year by meteosangonera), they must be regarded as not so very useful. There are arguments to top of the value and I may do that sometime.

1. The equations seem to become unstable and meaningless with high VPD combined with high winds;
2. There is no such thing as a catastrophic catastrophe;
3. Instruments may become inaccurate/unstable in the extreme regions. Notably RH become inaccurate at high and low region;
4. I already mentioned the urban positioning of Chaffey Common, this will influence the value of pwsFWI upward, (temperature will be higher and RH will be lower).

Nevertheless, it is clear that a value of far beyond 800 may have an unnatural upward bias, there is very much a reason to use the warning level associated with it because without the bias, it would very likely still be extremely high.

If you would like the data to feed directly into your model and play with it, please let me know how much and what format; I have 30-minute data from 2013-11-13 17:30 to 2017-01-01 16:30 and 5-minute data from 2017-01-01 16:45 to current.

We're 17km apart horizontally but over 400m apart vertically,

I use youweather's prediction for Crafers which is usually about 3 or 4 degrees cooler than Adelaide's. that will account for some of the difference, but that said, they are quite extreme, given the values are not too far apart.

I agree - the conditions are often quite different between Crafers (Mount Lofty station) and the stations on the Adelaide Plains. Adelaide Airport is another station which often has rather different conditions to the other stations on the Plains but that's because it's basically at the beach.

Finally, a couple of bugs to report:

1. On the FWI page, the column for RH is actually not showing the percentage (values from 0-100), it's showing the ratio (values from 0-1) - the values should be multiplied by 100, at least for display;
2. On the YADR page, the heading for humidity is "Daily Values Humidity (°C)" - this should obviously be "Daily Values Humidity (%)"

Thanks again,
Scott.

[HansR]

@sjmcinness:
Sorry I had a pretty long reply with quotes and all but apparently I did not submit.
I am not going to redo it, but the short version is:

• Yes apparently very low RHs exist in Australia. Apparently a good drought breeds big fires. Can't believe you live there (difficult to understand for someone from a sealevel swamp where RH hardly gets below 50% and is normally around 80% )
• Thnx for the context text above the table.
• I'll see the menu items appear.
• I don't use data from the monthly logs, I use dayfile.txt. If there are issues I may ask your dayfile.txt for testing.
• I think it's great to have two stations close together to stress the point of local differences, my own station has a neighbour (t'Zandt) also close by but there is hardly any difference there
• I'll fix the numbers and the other bugs. Currently testing phase. should be released soon.

Cheers, Hans

[sjmcinness]

Never mind, I understand

• Ahh, dry heat is good - it makes sweating worthwhile, instead of just making one wet. I much prefer warm over freezing; we rarely have a day under 5°C. Bushfires aren't good, though, and over 6 million hectares have been burnt in Australia since the current bushfire season started in August (lots more detail on Wikipedia) which, to give some perspective, is nearly 1.5 times the size of Netherlands.
• I've added the other pages to the link list at the bottom of each page now and I added the map to the bottom of the FWI page. There's probably some more formatting to do and some more consistency to fix but it will do for now (I should probably change to CumuluxMX as well, since it's being developed again; I stopped checking when Steve stopped developing and feature parity was still poor back then).
• Yes, I understand that you use data from dayfile - I was offering more
• I wonder whether using the data from dayfile is the best option for temperature and humidity - it does give you the highest temperature and the lowest humidity but it doesn't give you the humidity at the same point in time as the highest temperature - which FWIcalc did (well, it preferred to use midday but provided the option to update based on a higher temperature later in the day).
• Oh, the fact that Phil's and mine are relatively close but have vastly different conditions is a great thing - more points of data is usually a good thing and it can only make the model better
• Thanks for the prompt fixes I see that you're running the updated version and YADR looks good now but you have removed the % symbol from the RH column on the FWI page - if it's possible, would you please be able to add an option so that the RH values could be displayed as percentages? I find RH easier to read that way and then it makes it consistent with RH on all of the other pages.

Cheers,
Scott.

[HansR]

Version 2.0.3 is on-line now.
No functional changes, just some fixes.

@sjmcinness / Scott : I'll answer your questions below.

[HansR]

Ahh, dry heat is good - it makes sweating worthwhile, instead of just making one wet. I much prefer warm over freezing; we rarely have a day under 5°C.

It's a different world. Looking at the graphs and tabulated daily data from the three Australian stations, I can tell you this half year of CumulusUtils and looking at all Utils stations daily has taught me a lot.

Bushfires aren't good, though, and over 6 million hectares have been burnt in Australia since the current bushfire season started in August (lots more detail on Wikipedia) which, to give some perspective, is nearly 1.5 times the size of Netherlands.

I know. You're all over the news and discussion with Phil (NSW) in September made me decide to speed up development. You may read my blog on this issue.

I've added the other pages to the link list at the bottom of each page now and I added the map to the bottom of the FWI page. There's probably some more formatting to do and some more consistency to fix but it will do for now (I should probably change to CumuluxMX as well, since it's being developed again; I stopped checking when Steve stopped developing and feature parity was still poor back then).

OK, looks great. Saw that beteljuice gave you some advice above. You may want to follow his advice as he's responsible for the beteljuice-interface you're using If necessary I'll make some changes.

I wonder whether using the data from dayfile is the best option for temperature and humidity - it does give you the highest temperature and the lowest humidity but it doesn't give you the humidity at the same point in time as the highest temperature - which FWIcalc did (well, it preferred to use midday but provided the option to update based on a higher temperature later in the day).

Ah, well, the timing on the day is more or less irrelevant. I had several discussions on this with Phil (NSW) and I regard the process as seen from the forest. Moisture related processes in forests are not very fast, neither in absorbing moisture as in releasing it. So an hourly estimation of FWI is not really necessary and neither is it important to take the highest temperature point specifically. It's a different approach from FWIcalc.

Oh, the fact that Phil's and mine are relatively close but have vastly different conditions is a great thing - more points of data is usually a good thing and it can only make the model better

It is. It actually is the most stunning lesson of my CumulusUtils exercise (I run daily through all stations participating if I can, weather is stunning)

Thanks for the prompt fixes I see that you're running the updated version and YADR looks good now but you have removed the % symbol from the RH column on the FWI page - if it's possible, would you please be able to add an option so that the RH values could be displayed as percentages? I find RH easier to read that way and then it makes it consistent with RH on all of the other pages.

You're welcome. Thanks for the observations!
With respect to the RH (and other items on the pwsFWI page): I will keep it in fractions, while on all other pages RH will be in percentages. The reason is that the pwsFWI values are those actually entered in the equations. So even if you would be on imperial system, you would see the values in degree Celsius while on the YADR page you would see Fahrenheit. That's how it is, it's for analysis and information. The same holds for the resulting csv-file.

Cheers,
Hans