Performance Verification Statement for the Sunburst SAMI-pH Sensor.

Abstract

The Alliance for Coastal Technology (ACT) conducted a sensor verification study of in situ pH sensors during 2013 and 2014 to characterize performance measures of accuracy and reliability in a series of controlled laboratory studies and field mooring tests in diverse coastal environments. A ten week long laboratory study was conducted at the Hawaii Institute of Marine Biology and involved week long exposures at a full range of temperature and salinity conditions. Tests were conducted at three fixed salinity levels (0.03, 22, 35) at each of three fixed temperatures (10, 20, 30 oC). Ambient pH in the test tank was allowed to vary naturally over the first five days. On the sixth day the pH was rapidly modified using acid/base additions to compare accuracy over an extended range and during rapid changes. On the seventh day the temperature was rapidly shifted to the next test condition. On the tenth week a repeated seawater trial was conducted for two days while the temperature was varied slowly over the 10 – 30 oC range. Four field-mooring tests were conducted to examine the ability of test instruments to consistently track natural changes in pH over extended deployments of 4-8 weeks. Deployments were conducted at: Moss Landing Harbor, CA; Kaneohe Bay, HI; Chesapeake Bay, MD; and Lake Michigan, MI. Instrument performance was evaluated against reference samples collected and analyzed on site by ACT staff using the spectrophotometric dye technique following the methods of Yao and Byrne (2001) and Liu et al. (2011). A total of 263 reference samples were collected during the laboratory tests and between 84 – 107 reference samples were collected for each mooring test. This document presents the results of the Sunburst SAMI-pH which measures pHT spectrophotometrically using a dye-based colorimetric reaction. The SAMI-pH operated continuously throughout the entire ten week lab test and generated 3140 useable measurements out of a total possibility of 3154 timepoints. The total range of pH measured by the SAMI-pH was 6.273 to 8.441, compared to the range of our discrete reference samples of 6.943 to 8.502.The SAMI-pH showed some differences in accuracy across the various test solutions and pH ranges, as would be expected based on the specific setup of the instrument in terms of the indicator dye used and its known operational range. For the complete Lab test, the mean of the differences between the SAMI-pH measurement and reference pH was -0.017 ±0.970 (N=263), with a total range of -1.159 to 0.140. The larger offsets only occurred during the rapid pH shifts (see Appendix 1) and comparisons for only the stable testing conditions had a mean difference from reference measurements of -0.010 ±0.058 (N=193), with a total range of -0.048 to 0.063. Instrument measurements conducted at ten weeks with the second seawater trial exhibited a similar level of accuracy with a mean difference from reference of -0.010 ± 0.033 (N=8) compared to 0.014 ± 0.003 (N=27) for measurements in the first week of the lab test. At Moss Landing Harbor the field deployment test was conducted over 28 days with a mean temperature and salinity of 16.6 oC and 33. The measured pH range from our 84 discrete reference samples was 7.933 – 8.077 compared to a range of 7.798 – 8.150 reported by the SAMI-pH for 2564 observations conducted continuously at 15 minute intervals. The average and standard deviation of the measurement difference between the SAMI-pH and reference samples over the total deployment was 0.039 ±0.003 with a total range of -0.123 to 0.158. The useable data return for the deployment was 100%. At Kaneohe Bay the field deployment test was conducted over 88 days with a mean temperature and salinity of 24.5 oC and 34.4. The measured pH range from our 101 discrete reference samples was 7.814 – 8.084 compared to a range of 7.678 – 8.220 reported by the SAMI-pH for 4211 observations conducted continuously at 30 minute intervals. The average and standard deviation of the measurement difference between the SAMI-pH and reference samples over the total deployment was -0.014 ±0.016 with a total range in the differences of 0.068 to 0.025. The useable data return for the deployment was 100%. At Chesapeake Bay the field deployment test was conducted over 30 days with a mean temperature and salinity of 5.9 oC and 12.8. The measured pH range from our 107 discrete reference samples was 8.024 – 8.403 compared to a range of 8.121 – 8.479 reported by the SAMI-pH for 1842 observations conducted continuously at 15 minute intervals during the first 20 days of operations. Following this date, the instrument reported 175 non-determined values and exhibited a noticeable increased offset so the remaining data was considered compromised. The average and standard deviation of the measurement difference between the SAMI-pH and reference samples for the period up to April 1 was 0.041 ±0.033 (n=69), with the total range of differences between -0.086 to 0.114. The reported data return for the deployment was 94% but again only 69% of the data was considered fully reliable. At Lake Michigan the field deployment test was conducted over 29 days with a mean temperature and salinity of 21.2 oC and 0.03. Two SAMI-pH units were deployed at this test site, one equipped with MetaCresol Purple (mCP) as the indicator dye and one with Phenol Red (PR) as the indicator dye. The Phenol Red method was considered to be experimental for the company at the time and was being tested for freshwater applications. The measured range in ambient pH for the 98 reference samples was 8.013 – 8.526. The original range in pH reported by the SAMI-pH (PR) was 6.606 – 9.305 for 2564 observations conducted continuously at 15 minute intervals. After removing fifty-one observations that were identified as outliers based on a difference between the paired SAMI-pH units of more than two times the mean, the measured range was reduced to 7.658 – 8.687. The average and standard deviation of the measurement difference between the SAMI-pH (PR) and reference samples for the edited dataset was 0.021 ±0.047 with a range of -0.095 to 0.177. The omitted data represents 1.7 % of the returned data but represents a more realistic summary of the expected accuracy and uncertainty for the technology. The measured range of the SAMI-pH (mCP) samples was 7.813 – 8.405 based on 2665 observations conducted continuously at 15 minute intervals. The average and standard deviation of the measurement difference between the SAMI-pH (mCP) and reference samples over the total deployment was range was -0.162 ±0.045 and the total range of the differences was between -0.333 to -0.068. The useable data return for this unit was 100%. A summary plot of the SAMI-pH versus dye reference pH for all sites indicated that the SAMI-pH responded consistently among all test sites which covered a salinity range of freshwater to full seawater and temperatures from 3-26 oC. As expected there was a consistent offset in results reported within the Great Lakes test for the SAMI-pH with MetaCresol Purple indicator dye compared with the unit operating with the Phenol Red dye indicator. The SAMIpH (PR) results closely match the reference samples since ACT also used Phenol Red as the indicator dye for their analysis. Lastly, it is worth emphasizing that the continuous 15 – 30 minute time-series provided by the test instrument was able to resolve a significantly greater dynamic range and temporal resolution than could be obtained from discrete reference samples. There were no obvious changes in the differences between instrument and reference measurements during the duration of the mooring test, indicating that biofouling and instrument drift had not significantly affected measurement performance. Continuous in situ monitoring technologies, such as the Sunburst SAMI-pH, provide critical research and monitoring capabilities for helping to understand and manage important environmental processes such as carbonate chemistry and ocean acidification, as well as numerous other environmental or industrial applications.