Overview


       This document is intended to examine if dissolved Aluminum (test codes 01106A and 01106H) can be subsituted for monomeric Alumninum (test code 01106D).

       Monomeric Al is a dissolved species of aluminum that is filtered through a 0.1 micron filter. Dissolved aluminum concentrations are also derived after filtering raw water samples; however the pore size of the filters is larger (size TBD).

       Specifically, the following objectives will be addressed:

  1. How many samples have included both monomeric and dissolved aluminum? Is this sample size adequate to draw inferences?

  2. What is the range and central tendency of differences in results for these paired samples?

  3. Are the mean differences significantly different from 0?



SIS Query in R


       All sample results that included one or more of the monomeric or dissolved aluminum test codes were queried from SIS using R. The query resulted in 33,736 observations from 30,926 unique samples. The vast majority of the observations were dissolved aluminum. Only 426 monomeric aluminum samples have been collected statewide to date (Table 1).


       Of note, 67% of the observations that resulted from this query were not associated with an NHD flowline and 39% of observations did not have any coordinates associated with the sample.


Table 1. Number of observations queried from SIS, by test description and test code.
Parameter 01106A 01106H 01106D
Al Diss 0.1u (monomeric) 426
ALUMINUM D (dissolved) 17248 16062



Data Reduction


       Next, samples were identified where dissolved and monomeric aluminum were collected together. After blanks were removed, this resulted in only 14 samples where both were collected together. Detection limits were consistent at 10 ug/L for each parameter across all samples. Differences in concentration were calculated between dissolved and monomeric aluminum in paired samples. Differences ranged from -20 ug/L (monomeric lower) to 60 ug/L (monomeric higher) (Table 2).


Table 2. Sample information, location, and concentrations of monomeric and dissolved aluminum in paired samples (n=14).
SAMPLE_ID DATE_COLLECTED MONITORING_POINT_NAME MONITORING_POINT_ALIAS_ID monomeric_Al_ug_l dissolved_Al_ug_l diff_ug_l diff_ug_l_abs
2023565 2016-03-17 14:34:00 Hyner Run WQN0464 70.00 10.000 60.000 60.000
2284193 2019-03-11 11:30:00 2030.00 2000.000 30.000 30.000
2029613 2016-04-07 07:26:00 Hyner Run WQN0464 18.40 10.000 8.400 8.400
2023568 2016-03-18 08:21:00 Carbaugh Run WQN0465 37.30 35.000 2.300 2.300
2030200 2016-04-08 10:00:00 Carbaugh Run WQN0465 34.40 34.300 0.100 0.100
2023566 2016-03-17 14:34:00 Hyner Run WQN0464 10.00 10.000 0.000 0.000
2029612 2016-04-07 07:26:00 Hyner Run WQN0464 10.00 10.000 0.000 0.000
2028826 2016-03-17 09:41:00 Jones Mill Run WQN0734 19.80 20.300 -0.500 0.500
2023564 2016-03-17 10:41:00 Rock Run Lycoming County WQN0463 27.70 30.400 -2.700 2.700
2283995 2019-03-11 15:00:00 54.90 58.800 -3.900 3.900
2283997 2019-03-12 08:15:00 28.50 34.200 -5.700 5.700
2283994 2019-03-11 14:40:00 21.50 28.900 -7.400 7.400
2109809 2017-02-08 11:45:00 Jones Mill Run WQN0734 36.86 55.421 -18.561 18.561
2283996 2019-03-12 07:50:00 181.00 201.000 -20.000 20.000



Visualization


       Most paired samples had differences between -5 to 5 ug/L. There did not appear to be a consistent difference in the directionality of concentration differences. Monomeric aluminum was both greater than and less than dissolved aluminum in the small dataset (Figure 1).


Figure 1. Histogram of differences in monomeric and dissolved aluminum concentrations.


       Viewed slightly differently, the concentrations of these related parameters appear to be comparable, except for two instances where monomeric is clearly higher than dissolved (Figure 2).


Figure 2. Boxplot of monomeric and dissolved aluminum concentrations. The individual observations are overlaid as points, and lines connect the paired samples.


       The final visualization of the dataset shows the differences in concentrations at a sample level.


Figure 3. Points indicate observed concentrations of monomeric and dissolved aluminum in each sample. Bars indicate differences (in ug/L) between the parameters at a sample level. The text overlaid on each bar quanitifies the difference (in ug/L).



Statistical anlalysis


       A paired sample t-test is a statistical procedure used to determine whether the mean difference between two sets of paired observations is zero. This paired sample t-test was applied to the 14 paired samples to determine whether the mean difference was significantly different from 0. The mean difference in concentrations in this dataset was found to be ~ 3.0 ug/L, which was not significantly different than 0 (P = 0.588).

## 
##  Paired t-test
## 
## data:  diff_df$monomeric_Al_ug_l and diff_df$dissolved_Al_ug_l
## t = 0.55623, df = 13, p-value = 0.5875
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -8.659926 14.665497
## sample estimates:
## mean of the differences 
##                3.002786



Implications


  • Although differences in concentrations in monomeric and dissolved aluminum were not significant, the sample size was exceedingly small (n=14).
  • Additional paired samples should be collected in the future, and a similar analysis repeated to increase confidence in results.
    • Monomeric and dissolved aluminum should be collected together in a variety of settings, including AMD, suspected acid deposition affected sites, and in settings where AMD and acid dep are not present. Settings should also span a gradient of stream size.
  • Instruction on correct collection method and supplies necessary should be provided to staff willing to collect samples.
  • Increased collected of paired monomeric and dissolved aluminum samples will help to elucidate whether or not dissolved can be substitued for monomeric aluminum when the former has not been collected. This could potentially benefit impairment decision making and source/cause determinations.