Mike Vandeman wrote:

On Fri, 29 Dec 2006 19:58:21 GMT, Michael Halliwell wrote:



Gee Mike.... You didn't answer my question the last time you tried posting this opinion paper.....here, let me paraphrase where we left off (and this is just on Wilson and Seney): In early December 2006 on alt.mountain-bike I posted...



Don't YOU read the reports you claim are "junk science"? Or maybe you are intentionally leaving out the full quote of Wilson and Seney:



“The initial regression results were not very encouraging in that none of the relationships between water runoff and soil texture, slope, antecedent soil moisture, trail roughness, and soil resistance was statistically significant. The switch to multiple regression and the inclusion of soil texture as a series of indicator variables improved the model performance.”



But not enough to make the measure of erosion VALID: " Water run-off (9%) was one of three variables that made smaller contributions.” 9% is too small to validate the measure.







and later when discussing the multiple regression model: “…ten independent variables and cross-products combined to explain 70% of the variability in sediment yield. Treating the cumulative contributions of the different variables to the final result as a rough guide to their contributions confirmed that soil texture (37%), slope (35%) and user treatment (35%) had the most impact. Water run-off (9%) was one of three variables that made smaller contributions.” Or did the fact that it was the initial model that had the poor fit and didn't account for slope, etc. which was corrected by using a different model escape you?



Nope. The measure of erosion is STILL not valid. It wasn't "corrected". It was only "improved". 9% is still a very poor performance.


Ummm...Mike...their model correlates 70% which is significant....but the run-off only contributed a small portion of that...granted, Wilson and Seney do state that thier test apparatus could be improved...but in order to get uniform conditions to isolate the variables being tested, you are forced to use a manmade rain source. But, you have yet to provide anything other that statement of your OPINION of what is "obvious" to counter or invalidate their results.  As an environmental engineer, I can say it is obvious, from both personal observation and geotechnical science, that soil texture/type, slope and force application will have a greater impact than run-off, especially in cohesive soils where run-off does not penetrate far into the soil, which is consistent with the findings of Wilson and Seney. 






Michael J. Vandeman replied:



If water run-off had only a 9% correlation with the measure of erosion, it was obviously NOT a valid measure of erosion. QED



To which I replied:



You have no research (including of your own) to prove this



assertation.



My Ph.D., you forgot, is in PSYCHOMETRICS. In other words, this area is my specialty. A 9% correlation is not significant, proving that the measure of erosion is not VALID. A VALID measure of erosion should correlate highly with slope.

Mike, their model correlated 70% with the sediment yield. From there, they attempted to apportion it to sources...they are not directly trying to correlate run-ff with the sediment yield.  Read it again:


"Treating the cumulative contributions of the different variables to the final result as a rough guide to their contributions confirmed that soil texture (37%), slope (35%) and user treatment (35%) had the most impact. Water run-off (9%) was one of three variables that made smaller contributions.” (emphasis added)



Please cite your source. Don't quote your "literature



review" as your arguement there >was on the initial model not being able to correlate slope, not on the multiple regression model only indicating run-off having a 10% correlation (vs higher values >for slope, soil texture and user treatment). Also, don't forget that user treatment (hiking vs mt. biking) has been your soap box for 12 years. Would you care to actually respond this time?



Besides, it is obvious on the fact of it that their measure of erosion (run-off from artificial rain) is not a valid measure of erosion.

Your statements here suggest that you believe run-off has a higher impact on soil erosion. Fair enough..their test apparatus may not have been sufficient to get a full handle on the "three variable that made smaller contributions." Wilson and Seney did note this limitation to their research and stated that the energy imparted in their artifical system may not be typical of natural rainfall. If you want a test in "natural conditions" either state where on the Earth there is consistent, uniform natural rain to provide uniform test conditions or create your own testing apparatus that counteracts this deficiency, re-run the test and publish the results.

In spite of the fact they had poor overall correlation with runoff (which will also be influenced by soil texture and slope, variables with significantly larger contributions) do you seriously expect run-off to have a contribution greater than user application, slope and and soil texture? If so, that would appear to contradict your statements of the last 12 years where you have claimed that user application is the most important factor. If not, than you are arguing the point for no reason.

Michael Halliwell