January 07, 2008: The "Extended Column Test"
The promotion of this "test" as something useful in decision making in the field has bothered me for quite a while. Here is the short version of my concerns, as well as a suggestion for what to ask anyone promoting this.
[[ Note: This is being reposted on this blog in December 2014, about 7 years after it first appeared on a blog that has been discontinued. I have not followed the technical literature over this period very closely and more work has been done. However, my fundamental concerns remain. Especially for use outside a large data collection operation. (I have reservations about what I consider an over reliance on stability tests in general outside of such programs.) At the end I have added a few comments as of reposting this at the end of 2014. ]]
First, what does it supposedly tell us and is that useful? Well, it's not clear that it tells us whether weak layers exist or how strong they are. Rutschblock and compression tests are still best for that. The Extended Column Test (ECT) supposedly tells us, given a weak layer, whether it is likely to propagate a failure or not. This presumes the existence of a weak layer where a failure can be initiated. One person wrote that they had tried this test along with traditional ones and this one would have made them less conservative in every case. Is this a surprise? You won't rule out slopes with this that wouldn't be ruled out with established tests. All you will do is justify a slope as safe, despite a weakness, based on its supposedly poor potential to propagate a failure.
Do you really want to identify a weakness then ski a slope anyway because you don't think a failure will propagate? I don't. Especially if the test is less than perfect, and all of them are.
This test was first presented at the ISSW in the fall of 2006. Also presented was a Prototype Fracture Propagation Test by Canadian researchers Gauthier and Jamieson. The ECT had been done a few hundred times, in two general locations. There were not well planned research plots to collect meaningful and controlled data. Nothing about the test had ever been published in a peer reviewed journal. I'm not sure anything has been yet. On the other hand the Canadian prototype test had been very well studied. Data had been collected over several seasons from well-planned study plots. All kinds of variations in the test configuration were studied and compared statistically. And this had resulted in numerous publications in peer reviewed technical journals.
Now, if you were inclined to attempt to measure fracture propagation propensity which test would you want to rely on?
I believe there are two reasons the ECT has been hyped in the US without much reference to the Canadian test. The first is that the Canadian test was never presented as a final test ready to use in the field. It was always preceded with the word "Prototype". The researchers working on this would not have wanted it promoted for general use at that time. In the December 2007 issue of Avalanche Review it is presented to the public in simple terms as a true test, for the first time I believe. The status of the ECT was not qualified in this manner very clearly.
The second reason is that presentations and paper on the Canadian test, until recently, were presentations on research. Involving data collection methods and plans, statistical analysis of results, and comparisons to complicated theoretical calculations on fracture mechanics. There is not much of a community of true scientists and engineers in snow science in the US. In fact, science education in the US has deteriorated in recent years. So people attending the ISSW aren't going to pay any attention to anything that technical. They are going to pay attention to something they think they can walk out the door and use. Whether they clearly understand its basis or not.
So if somebody promotes the ECT, or implies in their teaching or writing that it is important or useful, I suggest asking them about it. Ask if they've compared it to the Canadian test and if they are familiar with that work. (Most likely they will be vaguely aware of it but not very familiar with it.) Ask them why they feel the ECT can be reliably interpreted - on what basis? And ask them what they think about the number of studies and details analyzed for each test. If they can't answer these questions, or engage in an intelligent and informed technical discussion, they should not be promoting the test.
The ECT test may prove to be a viable tool. That will require significant further study. Well planned study, with results that can be analyzed with statistical significance, and with fracture mechanics theory to help explain the mechanics of the test. Until then it's simply an interesting idea.
Update - December 2014
I can think of three things to add despite not following any specific well-planned research or peer reviewed studies.
The first is that I can recall at least one incident report from a fatality where this test did not identify the appropriate weak layer at all. I believe traditional tests did, although I cannot remember for certain. While I have not followed research closely I have continued to archive thousands of incident reports and I have not seen any which this test would have prevented. Very few would be prevented by stability tests in general - most came down to poor planning, followed by poor decisions in the field.
The second is that I do recall one short paper from the Swiss. They concluded that this test was slightly more accurate but was also harder to perform properly. Is the slight increase in accurate results really gained in general use, or is the test frequently not performed exactly as required? This paper also did not address the third issue because the depth and nature of the weak layers was not mentioned.
The third comment is that when this test was newer I heard different figures for the depth to which it is useful. One was an off-the-cuff comment at a meeting by a researcher who had done this test next to others many times but only informally (and not part of any study). His impression was that below about 17 cm it became less reliable. I later heard a US avalanche specialist state a somewhat higher number. I believe it could have been as high as 50 cm, but it may have been less. Neither number is very deep, and as far as I know there has not been a study to determine the usefulness vs the depth of the layer. This most likely depends on the nature of the weak layer itself.
The other test mentioned, the Propagation Saw Test, has become well established for use in programs with databases. It was developed for deeper layers in general, and specifically for propagation without mixing it up with initiation. We have no good way of assessing how likely an initial failure is for a deep layer - which is what makes them problematic. This test can take quite a bit of time to execute, especially for deep layers. So while it can be a useful data point it is probably not particularly useful for recreation. Less so than the quicker established tests.
[[ Note: This is being reposted on this blog in December 2014, about 7 years after it first appeared on a blog that has been discontinued. I have not followed the technical literature over this period very closely and more work has been done. However, my fundamental concerns remain. Especially for use outside a large data collection operation. (I have reservations about what I consider an over reliance on stability tests in general outside of such programs.) At the end I have added a few comments as of reposting this at the end of 2014. ]]
First, what does it supposedly tell us and is that useful? Well, it's not clear that it tells us whether weak layers exist or how strong they are. Rutschblock and compression tests are still best for that. The Extended Column Test (ECT) supposedly tells us, given a weak layer, whether it is likely to propagate a failure or not. This presumes the existence of a weak layer where a failure can be initiated. One person wrote that they had tried this test along with traditional ones and this one would have made them less conservative in every case. Is this a surprise? You won't rule out slopes with this that wouldn't be ruled out with established tests. All you will do is justify a slope as safe, despite a weakness, based on its supposedly poor potential to propagate a failure.
Do you really want to identify a weakness then ski a slope anyway because you don't think a failure will propagate? I don't. Especially if the test is less than perfect, and all of them are.
This test was first presented at the ISSW in the fall of 2006. Also presented was a Prototype Fracture Propagation Test by Canadian researchers Gauthier and Jamieson. The ECT had been done a few hundred times, in two general locations. There were not well planned research plots to collect meaningful and controlled data. Nothing about the test had ever been published in a peer reviewed journal. I'm not sure anything has been yet. On the other hand the Canadian prototype test had been very well studied. Data had been collected over several seasons from well-planned study plots. All kinds of variations in the test configuration were studied and compared statistically. And this had resulted in numerous publications in peer reviewed technical journals.
Now, if you were inclined to attempt to measure fracture propagation propensity which test would you want to rely on?
I believe there are two reasons the ECT has been hyped in the US without much reference to the Canadian test. The first is that the Canadian test was never presented as a final test ready to use in the field. It was always preceded with the word "Prototype". The researchers working on this would not have wanted it promoted for general use at that time. In the December 2007 issue of Avalanche Review it is presented to the public in simple terms as a true test, for the first time I believe. The status of the ECT was not qualified in this manner very clearly.
The second reason is that presentations and paper on the Canadian test, until recently, were presentations on research. Involving data collection methods and plans, statistical analysis of results, and comparisons to complicated theoretical calculations on fracture mechanics. There is not much of a community of true scientists and engineers in snow science in the US. In fact, science education in the US has deteriorated in recent years. So people attending the ISSW aren't going to pay any attention to anything that technical. They are going to pay attention to something they think they can walk out the door and use. Whether they clearly understand its basis or not.
So if somebody promotes the ECT, or implies in their teaching or writing that it is important or useful, I suggest asking them about it. Ask if they've compared it to the Canadian test and if they are familiar with that work. (Most likely they will be vaguely aware of it but not very familiar with it.) Ask them why they feel the ECT can be reliably interpreted - on what basis? And ask them what they think about the number of studies and details analyzed for each test. If they can't answer these questions, or engage in an intelligent and informed technical discussion, they should not be promoting the test.
The ECT test may prove to be a viable tool. That will require significant further study. Well planned study, with results that can be analyzed with statistical significance, and with fracture mechanics theory to help explain the mechanics of the test. Until then it's simply an interesting idea.
Update - December 2014
I can think of three things to add despite not following any specific well-planned research or peer reviewed studies.
The first is that I can recall at least one incident report from a fatality where this test did not identify the appropriate weak layer at all. I believe traditional tests did, although I cannot remember for certain. While I have not followed research closely I have continued to archive thousands of incident reports and I have not seen any which this test would have prevented. Very few would be prevented by stability tests in general - most came down to poor planning, followed by poor decisions in the field.
The second is that I do recall one short paper from the Swiss. They concluded that this test was slightly more accurate but was also harder to perform properly. Is the slight increase in accurate results really gained in general use, or is the test frequently not performed exactly as required? This paper also did not address the third issue because the depth and nature of the weak layers was not mentioned.
The third comment is that when this test was newer I heard different figures for the depth to which it is useful. One was an off-the-cuff comment at a meeting by a researcher who had done this test next to others many times but only informally (and not part of any study). His impression was that below about 17 cm it became less reliable. I later heard a US avalanche specialist state a somewhat higher number. I believe it could have been as high as 50 cm, but it may have been less. Neither number is very deep, and as far as I know there has not been a study to determine the usefulness vs the depth of the layer. This most likely depends on the nature of the weak layer itself.
The other test mentioned, the Propagation Saw Test, has become well established for use in programs with databases. It was developed for deeper layers in general, and specifically for propagation without mixing it up with initiation. We have no good way of assessing how likely an initial failure is for a deep layer - which is what makes them problematic. This test can take quite a bit of time to execute, especially for deep layers. So while it can be a useful data point it is probably not particularly useful for recreation. Less so than the quicker established tests.
admin wrote:
"I don't see how one or two ECT tests here or there is supposed to be used in a persons final decision making process"
So if you do an ECT and get a stability failure at 6, you are still in douht whether to ski or not? The test can show you when NOT to ski but never be used as an "Lets go skiing" on its own.
Regards Espen, Norwegian Freeskier