Acer ventura

I'd forgotten about this and in effort to keep the thread on topic have replied here. http://arbtalk.co.uk/forum/general-chat/72575-quantified-tree-risk-assessment-qtra-version-5-questions-answers-2.html#post1115583 Cheers Acer ventura

A bit of tidying up from here... http://arbtalk.co.uk/forum/general-chat/73829-isa-best-management-practices-tree-risk-assessment-traq-5.html#post1113628 ...which I had forgotten about. Eh? This makes no sense. You need the Monte Carlo simulations to generate probability distributions before you can define the standard deviations for it. The Palisade website has an easy to grasp description of what Monte Carlo simulations do; Monte Carlo Simulation: What Is It and How Does It Work? - Palisade I have no idea what you mean by “risk is preceived on a linear scale”. Cheers Acer ventura

A bit of tidying up from here... http://arbtalk.co.uk/forum/general-chat/73829-isa-best-management-practices-tree-risk-assessment-traq-5.html#post1113787 ...which I had forgotten about. No it doesn’t. A life is a life. A risk of 1/10 000 for the loss of a life is an advisory threshold of tolerable risk. The Value of Statistical Life (VOSL) is used to apportion a monetary value of damage to property, as a fraction of the VOSL, and determine whether a risk is ALARP. The Risk of Harm was coined because it covers, fatality, injury, and damage to property. Cheers Acer ventura

Hi Craig And that’s one of the beauties of using probability to estimate the likelihood of an uncertain event. You’re not saying it will, or will not happen. You’re estimating the odds of it happening. One of characteristic of probabilities that many don’t immediately grasp, when you use them in a tree risk assessment, is that you’re conveying your uncertainty about what might happen in the event of a tree or branch failing and hitting someone or something. There’s a really neat example that Douglas Hubbard uses which illustrates this, and also neatly nails unfounded criticisms about the use of probabilistic numbers implying false precision. It’s from Chapter 7 - ‘Worse Than Useless; The Most Popular Risk Assessment Method and Why it Doesn’t Work’, in ‘The Failure of Risk Management: Why it's Broken and How to Fix it’. It's a great book, and is part of the foundations on which this thread is built. If you have no idea whether it’s going to rain tomorrow you could express this likelihood with, ‘there’s 50% chance of rain tomorrow’. 50% looks like a very precise figure but what it means is that you’re expressing the highest level of certainty that you can about whether it will rain tomorrow. There isn't a higher level of uncertainty. Cheers Acer ventura

Hi Paul Initially, it can take some time to get your head around numbers if you’re not used to working with them. Doing the research into this thread, I spent a long time scratching my head during the working out, discussing it with others, and checking it over and over. It’s worth a re-visit after some digestion, taking things in bit-by-bit because what the numbers do is remove the subjectivity and ambiguity. What they give is objectivity and clarity. Asking questions, like you’ve done here, is also a good way of getting your head around things. I tend to the view, there's no such thing as a dumb question and would be great to have some more. All I've done here is outline what I think I've worked out. I could be wrong, and I would welcome anyone posting to explain where and why I've gone wrong. One of the reasons I decided to set this thread up the way I have is so it will be there, step by step, and completely transparent, for anyone to go through and agree or disagree with.

Hi Tony I was a 110% sure you were going to say that. Good call on my wrong use of tense. Cheers Acer ventura

There’s not been that much traffic in the way of contributions to this thread so far (more off forum), but clearly there’s a lot of interest judging by the amount of views after a new post. The lack of contributions doesn’t bother me because the way I figure it, this being the interweb, and a discussion forum for arborists, if I’ve got something wrong, there will be at the very least someone eager to point this out. And what we’ve gone through so far is largely mathematical and binary. I’ve reached a crossroads here and I’m not sure which section to develop next. The purpose of the thread is to explore the possibilities of QTRA and the Risk BMP working in tandem, by analysing the Risk BMP for common ground. When I first started to look at this I thought the main points of potential conflict were going to be the ‘risk tolerance’ issue, and that there’s no scope for a risk to be As Low As Reasonably Practicable (ALARP) with the Risk BMP. That was before I started dissecting the matrices. Looking at the matrices in detail was an interesting exercise and there are more issues with the matrices than I’ve gone through so far - which might be the underlying reasons to explain why the output risk ratings look like they could be untenable - and I was going to explain what they are. Upon reflection, I’ll go with what I think is the more important issue, which is ‘risk tolerance’, on Monday. This is less of a yes/no binary subject and will hopefully generate a bit more forum debate. As ever, any comments, on or off forum, are very welcome. Cheers Acer ventura

Hi Guy A point well made about a risk assessors own definitions of what Possible means to them overriding guidance. For me, Possible is the most problematic descriptor in the Likelihood of Failure ratings because “Failure could occur, but is unlikely” covers everything apart from Imminent. I’m well aware that part of using the Risk BMP is that you buy into their definitions of what Possible means relative to other Likelihood of Failure descriptors, but boy does it cause some cognitive dissonance. However, are you sure you've got the Possible = No Low Risk conclusion right? According to the Likelihood Matrix, when the Likelihood of Failure is Possible you can get an Unlikely with Medium, Low, and Very low Likelihood of Impacting the Target. In the Risk Matrix an Unlikely Likelihood of Failure & Impact = Low Risk rating for all Consequences. Something I’ve not figured out yet, and so far found nothing published, is reference to risks being assessed and rated using two matrices. I’ve not encountered this double matrix approach before, and it has occurred to me the intrinsic limitations of using risk matrices might be especially compounded where two are used and the outputs of one are fed into the other. For example, the lack of exclusivity and significant overlap in the Likelihood of Failure and Impact ranges in the Risk Matrix are the primary reason that qualitative lower risk ratings can be quantitatively higher than the qualitative higher risk rating above them. Does anyone know of any double matrix risk assessments? Cheers Acer ventura

Hi Jules I think everyone else is likely to be pretty bored of this sub-thread, and it's way off topic. I've transplanted it to the QTRA v5 thread, as I will do my reply to you about Monte Carlo simulations, and what you think the Risk of Harm means, when I get the time. I've replied to you here. http://arbtalk.co.uk/forum/general-chat/72575-quantified-tree-risk-assessment-qtra-version-5-questions-answers.html#post1114407 Cheers Acer ventura

I've moved this probability debate over because it's getting way off topic, and I was going to drag one more subject in that thread because it relates to Monte Carlo simulations in QTRA v5. It starts here. http://arbtalk.co.uk/forum/general-chat/73829-isa-best-management-practices-tree-risk-assessment-traq.html Hi Jules You almost answered your own question with your contradiction in your penultimate post. Arsing about with probabilities can all too easily mess with people’s heads, and I include myself here. Try this; Take 30 unbiased coins. Toss each one in turn. On the table you now have 30 coins with 15 coins tails up and 15 coins heads up. Is the probability of 15 of the 30 coins being tails up? 15/30 = 1/2? Or is it 1/2 x 1/2 etc 15 times = 1/33 000? Cheers Acer ventura

Here’s the BMP Risk Matrix, with the cells completed for their respective risk ratings, but coloured and populated according to QTRA advisory thresholds (second slide) and the Risks of Harm, as probabilities, for damage to property only. The QTRA Target Ranges for property are labelled ‘T1’ etc, in the Consequences column titles, and were derived from the descriptions of damage to property in the Risk BMP and TRAQ Manual. T1 = £1 500 000 T2 = £150 000 - >£15 000 T3 = £15 000 - >£1 500 T4 = £1 500 - >£150 T5 = £150 - >£15 T6 = £15 - £1 This is a tentative categorisation because the monetary damages in the Risk BMP are not clearly defined and ambiguous. They are described as; Negligible = Low monetary damage Minor = Moderate monetary damage Significant = High monetary damage Severe = There is no monetary damage descriptor There is no indication of what Low, Moderate, or High monetary damage are in terms of actual monetary value. There are other descriptors to give some clues; Negligible “A small branch striking a fence.” “A medium-sized branch striking a shrub bed” “A large part striking a structure and causing low monetary damage” Minor “A small branch striking a house roof from a high height.” “A medium-sized branch striking a deck from a moderate height.” “A large part striking a structure and causing moderate monetary damage.” Significant “A medium-sized part striking an unoccupied new vehicle from a moderate or high height” “A large part striking a structure and resulting in high monetary damage” Severe No monetary damage descriptor. It’s not clear; At what diameter and/or length a ‘small branch’ becomes a ‘medium-sized branch’, or a ‘medium-sized’ branch becomes a ‘large part’. Where a ‘low height’ becomes a ‘moderate height’, or a ‘moderate height’ becomes a ‘high height’. Or, to play out one example, whether a medium-sized part striking an unoccupied new vehicle from a ‘low’ height means the Consequences are now Minor with a ‘moderate monetary damage’ rather than Significant with ‘high monetary damage’. Or, how older a vehicle has to be than a new one in the Significant example, before the Consequences of being struck by a medium-sized part from a moderate or high height go from Significant high monetary value down to Minor moderate monetary value. And so forth. As you can see from the matrix, the QTRA Risks of Harm are the same for all the Likelihood of Failure and Impact input ranges of Very Likely, Likely, Somewhat likely, and Unlikely (the upper values for Unlikely) within each Consequence column. This is because all the Likelihood of Failure and Impact input ranges (the upper values of Unlikely range) are all extremely compressed into one QTRA Probability of Failure Range 1 (1/1 – 1/10). So, the 'Severe' Consequences risk ratings column of Extreme, High, Moderate, and Low are all red unacceptable risks at 1/3 with Low being less than <1/3. The 'Significant' Consequences risk ratings column of High, Moderate, and Low are all red unacceptable risks at 1/300 with Low being less than <1/300. The 'Minor' Consequences risk ratings column of Moderate and Low are all amber unacceptable risks to impose (unless there is great value or the stakeholders agree) at 1/3 000 with Low being less than <1/ 3 000. The 'Negligible' Consequences risk ratings column are all Low, and are yellow risks at 1/30 000, except the 'Unlikely' cell which is <1/30 000, which are tolerable risks that can be imposed if the risk is ALARP. There is considerable compression in the Severe Consequences range because (T1 – T2) £1 500 000 - >£15 000, of the QTRA property damage range, falls within Severe. Any suggestions as to where the QTRA Target Ranges might be any different for Negligible, Minor, Significant, and Severe Consequences to property? One thing I haven’t been able to figure out is how or why the “disruption of arterial traffic or a motorways” is in the same Consequences range, Severe, as a death. Cheers Acer ventura

My apologies for the delay because I’m now going to post the QTRA populated BMP Risk Matrix tomorrow morning. Discussing the Consequences input to the Risk Matrix, relative to the QTRA inputs and Risk of Harm outputs, means the accompanying text is lengthier than I hoped it would be. I’m sure it will benefit from a night’s sleep, some editing, and my near hapless ability to effectively proof read out the inevitable word blindness over very short time periods (though I quite like the irony of a pendant hat). Jules, I will reply to some of your posts. However, I’m going to do so on the QTRA v 5 thread in an effort to prevent this thread from being driven into peculiar backwoods cul-de-sac where feet, footwear, and the possible lack of them, are not significant in either a figurative or standardly deviant way. In the meantime, see whether you can get your head around the difference between ‘independent’ and ‘dependent’ probabilities and why you don’t cumulatively multiply the former. Cheers Acer ventura

Hi Jules You surprise me because the use of Monte Carlo simulations in version 5 is one of the biggest turn-ons for me, and takes the credibility of QTRA to another level as a professional robust means of assessing tree risk. Are you aware of how often it’s used in risk assessments by other professions? Do you want to raise your concerns about the use of Monte Carlo simulations in the QTRA v5 thread, which is gathering dust here... Quantified Tree Risk Assessment (QTRA) Version 5 - Questions & Answers - Arbtalk.co.uk | Discussion Forum for Arborists ...so the thread can get back on TRAQ? I've been busy today, so will post up the QTRA probabilities, as far as I've been able to work out, in the Risk BMP Risk Matrix tomorrow. In the meantime, hopefully the better angels of James' nature will persuade him to be generous and share his explanation of where he thinks I've gone wrong in working out the Likelihood of Failure and Likelihood of Impacting the Target probabilities. Cheers Acer ventura

Hi Jules The 'Risk of Harm' at the top of the slide are my words, not NASA’s which are in quotation marks, because the slide is in the context of a QTRA presentation and that’s how QTRA refers to the risk probability. Nicholas Johnson referred to a ‘human casualty’, which is a euphemism for a fatality. I suspect part of NASA's model is that if someone is unfortunate enough to be hit by a bit of falling satellite they're likely to die. If you look online, at their website and articles about what the Orbital Debris researchers have been looking to achieve, it seems clear the consequences are a fatality. Cheers Acer ventura

Hi Jules Perhaps worthy of another thread? Given Nelda Matheny is one third of the principal authors of the Risk BMP, I think she might disagree with you that’s she’s misunderstood her previous work with Jim Clark. I agree Matheny & Clark were pioneers with their approach in 1994. But that’s 20 years ago and things move on and develop. Moreover, there are some really important caveats in their publication about how their hazard ranking system should be applied which are still frequently ignored. One of the interesting things I’ve found researching tree risk, and running training workshops, is just how many versions of Matheny & Clark’s hazard ranking system are knocking around out there as methods of assessing risk, instead of ranking hazard. Often with additional numbers bolted on to try and massage the end result to reflect a level of risk that the designer thinks the tree should have. Cheers Acer ventura

Hi Jules Pendant’s hat on - it would be less than <0.000061 for a Low risk rating of less than <1/2 for a fatality. But that would only be the case if you pick the 14 out of the 30 trees that all have an equal Low risk rating of less than <1/2 of failing and causing a fatality, AND then identify the correct sequence in which each one of them fail and cause a fatality over the year. What you’ve done is calculate the odds of picking the correct sequence of heads or tails with 14 consecutive tosses from an unbiased coin. Cheers Acer ventura

Hi James Oh, I don’t know. Sharing your wisdom and insight on the Arbtalk discussion forum for the greater good? Given the subject matter, how many trees could be affected, how much money could be spent on tree risk management, potential loss of tree benefits, and the duty of care issues on the part of the risk assessor and risk manager, I’d of thought the content of this thread would have a much wider audience than QTRA Registered Users. If I thought it was just of interest to them then I’d have posted on the QTRA Discussion List. I did invite contributions to the thread from the QTRA Discussion list, UKTC, and the Linkedin American Society of Consulting Arborists (ASCA) Group.

Hi Jules There’s plenty. Google Tolerability of Risk Framework, and/or Tolerable risk 1/10 000, with the respective country of interest and take your pick. I often use this in presentations, which some of you might remember from a UKTC thread some years back. I followed up the story and found a guy called Nicholas L Johnson, NASA Chief Scientist for Orbital Debris, who I emailed to ask about how they calculated the risk, and what level they managed their risk to. Here’s part of his reply. What’s really neat about this is that those agencies who have signed up to the agreement now engineer satellites so that upon re-entry no bit survives is large enough to have a risk greater than 1/10 000. QTRA users are able to apply exactly the same principle when doing walkover surveys when they have the Target Range because by back calculating they can work out which Size and Probability of Failure is required to get a risk higher than 1/10 000, and restrict the assessment to those parts. Something else I found interesting about the news story is it looks like the press release prefixed the 1/3 200 with the word ‘just’ and no one batted an eye-lid because it ‘anchored’ them to the word. Whereas, had it been as ‘high’ as 1/3 200 there may have been more concern because of where it sits in ToR. It's an insight into risk communication. BTW the UARS was launched before the inter-agency agreement. I followed this up awhile back and China and India have also signed up to the 1/10 000 threshold. North Korea, on the other hand... Cheers Acer ventura

Hi Jules I was going to run the QTRA (Monte Carlo simulation) outputs tomorrow and populate the Risk Matrix as far as I’ve been able to work out. As you might have figured out from my response to Paul, the overlapping Risk BMP Likelihood ranges all fall within one of the QTRA ranges so it’s not an issue with running the QTRA side of things. There are some other problems with compatibility though, which I’ll go through. Cheers Acer ventura

Hi Jules I was looking to express what the upper value of a risk ranking of less than <1/2 per annum for a fatality would mean in language outside of probabilities. Another way. Consider there are 30 trees in the school with a Risk BMP ranking of Low risk <1/2 for a fatality per annum. What’s the worst that could happen and the risk ranking still be correct? Does that clarify it? Cheers Acer ventura

Hi Jules I’m not. I’m representing myself. My personal interest and professional development in tree risk assessment and tree risk assessment systems. And because it was raised by people I’d trained and talked to who are ISA TRAQ accredited, or who are interested in becoming so. On the commercial interest side. Just to clarify, I don’t draw a salary from QTRA. I do get paid by QTRA only when I run workshops, which I disclosed in my opening post. I use QTRA because I think it’s the best tree risk assessment methodology, and for that reason I’m happy to train others in its use, help develop it, and discuss its application on forums such as this. I used to use Matheny & Clark's Hazard Rating system in the 1990s until QTRA came along. If tree risk assessment system comes along that’s better than QTRA then I’ll drop it and use that.

Hi Paul I reckon most things that make you stop and give something a bit of thought, no matter how briefly, are often worthwhile exercises. The main issue with your suggestion of the upper value of ‘Improbable’ being at 1/1 000 would be that with the highest Target value (1/1), and a large tree or very large branch (1/1), we would be looking at a risk of 1/1 000. 1/1 000 x 1/1 x 1/1 = 1/1 000 A 1/1 000 risk is extremely high by ToR thresholds and is not a risk you can reasonably impose on the wider public. Many of our high target mature urban trees would be an unacceptable risk. Similarly, I think the lowest value of the Improbable Likelihood of Failure range at 1/1 000 000 would be too high as well. If we take an actuarial perspective and look at some figures that we do know; Number of trees in the UK around 5 000 000 000 (5 billion) Population of the UK around 62 000 000 Risk of death from tree failure in the UK 1/10 000 000 Nearly all the trees don’t fall down each and every year Then I think setting the lower value of Improbable Likelihood of Failure at 1/1 000 000 is erring on the risk averse ‘safe’ side. When looking for common ground between QTRA and the Risk BMP one of possible problems that popped up is that 3/4 of the Risk BMP’s Likelihood of Failure ranges Imminent = 1/1 Probable = 1/1 - >1/2 And the top value of Possible = <1/2 Fall into the highest Probability of Failure (PoF) Range 1 : 1/1-1/10 with QTRA. The Possible Likelihood of Failure range goes from <1/2 down to “but is unlikely” and looks like it covers all seven QTRA PoF ranges if the upper value of ‘Improbable’ is less than <1/1 000 000; which it must surely be. Cheers Acer ventura

Where does the Risk BMP’s risk ranking of Extreme, High, Moderate, and Low for a fatality sit in ToR? We can see part of the Low risk range for a fatality, but can’t make out where Extreme, High, and Moderate are, so let's zoom in. Can still only see part of Low risk range for a fatality, which extends off the bottom. Zoom in some more. Can still only see part of Low risk range for a fatality. Zoom in some more. Here they are, with the Low risk range for a fatality extending off the bottom. To put what the Risk BMP risk rankings are saying in terms of tangible numbers in relation to say schools, which I opened the thread with. If you have a class of 30 children and a year later 14 of them are dead, this would be ranked as a Low risk. If 29 of the 30 died, this would be ranked as a Moderate risk. By looking at the risk ranking ranges it is also becomes apparent; A Low risk of death can be higher than a Moderate risk of death A Moderate risk of death can be higher than a High risk of death. A High risk of death can be equal to an Extreme risk of death. When I went through this exercise my gut screamed this can’t be right. However, my head can’t come to any other conclusion because that’s what the logic and maths is saying. As I’m sure you can imagine, I’ve gone back over this a number of times and can‘t find fault. It's one of the reasons why I've laboriously and transparently shown the workings out in previous posts, because these risk rankings look like they're a bit of a problem. Can anyone point out where I’ve gone wrong here? As I said in the opening post, one the reasons I've had my sleeves rolled up and been looking at this in so much detail is because of requests about whether QTRA and the Risk BMP can work in tandem from TRAQ accredited arborists and QTRA Registered Users who want to be ISA accredited to assess tree risk. I’ll post the QTRA outcomes in the Risk Matrix tomorrow. Cheers Acer ventura

Still work in progress with the upper values and boundaries of the Likelihood of Failure range Improbable, and Likelihood of Impacting the Target range Very Low, in the Likelihood Matrix. Nonetheless, this is what we know so far about the four Likelihood of Failure and Impact ranges from the Likelihood Matrix, which can now be fed into the Risk Matrix. One thing that is readily apparent when the Likelihoods are viewed like this is the ranges are no longer explicit and there are overlaps. The other axis in the Risk Matrix is populated by four rating ranges of 'Consequences', descending in magnitude; ‘Severe, Significant, Minor, and Negligible’. The Risk Matrix cells are partitioned into four risk rankings, descending in magnitude; ‘Extreme, High, Moderate, and Low’. The easiest, and probably most important, of the Consequences to have a look at first is a fatality. A fatality is the highest value in the Severe Consequences range, and in terms of probability is 1/1. The probabilities for a fatality are the same as they are for the four Likelihood ranges because they are simply multiplied by 1/1. Cheers Acer ventura PS Thanks to whoever fixed my post No.12 on Saturday.

Hi Craig Wouldn't your ranges for 'Likelihood of Failure' be t'will , t'might, t'could, t'unlucky then? Cheers Acer ventura