REV94025 - Laserfiche WebLink

Home Browse Search

" `~~ STATE OF COLORADO DIVISION OF MINERALS AND GEOLOGY Department of Natural Resources ~13736herman St., Room 215 ~ ~ COLORADO Denver, COlorad0 80203 DIVISION O F Phone: (303) 866-3567 M I N E~ RA L S FAX: (303) 832-8106 G E O L O G Y ~. REC LAN ATION•NINING SAFETT•SCIENCE DATE: August 16, 2004 BiIIITvens Governor T0: Janet Binns ~~ Russell George Executive Direttor FROM: Dan Mathews Ronald W. Canany Division Direttor RE: Trapper 2004 Cropland Production Data Natural Resource Trustee Janet, as you know, Forrest Luke had provided the 2004 production data from the cropland reclamation parcel and the associated reference area, due to the problem they had with-the reference area sampling. Due to high variability (apparently associated with a late frost and very dry conditions), they were unable to meet sample adequacy after running 107 transects (with three quadrats per trensect) in the reference area. We suggested a two sample, one sided, reverse null success demonstration approach as an alternative. ~, L J Forrest provided the data in EXCEL, along with the results of a two sample, one sided, reverse null confidence limit comparison. Iran the same confidence limit comparison on the data and confirmed the operator's results. The reclaimed area lower 90% confidence limit exceeds the reference area upper 90% confidence limit by a wide margin, indicating that the means are different, and that the reclaimed area mean exceeds the reference area standard. I also ran a two sample reverse null t-test (since the t-test approach is specifically referenced in the proposed new regulations). Because sample adequacy was not met in the reference area, it was necessary to employ a two sample t-test, rather than the one sample test that we typically accept for reference area comparison success demonstration. Determination of the appropriate standard error term and degrees of freedom for use in the denominator of the t-test is more complicated with the two sample test. As a first step, it is necessary to test for homogeneity of variances to determine whether a pooled variance is appropriate, or whether the variance and degrees of freedom need to be approximated. I used Levene's test as recommended by McDonald et.al., 2003 (two sided test, alpha = .O1). Result indicated that the variances were not equivalent, and as a result I applied Satterthwaite's Approximation, as further recommended by McDonald et. al. Based on this procedure, appropriate degrees of freedom was indicated to be 32.7, and the Satterthwaite Standard Error was 13.4. The two-sample t-test incorporating these values confirmed the results of the confidence limit comparison; the reclaimed area mean exceeds the reference area standard (90% of the mean) at the 90°k level of statistical confidence. My calculations are attached. C: Sandy Brown Forrest Luke, Trapper Mine Office of Office of Colorado