Statistical Analysis of Left-Censored Geochemical Data
PacIOOS WQB Data QC
1. QC Challenges of Collecting
Time-Continuous Water
Quality Data:
PacIOOS Water Quality Buoy (WQB) Data QC
by
Michael Tomlinson & Eric De Carlo
2. Handout &Outline
Handout contains most data products in presentation
Overview of Pacific Islands Ocean Observing System
PacIOOS water quality components
Presentation focuses on the NOAA/PMEL MapCO2 /
PacIOOS specifically Water Quality Buoy data QC
Manual data processing & QC (different time scales)
Annual (and beyond) QC & data adjustment with
Aquarius® software
Independent data (& estimate) validation
Final data release
What have we learned from 5 years of monitoring?
5. What does
PacIOOS do?
Data Collection & QC
Data Aggregation
Data Visualization
Real-time Warnings
Provides Research Data
Outreach/Communication
6. Examples of Applications
Understanding Ocean Acidification
Assessing the effects of storms on coastal
water quality (e.g., Tomlinson et al., 2011)
Water quality monitoring & sounding “all
clear” after tsunamis (e.g., Fiedler et al., 2014)
Impacts of groundwater / surface water on
coastal marine ecosystems
12. Twice-Daily Checks (continued)
Gaps in the WQB record on
the website that did not
exist (see graph below);
caused by a problem with
the Data Turbine
17. WQB Data QC 1 Year & Beyond
Aquarius "whiteboard" used to QC and adjust, as
necessary, data from the PacIOOS WQBs. In addition to
the WQB data, other data are used as surrogates to
help QC the data and guide any necessary
adjustments. The raw, adjusted, surrogate, and discrete
(C&CH) data are plotted as time-series.
Figure 2
28. Problems Evident After 1+ Years
(SBE43® Membrane DO Sensor)
Sensor began to drift immediately after deployment and continued
throughout the entire deployment period, regardless of the time of
year. SBE conceded there was a problem but cannot explain it.
Figure 12
35. What have we learned in 5 years?
QC must be conducted at different time scales
Need one extra suite of sensors for annual servicing
Some sensors (e.g., SBE43® DO membrane sensor)
do not work well in shallow, subtropical water
Although expensive, need another suite of sensors with
comparable precision & accuracy for servicing checks
Data QC is not part-time job nor can it be automated …
… unless we emulate the National Data Buoy Center &
have redundant sensors (very expensive)
Routine monitoring provides valuable research data,
e.g., tsunami water quality effects (Fiedler et al., 2014)
36. Questions & Contact Information
www.pacioos.org
Michael Tomlinson
UHM Oceanography
Flagstaff, AZ
mtomlins@hawaii.edu
928-266-2236 (cell)