127
In some cases, such as the Clark-type polarographic oxygen electrode, instrument
drift is intrinsic to its operation. Redundant instrument deployment is always recommended to more readily detect sensor drift and to allow for failure of one
component.
Profiling platforms pose special challenges to sensor systems due to abrupt
gradients across physical, chemical, and biological discontinuities, thermoclines,
haloclines, pycnoclines, chemoclines, and other gradients. Many instruments are
thermosensitive such that their signal varies with temperature, requiring dedicated
adjustment algorithms. Derived variables such as salinity (computed from temperature and electrical conductivity) are especially troublesome since thermal mass of
the conductivity cell causes a lag in cell temperature relative to that sensed by the
rapid response thermistor. Furthermore, the slower response time of the conductivity cell causes additional lag relative to the thermistor when traversing abrupt gradients. Manufacturers commonly provide software that can be programmed to carry
out corrections to address these issues. Appropriate offsets depend on vertical
vehicle velocity as well as relative sensor position on the vehicle and must be determined through testing.
Quality control consists of automated analysis of the data stream to detect flaws
in transmission or instrument performance. IOOS protocols for quality assurance of
real-time oceanographic data (QUARTOD) prescribe rigorous data testing prior to
release. Thirteen standard tests are indicated for QUARTOD Q/C of T/S data (U.S.
Integrated Ocean Observing System 2015a) provided as an example in the list
below:
1. Time/gap test
2. Syntax test
3. Location test
4. Gross range test
5. Climatological test
6. Spike test
7. Rate of change test
8. Flat line test
9. Multivariate test
10. Attenuated signal test
11. Neighbor test
12. TS curve/space test
13. Density inversion test
Tests 1 and 2 address the quality of data transmission controlling for the arrival of
the expected data package in the format required. Test 3 checks for proper georeferencing. Tests 4 and 5 ensure that the data received do not exceed the native instrument range or climatologically plausible values as determined by the operator. Tests
6 and 7 provide measures of proper sensor operation and require some form of user
input with criteria based on climatological or other data. Test 8, like the eponymous
movie, ensures that the sensor is not “dead.” Test 10 can serve to detect fouled or
obstructed sensors. Test 11 relies on a comparable neighboring instrument but additional instrument-specific tests may be recommended. The above tests are widely
7.2 Quality Assurance and Quality Control (QA/QC) for In Situ Ocean Observing Data
In some cases, such as the Clark-type polarographic oxygen electrode, instrument
drift is intrinsic to its operation. Redundant instrument deployment is always recommended to more readily detect sensor drift and to allow for failure of one
component.
Profiling platforms pose special challenges to sensor systems due to abrupt
gradients across physical, chemical, and biological discontinuities, thermoclines,
haloclines, pycnoclines, chemoclines, and other gradients. Many instruments are
thermosensitive such that their signal varies with temperature, requiring dedicated
adjustment algorithms. Derived variables such as salinity (computed from temperature and electrical conductivity) are especially troublesome since thermal mass of
the conductivity cell causes a lag in cell temperature relative to that sensed by the
rapid response thermistor. Furthermore, the slower response time of the conductivity cell causes additional lag relative to the thermistor when traversing abrupt gradients. Manufacturers commonly provide software that can be programmed to carry
out corrections to address these issues. Appropriate offsets depend on vertical
vehicle velocity as well as relative sensor position on the vehicle and must be determined through testing.
Quality control consists of automated analysis of the data stream to detect flaws
in transmission or instrument performance. IOOS protocols for quality assurance of
real-time oceanographic data (QUARTOD) prescribe rigorous data testing prior to
release. Thirteen standard tests are indicated for QUARTOD Q/C of T/S data (U.S.
Integrated Ocean Observing System 2015a) provided as an example in the list
below:
1. Time/gap test
2. Syntax test
3. Location test
4. Gross range test
5. Climatological test
6. Spike test
7. Rate of change test
8. Flat line test
9. Multivariate test
10. Attenuated signal test
11. Neighbor test
12. TS curve/space test
13. Density inversion test
Tests 1 and 2 address the quality of data transmission controlling for the arrival of
the expected data package in the format required. Test 3 checks for proper georeferencing. Tests 4 and 5 ensure that the data received do not exceed the native instrument range or climatologically plausible values as determined by the operator. Tests
6 and 7 provide measures of proper sensor operation and require some form of user
input with criteria based on climatological or other data. Test 8, like the eponymous
movie, ensures that the sensor is not “dead.” Test 10 can serve to detect fouled or
obstructed sensors. Test 11 relies on a comparable neighboring instrument but additional instrument-specific tests may be recommended. The above tests are widely
7.2 Quality Assurance and Quality Control (QA/QC) for In Situ Ocean Observing Data