Laserfiche WebLink
Corey Heaps <br />CAM Colorado LLC <br />August 10, 2011 Page 6 <br />This document also states on page 15 that in certain situations "systematic sampling may <br />be more efficient than random sampling." This concept is seems to have been ignored by <br />the DRMS who now states that random and systematic sampling cannot not be performed <br />together as this violates the "consistency in sampling" requirement found in Rule <br />4.15.11(1) and yields biased data. Seemingly unrecognized by the DRMS is the <br />comment on page 16 relative to the "drawbacks associated with simple random <br />sampling..." Which are reported in some instances to "be less efficient than a stratified <br />design, and further, the completely random selection of sample observation points may <br />result in a distribution in which the observations are not dispersed uniformly throughout <br />the block in a manner which appears to be `representative? ?' In instances where the <br />distribution of sample observation points is skewed to such an extreme that the block as a <br />whole is not adequately represented, additional sampling may be warranted." <br />Our understanding of the DRMS regulations and all associated guideline documents <br />which the DRMS has prepared over the years, is that the vegetation baseline study <br />performed on the CAM - Colorado -LLC Fruita Unit Train Loadout are consistent with the <br />regulations and the Bond Release Guideline which specifically allows for operators to <br />"give careful attention to sampling approaches which minimize variation among sample <br />observations" by using such approaches such as "within - parcel stratification" That was <br />the exact approach used herein. In fact we submit that the approach used herein, yields <br />less variation and bias in the data that the approved approaches of "proportional <br />allocation and area weighting sampling" which are approved by the DRMS. <br />With respect to the suggestion made by the DRMS that "inconsistent" sampling methods <br />were used between the Greasewood Predisturbance and Greasewood Reference Area we <br />submit that this is not accurate. Random points were used on both sites. The DRMS <br />suggestion that the data must be sampled in a "sequential" manner "to maintain random <br />selection" or it is biased, cannot be documented in the scientific literature we have <br />consulted and discussions we have had with numerous professionals working in this field. <br />As we documented in our previous discussion relative to the first round of Adequacy <br />Review Comments and Responses, from an extremely authoritative source, the order in <br />which the random samples are collected is irrelevant, as long as they are random samples. <br />The statistical concept of "Sample Adequacy" requires that a sufficient number of <br />random sample points be collected to describe the population being sampled at the 90 <br />percent confidence interval. What is important is the number, not the order in which the <br />samples were collected. The approach being suggested by the DRMS wherein random <br />sample points must be collected and that these points must be collected in a "systematic" <br />or "sequential" order is "inconsistent" with any definition of "random" we are aware off. <br />If the regulations require that random data be collected, then why can't they be collected <br />in a random order? Either the regulatory requirement found in Rule 4.15.11 (1) wherein <br />"both random and systematic sample designs are acceptable" is correct or the DRMS <br />must now provide documentation why their concept of "random" sampling can only be <br />performed if the samples are sampled in systematic or "sequential" order "to maintain <br />