1、A comparison of filter pads for the analysis of total suspended solids,2 June 2009,Analytical Methods and Quality Assurance Workgroup,Bill Romano MD Dept. of Natural ResourcesElgin S. Perry, Ph.D. Statistics consultant,Current procedures,DHMH and DCLS analyze TSS using a pre-rinsed Whatman GF/F 1.5
2、micron pore size filter pad, as outlined in Standard Methods CBL uses an un-rinsed Whatman GF/F 0.7 micron filter pad and reports the mean of two pads ODU also uses a Whatman GF/F 0.7 micron pad, pre-rinses, and has pre-combusted the pads since January 2002,CBL comparison of un-rinsed and rinsed GF/
3、F 0.7 pads,CBL weighed 100 pads straight from the box The pads were then rinsed with three successive 20 ml rinses of deionized water, dried overnight in a 100 degree Centigrade oven, and re-weighed It took over one hour and six liters of high quality deionized water to rinse the pads 150 pads are n
4、eeded for each main Bay cruise,CBL comparison continued,The difference in weights between un-rinsed and rinsed pads was tested using the Wilcoxon signed rank test The p-value of 0.2655 was not significant The arithmetic mean difference between un-rinsed and rinsed pads was -0.000014 grams (rinsed mi
5、nus un-rinsed) The CBL balance can only be accurately read to 0.0001 grams,Statistical mumbo jumbo,The observed instrument precision was compared to the analytical precision by calculating the variance of the differences between the un-rinsed and rinsed pads The calculated variance was divided by tw
6、o because the variance is additive The standard deviation of 0.0002 is very close to the analytical precision of 0.0001 There is no statistical or practical difference between un-rinsed and rinsed pads,Why use a smaller pore size pad,CBL started analyzing main Bay water samples in the mid-1980s The
7、designers of the main Bay monitoring program were primarily interested in the dissolved fraction of nutrients A smaller pore size pad does a better job of separating particulate and dissolved fractions of a water sample,Particulate,Dissolved,The CBL repeated measures study,Carl collected a carboy of
8、 water from the CBL pass-through sea water system Ten samples were drawn from the carboy for “high level” TSS analysis, then split for analysis using un-rinsed 0.7 micron Whatman pads and pre-rinsed, pre-weighed 1.5 micron Environmental Express pads The remaining water was diluted and eight samples
9、were drawn for “low level” TSS analysis,Repeated measures results,Statistical mumbo jumbo II,Although a statistical test was probably not needed to see that the TSS concentrations differ based on filter pad pore size, I did one anyway using SAS PROC ANOVA and the following model statement: log10TSS
10、= pad level pad*level The p-value for pad was 0.0006 and the p-value for level was 0.0001,Putting the difference in perspective,CBL analyzes two TSS pads and reports the average The variability between the CBL pad replicates and the repeated measures study was tested after controlling for variation
11、in the replicate pad data Mean square errors were used to calculate an F-statistic, which was significant at p0.0001 The variability between the replicate pads is greater than the variability between the different pore size pads,A long time ago in a Bay far, far away,DHMH and CBL analyzed split TSS
12、samples collected during December 1999 and May 2000 main Bay cruises The means for DHMH (1.5 ) and CBL (0.7 ) were 10.42 and 10.72 mg/L, respectively for the May 1999 data Although there is no difference between the concentrations, the p-value from the Wilcoxon signed rank test was significant (p=0.
13、0081) This was probably a Type I error,Quit while youre ahead,The May 2000 data have a different pattern than the December 1999 data The means for DHMH (1.5 ) and CBL (0.7 ) were 9.18 and 11.18 mg/L, respectively These results show that the smaller the pore size the higher the TSS concentration The
14、Wilcoxon signed rank test indicates the difference between the pads is significant (p0.0001) These results include possible differences in laboratory handling,Conclusions,There is no meaningful difference between un-rinsed and rinsed Whatman GF/F 0.7 pore size pads The method for analyzing TSS is ol
15、d and maybe back then filter pads were not of as high a quality as they are now and they had to be pre-rinsed There is strong evidence that changing pore size pads will introduce a step trend in the data Decreasing the number of pads CBL analyzes would represent a savings in time and money Although measurement error would increase, it is probably a small fraction of the overall error,Questions or comments,Photo by William Frece,
