This is an Open Challange to create a decision tree from limited information as is usually available from a basic electronic weather station situated in UK.

Trying to correctly interpret both current conditions and likely forecast, together with any advisories.

Open to everyone at every level, decisions can be as complicated or a basic as you like as long as the reasoning is sound. You can do a small bit or try it all !

PEOPLE OF A NERVOUS DISPOSITION NEED NOT APPLY !
It should be expected that anyones contribution will be expanded / contracted / modified by anyone else as they 'fly' with an idea.


Sort of things wanted:
How much rain to cause local flooding ?
If it's winter AND low (but not freezing ?) temp AND forecast is rain - is that sleet ?
If its May AND evening AND temp <5 AND forecast clear - is that a frost warning for gardeners ?
If Summer AND wind from (direction) AND temp certain value - rain ?
and so on ....

Probably let this 'run' for 4 / 5 weeks, please limit yourselves to the data outlined below - if successful then it could be expanded to include trends / historic data / observational data.

Information available:
'change since last' for almost any event can be made available either as none, rising, falling OR as a numeric value - I am using this as 'eye candy' because the last compared value could be as short as 10 sec. or as great as one hour, dependent upon the ftp frequency set by the user and therefore is unreliable as a 'trend'

Date (could be used for seasonal decisions)
Time (night / day - morning / evening decisions ?)

Current Temperature
* change since last

"Feels Like" temperature (calculated)

Comfort Index (calculated - Verbose description - eg. Uncomfortably Cold)

Windchill

Wind speed (all common units - inc. verbose, eg. Smoke rises vertically)
Wind direction (degrees / SSW / South-Southwest)

Humidity
* change since last

Dew Point

Current Barometer (relative / absolute - hPa / inHg)
5(?) hour barometer trend (as decided by the wx station - eg. falling, steady, rising)
* change since last

Station 'forecast' (clear, rain, cloud) [wx settable as 2 / 3 / 4 hPa change, default 4]

Station 'Storm Warning' [wx settable 3 - 9 hPa, default 5]
(NB. rate of change is not stated, presumed to be per hour unless you know different)

Rain last 24 hours (rolling count every hour, on the hour)
Rain last 60 minutes (rolling count, real time)

Cloud Base Height (m / ft - calculated, Above Ground Level)

Be aware of possible pitfalls !
Eg. rain.
Rain 60m greater than 0, must be raining - FALSE (may have rained 59 minutes ago)
Rain 60m less than last check, must be dry - FALSE (may be raining less than 59 minutes ago)
Rain 60m equals 0, must be dry - LIKELY (may not be enough rain to work the bucket )
Rain 60m greater last check, must be raining - TRUE (but that could be 10s ago - not reliable as a 'trend' because next check will probably be 'no change' unless it's a monsoon)
Rain 60m greater than 0 and 'no change since last check' must be dry - FALSE (may have rained 59 minutes ago the same amount it's raining now)


Is has been rumoured that the beteljuice "is working on something", well the idea is that I will probably stitch the 'group logic' into HeavyWeather - AJAX to give nice 'English' forecasts and / or advisories (some already exist)

NB. This is a Work In Progress 'beta' and is not yet ready for distrubution ! - Nor is it guaranteed to be working when you see it !! (Big thank you to Scotweather for making her data available)

__________________

This one is going to take some thought.....

__________________

10mm of rain in 1 hour here in Falkirk usually causes localised flooding but thats JUST Falkirk.

__________________

Thanks Kate - if I can just trample all over that as an example

You said "causes" - so
10 mm last 60m "Localised Flooding"
What about eg.
8 mm "Localised Flooding Likely"
6 mm "Possibility of Localised Flooding"

What about sustained rain (rain24hr) ? and also some 'fuzzy' logic ?
eg.
rain24 > 24 mm AND forecast = rain "Possibility of Localised Flooding"
rain24 > 30 mm AND forecast NOT rain "Possibility of Localised Flooding"
... BUT IF forecast = rain OR rain24 > 36 mm "Localised Flooding Likely"
... BUT IF rain24 > 42 mm "Localised Flooding"

I've used a mix of logic description just to show that as long as it's sort of understandable you don't have to 'code' the logic

We don't have to worry about temperature because if it's ice there is no count and no flood - when it thaws we get a count !

hmm .. now there's a thought !
rain24 > 2mm AND temp <1 "Caution: Icy Roads" ???? discuss

__________________

Like the first bit BUT the ambiguous area of it could still be 6mm or 8mm with still 59mins to go in that last hour. [scratches ones head !! ]

As to the second part my brain is fuzzy never mind the fuzzy logic seems reasonable

__________________

You mean all the rain happened this last minute ? or in the minute just about to 'fall off' the end ?

Doesn't really matter, a flash flood would still need time to drain away (say 59 minutes ) so any warning would still be valid.

If rain 'to come' during the 'rolling' 60 mins increases / decreases / stops, then the advisory would update.

The 6 / 8 mm figures are for illustration purposes only.

__________________

You have started an enormous project with this bj.

I can see the appeal of doing this, generating creative logic.
But it will take some serious data mining to make even tiny parts realistic.

I am doing a data gathering operation to find optimum conditions for re-energising and storing NiMH cells.
Its tryng to find which factors are relevant.
Only then determining optimum values.
I have been at this for months !

Good Luck and Have an Interesting Time, but be realistic about how much you can achieve.
After all you are trying single-handedly to do what the Meteorolgical Industry is trying to achieve.

Nothing so foolish PB - for a start I'm asking everybody else !

Seriously, there is enough 'talent' in-house to contruct some sort of basic flow-chart to determine what is happening 'now' and 'most likely' next.

We aren't talking jetstream analysis and sea temps, we're talking pine cones and seaweed !

And if I can then use parts of the logic to give easy speak English or graphics - Great !!

PS. Thank you for your concern - the beteljuice was always a disappointment at school, was it ever thus !

__________________

hi all - sounds a fascinating project bj - and good luck with it - but anything that requires other than point and click is way over my head

can i just throw a little spanner in the works - and that is this - variables - and by this i mean local variables - that could seriously compromise the outcome of any analysis - and i will use Kate's flooding scenario as an example if i may

she said "
10mm of rain in 1 hour here in Falkirk usually causes localised flooding but that's JUST Falkirk."

now why is that ?? - topography - block paving - or the council not cleaning the drains out regularly ??

tell you a little story : many many years ago - i made a statement along the lines of " well if we get any flooding up here ( Dartmoor plateaux ) then the rest of the world is in BIG trouble " - an hour later we where called out to flooding of one of the highest pubs on Dartmoor - reason a culvert under the road had become blocked and the pub sat in a slight gully - result 2 foot of flood water in the bars ( oh and it was blowing a hoolie and try stitching down a way wood roof in that lot - BIG FUN !!

now normally if the gully had been clear - it would not have mattered how much runoff or rain came down on the moor in that area and indeed it was several years before we got called there again for flooding -( same reason )

so you can see from this example - that whilst it is possible to say if x happens -( heavy rain ) then y may result ( flooding possible ) - somewhere you have to factor in BUT only if z ( the council did not clean the drains ) AND w ( its autumn and the gullies may be blocked by leaves ) are TRUE

IF - z and /or w are FALSE - then y becomes less probable

well i hope i have that something like right ?? - just trying to understand the enormity of the task you have set yourself - in a local context

but i do like the pine cones and seaweed analogy - as i think some times some forecasts /forecasters tend to baffle with science - when all the average user wants to know is - will it rain - if it does what will happen - and can i blame local flooding on next doors new block paving ??

sorry for the ramble folks - just ignore me - most people do

Brett

__________________

No problem - That's great
You have w Autumn (leaves) AND x heavy rain

But I am only trying to make sense of 'Acts Of God' - councils are beyond any kind of reasoning

__________________