The calval package contains a collection of functions to assist with
the flagging and visualization of validation data collected during
quality assurance testing of the Centre for Marine Applied Research
(CMAR) Coastal Monitoring
Program sensors. Sensor testing occurs prior to sensor deployment to
ensure sensors are recording within an appropriate range for each
variable. Testing is repeated after sensors are retrieved from
deployment to check for sensor drift and biofouling impacts.
calval is compatible with the following sensor models:
Table 1. Compatible Sensor Models
| Sensor | Variable(s) Measured |
|---|---|
| aquaMeasure DOT | Temperature, Dissolved Oxygen Percent Saturation |
| aquaMeasure SAL | Temperature, Salinity |
| Vemco VR2AR & VR2AR-X | Temperature |
| Hobo DO U26-001 | Temperature, Dissolved Oxygen mg/L |
| Hobo Temp U22-001 | Temperature |
You can install the development version of calval from
GitHub with:
# install.packages("devtools")
devtools::install_github("ntorrie/calval")Prior to deployment, sensors that can be calibrated are calibrated according to their specific sensor manual. Each sensor is then subjected to a validation test for each variable that it measures.
The first validation test (Test 1) evaluates accuracy of dissolved
oxygen percent saturation measurements. For a validation test of
dissolved oxygen percent saturation, sensors are set to record at 10
minute intervals and placed in an air-tight environment with
water-saturated air at 100 % humidity. The sensors are left to record
for a minimum of 6 hours (Figure 1). Upon completion of the validation
tests, data from each sensor are offloaded and evaluated using the
calval package functions.
The precision for these dissolved oxygen sensors is +/-5 % (InnovaSea 2021). Since the the sensors in the test environment are in water-saturated air, they should measure 100 +/-5 % saturation. Any sensors which record observations outside of the acceptable 95 - 105 % range for > 10 % of the test duration require re-calibration and re-validation. If a sensor fails more than 2 consecutive validation tests it will be sent back to the manufacturer for troubleshooting. Sensors that pass the tests are cleared for deployment.
After Test 1 is completed (when applicable), the second validation test begins. Test 2 evaluates accuracy of temperature, dissolved oxygen mg/L (when applicable), and salinity (when applicable) measurements. If no salinity sensors are present in the test batch, all sensors undergo Test 2a only. If salinity sensors are present in the test batch, all sensors undergo Test 2b only.
For validation Test 2a (evaluating temperature and dissolved oxygen measured in mg/L (when applicable)), sensors are set to record at 10 - 15 minute intervals and submerged in a well insulated tank of fresh water for a minimum of 12 hours. For validation Test 2b (evaluating salinity, temperature, and dissolved oxygen measured in mg/L (when applicable)), the same test setup and duration is used as in Test 2a, except sea water is used instead of fresh water. A minimum of 4 sensors are required in the test batch to ensure a valid test for both Test 2a and 2b. When possible, a mix of sensor types are included in each test batch.
Upon completion of Test 2a or 2b, data from each sensor are offloaded
and evaluated using the calval package functions. For each variable,
an acceptable observation should fall within the range of median
observation +/- the sensor-specific accuracy range. The accuracy range
for each sensor type varies slightly. Any sensor that records outside of
the acceptable range for > 10 % of the test duration will be
re-calibrated (when applicable) and re-validated. If a sensor fails more
than 2 consecutive validation tests it will be sent back to the
manufacturer for troubleshooting. Sensors that pass the tests are
cleared for deployment.
Figure 4. Visualizing results of a temperature test for 20 sensors. Sensor 548042 recorded outside of an acceptable range for 71.3% of the test period, and was therefore flagged for re-calibration and further testing.
After sensors are retrieved from deployment, they undergo post-deployment validation testing following the same procedures as the pre-deployment validation tests. Results from the post-deployment validation tests indicate whether the sensor data recorded during the sensor deployment may have been impacted by sensor drift or biofouling.
Table 2. Validation Test Details
| Test | Test Environment | Minimum Test Duration | Variable(s) Tested | Unit | Validation Algorithm | Loggers Tested |
|---|---|---|---|---|---|---|
| Test 1 | Air-tight environment with water saturated air at 100% humidity. Sensors not submerged. | 6 hours | dissolved oxygen | percent saturation | 100% dissolved oxygen +/- sensor accuracy range | aquaMeasure DOT |
| Test 2a | Insulated tank of fresh water. Sensors fully submerged. | 12 hours | temperature | degrees celsius | Median temperature +/- sensor accuracy range | aquaMeasure DOT, Hobo DO U26-001, Hobo Temp U22-001, Vemco VR2AR, Vemco VR2AR-X |
| Test 2a | Insulated tank of fresh water. Sensors fully submerged. | 12 hours | dissolved oxygen | mg/L | Median dissolved oxygen concentration +/- sensor accuracy range | Hobo DO U26-001 |
| Test 2b | Insulated tank of sea water. Sensors fully submerged. | 12 hours | temperature | degrees celsius | Median temperature +/- sensor accuracy range | aquaMeasure DOT, Hobo DO U26-001, Hobo Temp U22-001, Vemco VR2AR, Vemco VR2AR-X, aquaMeasure SAL |
| Test 2b | Insulated tank of sea water. Sensors fully submerged. | 12 hours | dissolved oxygen | mg/L | Median dissolved oxygen concentration +/- sensor accuracy range | Hobo DO U26-001 |
| Test 2b | Insulated tank of sea water. Sensors fully submerged. | 12 hours | salinity | practical salinity units (PSU) | Median salinity +/- sensor accuracy range | aquaMeasure SAL |