UKTC Archive

Re: Statistical Risk vs Ecosystem Services - does this compute?

Subject: Re: Statistical Risk vs Ecosystem Services - does this compute?
From: Rod Leslie
Date: Dec 05 2019 18:59:01
Quite right. There is no way a proportion of unpredicted failures is the same as random - there's no question that inspections can identify many of the factors - and likelihood - involved in failure, but we know, not all of them. We can even predict some unpredictable failures - for example, the shedding of branches by big old Beech - as far as I know, noone can accurately predict thyat a particular limb is going to be shed in a particular timeframe  - but we know for certain there is a particular and recurrent risk - so just don't park the BMW under your big old Beech in a hot summer.


Rod Leslie

On 05/12/2019 18:11, Bill Anderson wrote:
No one will read this...?
You're wrong Julian, I've just read it, twice, and I see what you mean. I
think....
Bill.

On Thu, 5 Dec 2019 at 17:43, Julian Morris <jamorris@xxxxx.com> wrote:

No-one will read this, but for the record....

If we are to rely on the science (or possibly more correctly the
mathematics) of statistics, we must speak the language with consistent
meaning. We shouldn’t mix the lay and the mathematical meaning of words. So
I have reached for a couple of text books I have in the office. They
clarify that ‘random’ means equally likely to occur. In the context of tree
failure and the hypothesis that tree failures are random, this would be to
say that, regardless of size, position, condition, species, soil etc. every
and any tree is as likely to fail as the next.

We all know that we don’t have a perfect database of tree failures (and of
course tree non-failures) but I think it self-evident that the hypothesis
is not true. I am confident this would be backed up by a questionnaire of a
huge number of people that have a working knowledge, or better, of trees.
But the matter of statistical significance has been raised, and merits
consideration. Statistical significance is a measure of the likelihood that
a relationship between two or more variables is caused by something other
than chance. ‘Chance’ is of course synonymous with ‘randomness’ in this
context. An important part of the concept of statistical significance is
that it is not a black-or-white matter. Significance must always be stated
and calculated with reference to a ‘confidence level’, which is quantified
(indirectly but nevertheless based on) the deviation between sample results
and a known normal distribution of outcomes. Large sample sizes can help
significance levels, but it’s only as a matter of degree.

What statistics and a hypothesis hopes to achieve is demonstration that
there is a statistically significant (to a persuasive confidence level)
correlation between two variables. This in turn can support (but not prove)
a causative link.

So say one postulates that Meripilus causes failure of  mature beech at
the base. And we might use this as a specific theory that, if not refuted,
debunks the theory that all tree failures of all species are random. So we
might use statistics to record and show that in a group of mature beech,
some of which have outward signs of Meripilus infection and some of which
don’t, over a fixed period of study nearly all that have Meripilus failed
and nearly all that didn’t, didn’t fail. We thus have a normal distribution
from the non-infected trees and can show to some confidence level that the
infected trees deviate substantially in failure rates from the normal
distribution.  We thus have correlation between infection and failure, a
significant absence of ‘random’. Now, one could deny that there is a
causative link, but that’s a different problem…

One has to be very careful with the terminology. A lack of statistical
data to make up normal distributions and a lack of data for a tested
hypothesis might mean that there is currently no statistically significant
evidence, but that’s all it means, it doesn’t mean that failure is random.
Absence of data does not mean absence of potential to demonstrate
correlation and in statistical terms does not mean random.

Stepping out of the specialist realm of statistics, I’d say there’s a
reason why no-one has statistically analysed the correlation between
Meripius and failure, and that’s because it’s self-evident. Search all the
authoritative text on the subject of the modus operandi of Meripilus, and
you won’t find any that have samples, statistics, confidence levels and
statistical significances. They all just say Meripilus caused failure. And
I’ll warrant none say that the absence of Meripilus causes failure.

Conclusion? Absence of statistical evidence does not mean random.
Inability to separate out one variable from many to show a correlation with
another variable does not mean an absence of causation. Only statistical
evidence that shows no correlation means random.
And stepping even further back from statistics, any book and any witness
will tell any judge that Meripulus causes failure and that if you spot it
you need to have very good reasons not to foresee it causing harm or
damage, because otherwise you will be found negligent for not acting (on
seeing it) by reducing or eliminating the risk. In that sense, this is why
our profession is important and useful, and why statistics plays at best a
supporting role. The law imposes a duty of care, and the non-randomness of
tree failure IS the line that we call ‘reasonably foreeable’.

I have a well-thumbed copy of ‘Statistics’ by F Owen and R Jones here,
which I bought second-hand a couple of decades ago, and I have just noticed
that the previous owner has written in block capitals inside the back cover
“HELP I AM TRAPPED INSIDE A HUMAN BODY”.  It somehow seems apposite. The
law only requires us to be human, no more and no less.


Julian A. Morris - Professional Tree Services
jamtrees.co.uk  and  highhedgesscotland.com
0778 XXX XXXX - 0141 XXX XXXX


Sent: Thursday, December 05, 2019 at 12:51 PM
From: "Jim Quaife" <jq@xxxxxxxxxxxxxxxxx.co.uk>
To: "UK Tree Care" <uktc@xxxxxx.tree-care.info>
Subject: RE: RE: Statistical Risk vs Ecosystem Services - does this
compute?
That tree failures are random is a factor in the science of statistics,
but in our terms it is merely that we cannot accurately predict them, nor
do we have any idea how effective our surveys are in any quantifiable sense.
This does not mean that we shouldn't survey trees, because intuitively
it is beneficial (isn't it?).  For one thing, without tree surveys insurers
would walk away.
We see things that we regard as requiring action and there is no doubt
that in so doing we simply must have prevented some incidents.  But we have
no precise idea with anything other than a general feeling that we have.
Being aware of this does not discredit our profession (it is more than a
'trade'), in fact quite the opposite - in fact (without citing Rumsfeld!)
that we are aware that we don't know something is often as important as
knowing something.
The difficulty is that in this day and age there is increasingly no such
thing as an "accident" inasmuch as there is always a search for someone to
blame.
All we can do is to use our knowledge and experience as best we can, and
that qualifies as being conscientious and professional.  I don't see a
conflict - the alternative is naivety.
Jim

-----Original Message-----
From: uktc-request@xxxxxx.tree-care.info [mailto:
uktc-request@xxxxxx.tree-care.info] On Behalf Of Rupert Baker
Sent: 05 December 2019 11:44
To: UK Tree Care
Subject: RE: RE: Statistical Risk vs Ecosystem Services - does this
compute?
Agreed, Michael- and even that can be influenced by its environment - 1
atom of U235 by itself - decays randomly.... 1 atom in a block of U235 of
say 2.5Kg upwards - does not decay randomly, but in concert with its
fellows....
To argue that tree failures are random does our trade no favours; it is
possible to assign probabilities of failure to trees with a degree of
accuracy; even if no-one can say exactly when a specific tree or part may
fail; we can be fairly sure (to take a reductio ad absurdum) that a 5-year
old seedling is unlikely to blow over, whereas a large mature beech with
half its root system cut off during a housing development, say 5 years
before; and evidence of extensive activity of decay fungi as a result,
probably wont survive a good blow from an unsuitable direction.....
Mind you, even if wind events are not random, they are still not
accurately predictable; and in UK at least, it is the interaction between
wind and trees that causes most failures.
Atb
Rupert

-----Original Message-----
From: uktc-request@xxxxxx.tree-care.info On Behalf Of Michael Richardson
Sent: 04 December 2019 13:14
To: UK Tree Care <uktc@xxxxxx.tree-care.info>
Subject: Re: RE: Statistical Risk vs Ecosystem Services - does this
compute?
I believe that only truly random event in nature is the decay of a
radionucleotide atom.

Michael Richardson B.Sc.F., BCMA
Ontario MTCU Qualified Arborist
Richardson Tree Care
Richardsontreecare.ca
613-475-2877
800-769-9183

   <http://www.richardsontreecare.ca/images/Tree_Doc_logo_email.png>



On Wed, Dec 4, 2019 at 7:53 AM Jim Quaife <jq@xxxxxxxxxxxxxxxxx.co.uk>
wrote:
We just have to disagree Julian.
Jim

-----Original Message-----
From: uktc-request@xxxxxx.tree-care.info [mailto:
uktc-request@xxxxxx.tree-care.info] On Behalf Of Julian Morris
Sent: 04 December 2019 12:43
To: UK Tree Care
Subject: Re: RE: Statistical Risk vs Ecosystem Services - does this
compute?

Jim, this havse come up before and I explained at length why I
disagreed with your suggestion that tree failures are random. I still
disagree with you. The timing of individual failures can be a result
of complex things and may not always be foreseeable, but it's not
random. The whole idea of tree risk assessment is that the law expects
a reasonable person to act on foreseeable harm or damage, and so
people like us are employed to assess trees to see if all the complex
things come together and amount to 'foreseeable'. In those cases the
failure is demonstrably not random, and since it's part of a continuum
with an arbitrary acceptable/unacceptable line marked on it, no
unforeseeable tree falures are random either.
I'd give you "The problem with tree incidents is that they are
complex, such that objective quantification and prediction is often so
imprecise that they appear random."


Julian A. Morris - Professional Tree Services jamtrees.co.uk  and
highhedgesscotland.com
0778 XXX XXXX - 0141 XXX XXXX


Sent: Wednesday, December 04, 2019 at 12:07 PM
From: "Jim Quaife" <jq@xxxxxxxxxxxxxxxxx.co.uk>
To: "UK Tree Care" <uktc@xxxxxx.tree-care.info>
Subject: RE: Statistical Risk vs Ecosystem Services - does this
compute?
Hi Chris,
The problem with tree risk statistics is that in the science of
stats
they are not significant - I didn't know this but I have been educated
by someone who does.
I thought that statistical significance was to do with the sample
size,
but it is not. Significance is determined by the absence of randomness.
This is not always possible and so stats which contain randomness have
to be adjusted to compensate, and that usually means that the reliance
one can put upon them decreases proportionately.
The problem with tree incidents is that they are random.
We like think that our tree surveys are comprehensive and
professional
(which they are - hopefully) but accurate prediction of tree failures
is virtually unheard of.  We specify work that requires attention
where we can anticipate failure, but we have absolutely no idea
whether in so doing we have actually prevented an incident.
Intuitively we think we have of course and I do not question the
integrity of surveyors (myself included!), but there is no way we can
prove it.
Regretfully it follows that any calculations based on a random data
are
of questionable worth in actuality.   Interesting yes, but applicable?
We all love numbers as they provide reassurance (particularly to
insurers who hover over all this), but I would be very wary of basing
any sort of policy or programme on such calculations.
Although it may sound "woolly", tree risk assessments are
justifiable
because the alternative of not conducting them is not.
Jim

-----Original Message-----
From: uktc-request@xxxxxx.tree-care.info [mailto:
uktc-request@xxxxxx.tree-care.info] On Behalf Of Corder, Chris
Sent: 04 December 2019 11:43
To: UK Tree Care
Subject: Statistical Risk vs Ecosystem Services - does this compute?

Dear all,

I am a sad case. I only have numbers to comfort me.
So I was wondering where the risk-benefit of trees might lie.
Maybe someone has already done this?
If not, does the following compute?

For arguments sake...lets agree that the background of risk of death
in
'public spaces' in the UK seems to be in the order of 1/10,000,000
Lets also assume/agree that the Value of Statistical Life is £2M
6 deaths per year = value of risk of £12M per year “A value of
statistical life of £1,000,000 is just another way of saying
that a reduction in risk of death of 1/100,000 per year has a value of
£10 per year” (HSE, 1996)"
Therefore, divide £12M by 1/10,000,000 = risk value of £1.20 The
i-Tree Eco London study found that 8,421,000 trees provided
£132,700,000 per year of ecosystem services i.e. £15 per tree per year
(or near as damn it). So lets assume £15 per tree might be about right.
£15 eco value per tree/£1.20 risk value = 12.5 So...is it right to
say that the background risk from trees would need
to be 12.5 times greater before the ecosystem benefits start to become
outweighed?
If so, then presumably the background risk from trees could increase
to
somewhere in the region of 1/800,000 before the risk starts to
outweigh ecosystem benefits?
(Which is sort of where the Tolerable/Broadly Acceptable region of
the
ToR Framework lies...is this coincidence?)
Does this compute or have I gone start raving mad?

p.s. I get the daily digest...so I won't see any replies in real
time.
So thanks in advance. And sorry in advance for delay in reply.
All the best,
Chris

Christopher Corder PDArb (RFS), BSc (Hons) in arboriculture, MArborA
Assistant Arboricultural Manager Hampshire Highways
Tel: 0300 XXX XXXX
Web: www.hants.gov.uk/roads
@Hantshighways

© Hampshire County Council 2017 | Disclaimer | Privacy Statement

-----Original Message-----
From: uktc-request@xxxxxx.tree-care.info <
uktc-request@xxxxxx.tree-care.info> On Behalf Of Alastair Durkin
Sent: 03 December 2019 08:56
To: UK Tree Care <uktc@xxxxxx.tree-care.info>
Subject: RE: Replacement trees in an area TPO

Hi Jon the situation is this:

If you have an area TPO and allow a tree to be removed, subject to
replacement planting then you MUST either make a new TPO on the
replacement OR formally 'vary' the TPO to include the new individual
tree. Otherwise the tree is not protected.
The 'C' business is for giving effect to planning conditions under
s197,
it's not for TPO app conditions. See section 4 of the model order.
Hope this helps.

Alastair


-----Original Message-----
From: uktc-request@xxxxxx.tree-care.info <
uktc-request@xxxxxx.tree-care.info> On Behalf Of Jon Heuch
Sent: 01 December 2019 14:14
To: UK Tree Care <uktc@xxxxxx.tree-care.info>
Subject: Replacement trees in an area TPO

Good folk of uktc



Remind me what should happen if the removal of a protected tree
covered
by an area order is allowed, but a condition is given to plant a
replacement tree? The order may or may not contain separate individual
protected trees.


The order can't be altered to include an individual tree shown
within
the protected area, can it?


The tree officer who gave permission will remember but what is there
on
record to show a protected replacement tree? The replacement tree will
clearly be younger than the order, so not seemingly protected to
subsequent tree officers.


Do area orders get conditional replacement trees? Do they get Tree
Replacement Notices?


Is this just a failing of the Area order?



Jon








--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/


--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/


--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/



--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/




--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/



--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/



--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/



--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/



--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/




--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/







--
The UK Tree Care mailing list
To unsubscribe send mailto:uktc-unsubscribe@xxxxxx.tree-care.info

The UKTC is supported by Bosky Trees arboricultural consultancy
http://www.boskytrees.co.uk/