1d ago

3 Views

0 Downloads

1.07 MB

10 Pages

Transcription

Appendix 21. Laboratory Comparison between YSI EXO andYSI 6136 Turbidity Sensors using pink clay at the KansasWater Science Center Laboratory, Lawrence, KansasComparison DescriptionStation name: Kansas Water Science Center Lab, Lawrence, Kansas.Equipment: A Yellow Springs Instrument (YSI) EXO water-quality monitor equipped with a YSIEXO turbidity sensor and a YSI 6 series equipped with a YSI 6136 turbidity sensor were deployedin a laboratory turbidity testing apparatus for comparison between the two sensors.(See“Performance Evaluation Tests,” “Laboratory Tests,” p. 7 of main report, for a full description oflaboratory methods.) The Hach model 2100AN laboratory turbidimeter with a flow-through cellwas used as a reference to measure the turbidity in the apparatus bucket every 15 minutesbefore adding more sediment. No datum corrections were applied to either dataset.Testing material and water: Pink clay and deionized water.Calibration standard used: Hach StablCal standards.Laboratory comparison date: February 15, 2017.DatasetsAll data were collected using USGS protocols (U.S. Geological Survey, variously dated) and arepublished in King (2021). Data were edited to remove periods where material was added to thetesting apparatus, leaving the steady-state data for analysis.21.1

Time Series21.2

Statistical Analyses - YSI EXO and YSI 6136 DataSlope comparisonThe following is a summary of final regression analysis for sensor-measured turbidity from a YSIEXO turbidity sensor and a YSI 6136 turbidity sensor at the Kansas Water Science Centerlaboratory, Lawrence, Kansas, on February 8, 2017; the data used in the final regressions wereaverages of turbidity for each step, each of which had a duration of approximately 15 minutesonce the sensor had stabilized:y 1.26x – 25.44wherey turbidity measured with YSI 6136 turbidity sensor (FNU)x turbidity measured with YSI EXO turbidity sensor (FNU).21.3

Paired t-test for YSI EXO and YSI 6136 DataSigmaPlot Statistical Output:Normality Test (Shapiro-Wilk): Passed (P 0.412)Paired t-test:Treatment NameYSI EXOYSI .804Std Dev277.873349.91473.817SEM105.026132.25527.900t -4.007 with 6 degrees of freedom.95 percent two-tailed confidence interval for difference of means: -180.073 to -43.535Two-tailed P-value 0.00706The change that occurred with the treatment is greater than would be expected by chance; there is a statisticallysignificant change (P 0.007)One-tailed P-value 0.0035321.4

The sample mean of treatment YSI 6136 exceeds the sample mean of treatment YSI EXO by an amount that isgreater than would be expected by chance, rejecting the hypothesis that the population mean of treatment YSI EXOis greater than or equal to the population mean of treatment YSI 6136. (P 0.007)Power of performed two-tailed test with alpha 0.050: 0.913Power of performed one-tailed test with alpha 0.050: 0.969Summary of ResultsThere is a strong linear association between measurements made with the two sensors (R 1.00). Relative percentage difference ranged from 14 to 24 percent (median: 18 percent; mean:18 percent). The data passed the Shapiro-Wilk test for normality (P 0.412); therefore, a pairedt-test was performed. The difference between mean values for the YSI EXO and YSI 6136turbidity sensors was statistically significant (P 0.05).Statistical Analyses - YSI EXO and Hach 2100AN DataSlope comparisonThe following is a summary of final regression analysis for sensor-measured turbidity from a YSIEXO turbidity sensor and a Hach 2100AN laboratory turbidimeter at the Kansas Water ScienceCenter laboratory, Lawrence, Kansas, on February 8, 2017; the data used in the final regressionswere averages of turbidity for each step, each of which had a duration of approximately 15minutes once the sensor had stabilized:y 0.50x 110.91wherey turbidity measured with YSI EXO turbidity sensor (FNU)x turbidity measured with Hach 2100AN turbidimeter (FNU).21.5

21.6

Paired t-test for YSI EXO and Hach 2100AN DataSigmaPlot Statistical Output:Normality Test (Shapiro-Wilk): Passed (P 0.525)Paired t-test:Treatment NameYSI EXOHach 19.260Std Dev277.873556.178283.224SEM105.026210.215107.049t -2.982 with 6 degrees of freedom.95 percent two-tailed confidence interval for difference of means: -581.198 to -57.322Two-tailed P-value 0.0246The change that occurred with the treatment is greater than would be expected by chance; there is a statisticallysignificant change (P 0.025)One-tailed P-value 0.0123The sample mean of treatment Hach 2100AN exceeds the sample mean of treatment YSI EXO by an amount that isgreater than would be expected by chance, rejecting the hypothesis that the population mean of treatment YSI EXOis greater than or equal to the population mean of treatment Hach 2100AN. (P 0.025)Power of performed two-tailed test with alpha 0.050: 0.70Power of performed one-tailed test with alpha 0.050: 0.837Summary of ResultsThere is a strong linear association between measurements made with the two sensors (R 0.99). Relative percentage difference ranged from 13 to 60 percent (median: 37 percent; mean:37 percent). The data passed the Shapiro-Wilk test for normality (P 0.525); therefore, a pairedt-test was performed. The difference between mean values for the YSI EXO sensor and Hach2100AN was statistically significant (P 0.05).Statistical Analyses - YSI 6136 and Hach 2100AN DataSlope comparisonThe following is a summary of final regression analysis for sensor-measured turbidity from a YSI6136 turbidity sensor and a Hach 2100AN laboratory turbidimeter at the Kansas Water ScienceCenter laboratory, Lawrence, Kansas, on February 8, 2017; the data used in the final regressionswere averages of turbidity for each step, each of which had a duration of approximately 15minutes once the sensor had stabilized:y 0.62x 112.04where21.7

y turbidity measured with YSI 6136 turbidity sensor (FNU)x turbidity measured with Hach 2100AN turbidimeter (FNU).21.8

Paired t-test for YSI 6136 and Hach 2100AN DataSigmaPlot Statistical Output:Normality Test (Shapiro-Wilk): Passed (P 0.413)Paired t-test:Treatment NameYSI 6136Hach 07.456Std Dev349.914556.178212.331SEM132.255210.21580.254t -2.585 with 6 degrees of freedom.95 percent two-tailed confidence interval for difference of means: -403.829 to -11.082Two-tailed P-value 0.0415The change that occurred with the treatment is greater than would be expected by chance; there is a statisticallysignificant change (P 0.041)One-tailed P-value 0.0207The sample mean of treatment Hach 2100AN exceeds the sample mean of treatment YSI 6136 by an amount that isgreater than would be expected by chance, rejecting the hypothesis that the population mean of treatment YSI 6136is greater than or equal to the population mean of treatment Hach 2100AN. (P 0.041)21.9

Power of performed two-tailed test with alpha 0.050: 0.581Power of performed one-tailed test with alpha 0.050: 0.737Summary of ResultsThere is a strong linear association between measurements made with the two sensors (R 0.99). Relative percentage difference ranged from 3 to 41 percent (median: 19 percent; mean:21 percent). The data passed the Shapiro-Wilk test for normality (P 0.413); therefore, a pairedt-test was performed. The difference between mean values for the YSI 6136 sensor and Hach2100AN was statistically significant (P 0.05).Selected ReferencesCleveland, W.S., 1979, Robust locally weighted regression and smoothing scatterplots: Journalof the American Statistical Association, v. 74, no. 368, p. 829–836.Helsel, D.R., and Hirsch, R.M., 2002, Statistical methods in water resources—Hydrologicanalysis and interpretation: U.S. Geological Survey Techniques of Water-ResourcesInvestigations, book 4, chap. A3, 522 p. [Also available at https://doi.org/10.3133/twri04A3.]King, L.R., 2021, Laboratory and field data for selected turbidity standard and sensorcomparisons, October 2014 to September 2017: U.S. Geological Survey Data Release,https://doi.org/10.5066/P9EVSDHH.U.S. Geological Survey, variously dated, The national field manual for the collection of waterquality data: U.S. Geological Survey Techniques and Methods, book 9, chaps A1–A10. [Alsoavailable at https://water.usgs.gov/owq/FieldManual/.]21.10

Statistical Analyses - YSI EXO and YSI 6136 Data Slope comparison The following is a summary of final regression analysis for sensor-measured turbidity from a YSI EXO turbidity sensor and a YSI 6136 turbidity sensor at the Kansas Water Science Center laboratory, Lawrence, Kansas, on Febr