Acer ventura

I recently posted a variation of this on the UKTC's ‘Defect led Tree Risk Assessment - let's change the title! Numbers and Words’ thread, and given where we are with this thread about ‘exactness’ and ‘uncertainty’, it seems appropriate to raise it here. Currently, QTRA is undergoing the next step in its development to better encapsulate the ‘uncertainty’ element of risk. We’ve adopted a tame mathematician and got them running Monte Carlo Simulations on thousands and thousands of possible outcomes within the three input ranges at particular confidence levels to generate probability distributions and their respective averages. Those average outputs can then be related to the three thresholds in the 'Tolerability of Risk Framework (ToR)' (1/1,000 - 1/10,000 - 1/1,000,000) – though we may have 4 to include 1/100,000. What all this means is we’ll end up with a RoH that better express the uncertainty in the risk assessment, rather than specific numerical probabilities. The result of this evolution is that QTRA will be probabilistically more robust, easier to use for the risk assessor, and easier to interpret for the risk manager. For those of you who are interested in what on earth this Monte Carlo jibber-jabber is all about, here's an easy to grasp short visual example about it which is pretty good. Well, pretty good until Kevin strays out of his field of expertise and drifts into European geography. Cheers Acer ventura [ame=http://www.youtube.com/watch?v=Xaymy3Blnq4]Episode 2: Monte Carlo Simulation - YouTube[/ame]

Hi Robin And neither is QTRA 'exact', nor does it claim to be. It's a 'reasonably practicable' and 'proportionate'* risk assessment. You can't get "an outcome of 1/10050 is safer than a tree with 1/9997" with QTRA. I thought I'd covered this in the previous post, so I'll expand and try with a different tack. Paradoxically, perhaps it's all the effort that has been made to take 'variability' by the hand and be inexact that has made you think that QTRA is trying to be unreasonably exact. Let's roll up our sleeves and delve into the nuts and bolts of the QTRA engine. This is dirty under the bonnet stuff poking around in the cylinder head of underlying principles, which it is not necessary to know in order to drive QTRA (line up three ranges), but may help you understand what is going on. RoH is not a measure of 'exactness' or 'precision', and neither is the PoF. It's a probabilistic risk assessment and therefore an expression of 'uncertainty', or 'inexactness', if you like. Through probability calculated from the highest value of the broad ranges, we can operate within the realms of 'reasonable practicability' and 'proportionality'. The element of the QTRA calculation that has the greatest level of uncertainty is the PoF. You may have noticed in the example I gave previously, unlike the Target and Impact Potential, there was no scope to refine the PoF range. We are particularly mindful of uncertainty, or inexactness, in PoF which is why in QTRA guidance PoF has broad ranges and there's no scope to refine it because with current knowledge we don't think you can do this with a greater level of certainty. To try and refine PoF is not 'reasonably practicable' or 'proportionate'. In the example I gave, PoF falls within a range at some point between 1/100 -1/900, so we're up front with the fact we're not sure, we're uncertain, we're inexact and therefore to be on the safe side it is the highest value of 1/100 in the range that is going into the risk calculation. That RoH is expressed to 1 significant figure only is an extension of this uncertainty. RoH is a worst case scenario expression because it is the product of the highest value of the three input ranges, and in the example I used in the previous post where we get; RoH = 1/10,000 (T = 3 x IP = 2 x PoF = 3). However, if you calculate the actual probabilities T= 1/72 x IP= 1/2 x PoH= 1/100 = RoH 14,400 Because PoF is expressed to 1 significant figure, then so is the RoH, and 14/400 is therefore expressed as 1/10,000. What we've done with QTRA is put a lot of effort into embracing 'uncertainty', and worked really hard at not being 'exact', because that's what you do when you’re being reasonably practicable and proportionate and quantifying risk with imperfect knowledge. Does that help? Cheers Acer ventura *'Reasonably practicable' and 'proportionate' means; Doing only what is necessary and reasonable in the given circumstances. Considering the benefits derived from the hazard as well as the risks. Expending resources only at a level that is proportionate to the projected reduction in risk. Reducing risk to a reasonable level but not necessarily eliminating it. Balancing Costs with Benefits.

Hi Robin I’m not sure what your question is here. Do you mean RoH or PoF when you say ‘PoH’. Or are you referring to ‘Pooh’, who clearly understands the principles of QTRA, ‘after careful thought’? (see attached). Are you asking a variation of the accusation of ‘implied precision’/‘exactness’ to the subjective probability/professional judgement PoF element of the risk assessment? Or similarly about the ‘implied precision’/‘exactness’ in the RoH? Just to reiterate' date=' the ‘exactness’ of the tolerable level of risk ‘threshold’, the 1/10,000, is not set or ‘quantified’ by QTRA. It’s one of the thresholds of tolerable or acceptable risk that has been established by a myriad of external independent bodies that we’re looking to assess and measure the risk from trees against; which is why we’re quantifying. Rather than carpet bomb your question with an answer to a question I think you’re getting at and ending up going Cambodia on it, could you please have a go at clarifying what you’re asking? Acer/David. I’m not that fussed. I’ve adopted Acer ventura as a forum name since January 1999 when out of a drunken Christmas catch up with Chris Hastie the UKTC germinated and I made the first posting. It’s borne out of a frustration that I decided to pretend to be a grown up and didn’t have the cahoonas to call my consultancy Acer ventura’s Tree Detective Agency, and it’s become habitual. Cheers Acer ventura

Hi Robin It's a great question. As I said earlier in the thread, there's no such thing as a daft question.

Hi David On the contrary, that's a really useful contribution and clarification because it highlights the issue that QTRA is about assessing the 'tree risk' and not looking at 'tree defects'. Target valuation is the most important element of tree risk. Getting this part of the risk equation wrong can mean the risk assessment is as inaccurate as missing a defect-riddled tree of poor vitality with symptoms of incipient failure adjacent to a high value target, or pointlessly pursuing the removal of deadwood in parks. One of the skills that an arborist needs to acquire with QTRA is working out Target values. With roads that's often quite easy because the Department of Transport has average road use statistics for each region and class of road, and the Highways Department of the Local Government Agency often has very accurate data. For pedestrian use, the owner or manager is usually the most informed and the risk assessor the least knowledgeable. Part of the skill of the assessor is teasing the answers out of those that know the site better. This is key element because trees can fail at any time of the day over the year whether people are there or not. You need an annual average occupancy, which includes winter and night time, and it's not uncommon for only the more apparent higher occupation rates to be recollected unless prompted and directed by the assessor. By quantifying the risk assessment, there are sometimes clues as to whether the information you're given about the Target might be unintentionally biased or inaccurate. So if there were 7 million going through Kenwood Gate a year than that would equate to just over 800/hour, 14/minute, or 1 every 4-5 seconds, pedestrians going through the gate, day and night, fair and foul. If the gates are closed at dusk, then these footfall rates would be roughly doubled - not that doubling of a rate over half the time would affect the average occupancy over the whole year. I was once asked to carry out a QTRA at the home of a barrister who had been involved in a high profile tree failure case. He had a very nice house adjacent to a single track country road, which I know from experience is unlikely to figure on Highways monitoring. I asked my client to be conservative about it and, "On average, over a year, how many cars/day do you think come down this lane?" After a bit of thought, his reply was “50”. I said this would be my Target and I would reference them as my source in my report. Initially, they seemed a bit shocked but then upon consideration, realising this was something they would know far more about than I would during my hour on site, they recognised it was a 'proportional' and 'reasonably practicable' solution to assessing the tree risk. Cheers Acer ventura

Hi Rob In the context of this thread, just have a read of David Lonsdale's 'Supplementary' at (9). It's not very long but a marvellously detailed forensic dissection and dismissal of the attempted criticism levelled at QTRA, but also addresses some of the questions you have raised on here. Cheers Acer ventura

On the RobArb QTRA thread that gave birth to this one, a gentleman by the name of showoffsummer raised Julian Forbes-Laird’s (Claimant's expert witness) - 'Supplementary Report on QTRA' from the Bowen & Others v National Trust court case (8). http://arbtalk.co.uk/forum/general-chat/50770-qtra-im-sorry-i-dont-agree-2.html#post771579 You might be interested in David Lonsdale’s response to the ‘Supplementary Report on QTRA’ (9), and how David used QTRA and the Health and Safety Executive's (HSE) Tolerability of Risk (ToR) framework in his expert report (6) when determining whether the risk posed by the tree was tolerable. There’s a whole bunch of files relating to the case which are numbered in chronological order, after the Judgment, which are available through the QTRA website: Quantified Tree Risk Assessment 1) The Judgment 2) The Particulars of Claim 3) The Defence 4) The Claimants' arboriculturist expert witness (Julian Forbes-Laird) report 5) Appendices to Julian Forbes-Laird's expert witness report 6) The Defendant's arboriculturist expert witness (Dr David Lonsdale) report 7) The arboriculturist experts' joint statement 8) Julian Forbes-Laird (Claimant's expert witness) - 'Supplementary Report on QTRA' 9) Dr David Lonsdale (Defendant's expert witness) - response to the 'Supplementary Report on QTRA' 10) Dr David Lonsdale’s Commentary on the case Cheers Acer ventura

Hi Rob Thanks for the appreciation and I’ve been happy to try and answer your and Tony’s questions. I take the position there’s no such thing as dumb or bad question, and one of the things I think is really important when running both the QTRA and VTA workshops is ensuring that everyone is comfortable so they don’t hold back. I wanted to do that on here as well. I’m very mindful QTRA can initially mess with arborists heads because it may seem contrary to what they have been brought up with and what they learned in college. An arborist called Dom Scanlon phrased it very nicely after he did the training and said he had to ‘unlearn’ what he had learned at college in order to get his head around it. Something I can sympathise with because I was at the same college with him. I think one of the main reasons for this ‘cognitive dissonance’ is because QTRA is founded on the principles of ‘risk’ and ‘probability’, which apply to any field, rather than having a look at defects or hazards in a tree. Hence the opening post on this thread was about what risk is, and what a tolerable or acceptable level of risk was. It’s interesting that on training seminars and during presentations non-arborists can often ‘get’ QTRA before arborists do. Dom also told me a tale about when he was explaining to a potential client, who was an engineer, how he assessed tree risk with QTRA and showed him the manual calculator (the wheel). Whereupon the engineer got something very similar out that he used in his profession. Needless to say, there was little effort required in persuading them in the merits of quantifying risk. The maths and probability side of risk is also new to many, but a lot of that is by way of understanding why QTRA does what it does. All the assessor has to do is work out the Target value in a broad range first, which is the bit they know the least about and may require a bit of work asking their client or talking to highways, and then the arborist can get onto their favourite bit of looking at trees.

Hi Tony Where is this? How many people/hour funnel through there when you have wind gusts hitting gale force 8 (39–46 mph 34–40 knot)? Say you have Gale force 7 - "Effort needed to walk against the wind", and gusts hitting Gale force 8, "Progress on foot is seriously impeded". 40 knots is when you start to get branch failure. That's the pedestrain occupency you would put into the Target. It's not thorny' date=' it's really important because it informs 'proportionality'. A subject for another thread. Perhaps when this one has run its course. Cheers Acer ventura

Hi Rob I don't follow the argument that the management and assessment of risk from trees should not be measured and relate to what we know are published tolerable or acceptable levels of risk because a part of the risk assessment requires some subjective judgement. How do we know what we’re doing otherwise? Don’t underestimate the immense value to duty holders of managing tree risk to a published tolerable or acceptable level of risk because they can adopt it as policy. Not only can they then compare their exposure to tree risk with all the other risks they have to manage, but defend claims against them in the event of acceptable risks being realised. What are the statistics which are subject to too much human interpretation?

Hi Tony Your post has reminded me I still owe you a reply on the UKTC’s "Limitations of Defect led Tree Risk Assessment" thread, which I’d begun to draft and then got distracted from by work, leaving it to curl up at the corners and gather dust, but I think I may have answered it above. The VTA approach is an integral part of QTRA. It’s not separate thing, it's what drives the Impact Potential (Size of Part) and Probability of Failure (PoF) components of the risk assessment. Though VTA is important in helping you work out the SoP and PoF, it alone won’t tell you what the RoH is relative to a published level of tolerable or acceptable risk, unless you quantify it and also measure the Target. This is something that is very much focused on during the VTA day, which runs paired with the QTRA day, and looks at how decay colonises trees, how trees with good vitality are mechanically self-optimising, the limitations of strength loss formulae, mapping decay/soundwood with the Thor 710 hammer, and calibrating PoF.

The next bit. Objectivity and Subjectivity. With an Objective numerical probabilistic threshold of risk to measure against - the 1/10,000 and 1/1,000,000 - what QTRA aims to do is maximise the Objectivity of the assessor and minimise the Subjectivity. It might be easier to illustrate how this works with an example. But first some context. The most important element of tree risk is the Target, because in the risk equation; Risk = Likelihood x Consequence Target = A proportion of the Likelihood of occupation AND the Consequence to what might be hit. Probability of Failure = A proportion of Likelihood Impact Potential (as a function of size) = A proportion of the Consequence I’ve attached a Powerpoint slide that illustrates the primary importance of Target in the relationship. Target is not only the most important element of tree risk assessment, it’s also the one that you can be the most Objective about. If we look at a recent site I was involved in, there is a 30mph residential road which was the highest value Target. Before I went to the site, the council’s highway department told me 4,441 vehicles a day used this road. Rounding this figure up to 4,500, and putting it into the QTRA software calculator, I get a Target value of 1/10. In other words, if a branch or tree falls onto this road, there is a 1/10 chance that it will either hit a car, or a car will not have time to stop before hitting the branch or tree. On a sliding scale of Objective – Subjective, I’m not sure it’s possible to get any closer to being Objective than taking this measured and quantified approach to tree risk assessment when assessing the Target. It’s an interesting exercise to compare this measured quantitative approach with trying to so do the same qualitatively, using words or ordinal numbers, even if you knew the road traffic figures. Or how you might then relate that to a tolerable or acceptable level of risk. The element of QTRA, as with any tree risk assessment, that is most Subjective is Probability of Failure (PoF). As you pointed out Robin, this is partially down to experience. We could also reframe a negative term ‘Subjectivity’, with a positive phrase ‘Professional Judgement’. After all, they both have the same factual meaning, but opposite emotional meanings. Let’s stay with the original dichotomy. I’ll explain how we go about minimising the Subjective element of PoF with QTRA. A self-optimised tree with good vitality has a PoF of less than 1/1,000,000, and we take this as a base point reference. So, where we encounter a tree with defects and the tree is perhaps of reduced vitality, isn’t showing positive signs of having mechanically adapted, or has incipient indications of failure, the question the assessor asks themselves is, within broad ranges, how many times more likely is the tree to fail, than the 1/1,000,000 tree. Is x10, x100, x1,000, and so forth. We’ve developed and taken this approach in QTRA training and Update sessions for a couple of years now. What I find interesting is that during the outdoor practical sessions, no matter the range of experience of those that attend, once we have calibrated an agreed starting point, and take this considered and measured approach, a very narrow consensus of PoF Subjectivity/Professional Judgement is reached within quite a short period of time. The widest divergence of opinion is seldom more than 1 x10 range out, and that tends to occur where the RoH is furthest from 1/10,000, so it’s not important when risks are getting down to the one in hundreds of millions. Once we start getting to a RoH that isn't tolerable, towards 1/10,000, there’s a convergence of consensus because the tree defects are more pronounced, and the tree’s poor vitality is more discernable. Apologies for the long post, but after spending a bit of time on it I figured I couldn’t break this part of the discussion down into anything less than one dump, so to speak. Let me know what you guys think. Does all this make sense? Or do you think I’ve been at the two-stroke and am speaking in tongues. Cheers Acer ventura QTRA Target Importance.pdf

Hi Robin Good to hear from you, and it would be great to hear from others who have something to say. I'll try to help explain and expand on this bit first; I suppose the conical ToR framework does resemble a hat that could be worn at a party. A party that would be incredibly dull and lacking the kind of physically reckless activities that often make such events memorably entertaining. Nonetheless, it’s a hat that’s had a lot of very careful thought and research go into its appearance and the numbers that make up its dimensions than ‘that’ll do’. From research by the Royal Society in ‘Risk Assessment: A Study Group Report’ in 1983 where 1/1,000,000 was raised as a level of risk that was so low that no benefits were necessary in order for society to accept it. To Henderson’s ‘Living with Risk’ – The British Medical Association Guide, in 1987, which concluded, “few people would commit their own resources to reduce an annual risk of death that was already as low as 1/10,000." Finally, the HSE in 1996, where they concluded, “For members of the public who have a risk imposed on them 'in the wider interest' HSE would set this limit at 1/10,000 per annum.” It goes way beyond our borders though. NASA adopted 1/10,000 as a human casualty threshold in 1995, and it was later accepted by the US government in 2001. The European Space Agency and the Japanese Aerospace Exploration Agency have also adopted the same threshold. Interestingly, satellites are now designed so that upon re-entry, when they fall apart, the pieces that fall back to earth will be small enough so the risk is less than 1/10,000. The same principles can be applied with QTRA, where, with a known target value you can work out the minimum size of part that could result in a risk greater than 1/10,000, or 1/1,000,000, and restrict your tree assessment to defects to those parts, and ignore anything with a smaller diameter. I'll get back to you with objectivity v subjectivity side of things this afternoon. As you've endured the UKTC's "Limitations of Defect led Tree Risk Assessment" thread you've probably already got some insight into this. Cheers Acer ventura PS Excuse me if I screw up the quote protocol, I'm still learning to drive the message box on here.

Hi Arbtalk RobArb’s thread http://arbtalk.co.uk/forum/general-chat/50770-qtra-im-sorry-i-dont-agree.html has been brought to my attention. It raised a number of questions and had some misconceptions about QTRA, so I thought I would trek over from the UKTC forum, dip my toes into Arbtalk, and chip in to answer some of the questions raised and clarify a few things. By way of introduction, I’m David Evans and outside of my consultancy work at The Arbor Centre I run QTRA and VTA workshops. I’m starting a new thread because the old one died, and it went off topic a fair bit. Please feel free to ask any questions and I’ll do my best to help out. I’ll begin at the beginning, with some of the points raised in RobArb’s first post. I appreciate the thread developed since then, but some of the issues raised are foundations on which everything else is built, so it’s pretty important to get them clear. <<I just can't be drawn into a system that relies on statistics as its main factor.>> QTRA doesn’t rely on statistics as its main factor, it’s founded on probability, which is the language of risk. Risk = Likelihood x Consequences. With QTRA, the likelihood and the consequences of someone or something being hit by a tree, or part of a tree. <<it has been deemed by QTRA that an acceptible risk is 1/10000. This from what i can tell has been derived from historical tree failures>> The 1/10,000 threshold does not come from QTRA, nor does it relate to historical tree failures. In summary, 1/10,000 is a ‘tolerable’ level of risk that can be imposed on the public for the wider good where the risk is As Low As Reasonably Practicable (ALARP); as outlined by the Health and Safety Executive (HSE). This is a really useful defendable threshold for managing tree risk to because it’s a level of risk that has been determined by an independent watchdog appointed by the government on behalf of society at large. It is not a level of risk that has been made up or is the opinion of an arborist. The reason QTRA was created was to enable the user to measure and calculate risk as a numerical probability, so the tree owner/manager can manage their liability and discharge their duty of care to a published and accepted level of tolerable risk. The 1/10,000 that is often cited. Rather than suffer me prattling on about its origins and context, what risk is, the 1/10,000 level of tolerable risk and its position in the HSE’s Tolerability of Risk (ToR) Framework, it can be seen in the first 3 pages of the QTRA Practice Note, which you can be downloaded here; Quantified Tree Risk Assessment The 2007 Goode Judgment from South Australia is interesting for a number of reasons. Not least because the Commissioner criticizes the 'implied precision' of the QTRA risk output and then bizarrely uses QTRA to back up his opinion but makes a fundamental error in applying it. The ‘implied precision’ issue he raised had been dealt with before the Judgment, and the Risk of Harm (RoH) output should not have been expressed with such precision to 4 significant figures, no matter how precise any of the inputs. The RoH with QTRA is expressed to one significant figure only. Mike Ellison did a presentation that included a detailed reply about the Goode judgment which I've extracted and attached. I’ll get onto the ‘subjectivity’ issue tomorrow. Anything to ask in the meantime, then please fire away. Cheers Acer ventura Goode Judgment Comments - Mike Ellison.pdf