quality data for their thesis/dissertation research; and the station owner gets the required
attention to the immediacies of station operation as required.
6. Data Validation
Successful interpretation of the incoming near-real time data by either experts or
expert system software (see Information Systems, below) is dependent on the quality of
the incoming data. In particular, in remote locations, with sensors chosen for long-term
stability and dependability, recorded values for many parameters may experience drift.
Automated expert system evaluation can identify suspect data after it has been
produced; however, it is necessary for establishing quality of the long-term data set to
periodically validate the data. These validations are performed by comparison of the in
situ sensors that have been exposed to the environment for extended periods against
recently calibrated instruments maintained and stored in a clean, controlled
environment. Validation of a standard CREWS station includes intercomparison of the
wind sensor data and verification of the quality of the conductivity-temperature-depth
(CTD) data. The wind sensor data can be monitored for self-consistency and against
local weather reports remotely. The CTD, on the other hand, requires a local technician
to visit the station for a calibration check.
Validating CTD sensors on the CREWS stations is made by comparison of the
CTD values from the station, as reported to the CREWS Web site, with readings from a
calibrated portable CTD at the station. To perform a validation the maintainer will
need a recently calibrated high-precision CTD, a (preferably) waterproof or water
resistant laptop computer with an RS-232 interface, CTD processing software, an RS232 communication cable, and access to the Web to see the currently reporting values.
Validation is done both before and after a simple in-water cleaning process, consisting
of including wiping down the sensor surfaces with a soft cloth, to ascertain the effect of
biofouling or other contaminants on the measurements. Since the reported values are
hourly averages of six discrete data points, the calibration CTD will need to be operated
on the same sampling schedule. Validation measurements should be collected for two
time periods before and two time periods after cleaning, so the total calibration time,
including the cleaning dive, is approximately five hours. It is possible for approved
divers to use the calibration time to enter the water to clean and inspect the mooring and
supporting hardware for signs of wear. If the CREWS station CTD is found to be out
of calibration, it is removed and replaced with the calibrated CTD. The field CTD that
was removed would then be returned for cleaning and calibration. Maintainers should
follow this validation process at least once every two weeks.
7. Information Systems
7.1 PRESENTATION OF REAL-TIME RAW DATA
At the station each instrument sends data (or is polled by a data logger), which
averages the number of readings over the hour and then transmits a stream of numbers
("datastream") up to a GOES satellite. The datastream is then relayed to Wallups
Island, Virginia, and then to AOML (see Figure 4). Individual sensor values are then
parsed from the datastream and presented as near real-time values on the CREWS Web
Page (http://www.coral.noaa.gov/crews), with the disclaimer that the data should only
be considered as provisional since they have not yet been reviewed. The disclaimer is
142
Hendee, Stabenau, Florit, Manzello and Jeffris
attention to the immediacies of station operation as required.
6. Data Validation
Successful interpretation of the incoming near-real time data by either experts or
expert system software (see Information Systems, below) is dependent on the quality of
the incoming data. In particular, in remote locations, with sensors chosen for long-term
stability and dependability, recorded values for many parameters may experience drift.
Automated expert system evaluation can identify suspect data after it has been
produced; however, it is necessary for establishing quality of the long-term data set to
periodically validate the data. These validations are performed by comparison of the in
situ sensors that have been exposed to the environment for extended periods against
recently calibrated instruments maintained and stored in a clean, controlled
environment. Validation of a standard CREWS station includes intercomparison of the
wind sensor data and verification of the quality of the conductivity-temperature-depth
(CTD) data. The wind sensor data can be monitored for self-consistency and against
local weather reports remotely. The CTD, on the other hand, requires a local technician
to visit the station for a calibration check.
Validating CTD sensors on the CREWS stations is made by comparison of the
CTD values from the station, as reported to the CREWS Web site, with readings from a
calibrated portable CTD at the station. To perform a validation the maintainer will
need a recently calibrated high-precision CTD, a (preferably) waterproof or water
resistant laptop computer with an RS-232 interface, CTD processing software, an RS232 communication cable, and access to the Web to see the currently reporting values.
Validation is done both before and after a simple in-water cleaning process, consisting
of including wiping down the sensor surfaces with a soft cloth, to ascertain the effect of
biofouling or other contaminants on the measurements. Since the reported values are
hourly averages of six discrete data points, the calibration CTD will need to be operated
on the same sampling schedule. Validation measurements should be collected for two
time periods before and two time periods after cleaning, so the total calibration time,
including the cleaning dive, is approximately five hours. It is possible for approved
divers to use the calibration time to enter the water to clean and inspect the mooring and
supporting hardware for signs of wear. If the CREWS station CTD is found to be out
of calibration, it is removed and replaced with the calibrated CTD. The field CTD that
was removed would then be returned for cleaning and calibration. Maintainers should
follow this validation process at least once every two weeks.
7. Information Systems
7.1 PRESENTATION OF REAL-TIME RAW DATA
At the station each instrument sends data (or is polled by a data logger), which
averages the number of readings over the hour and then transmits a stream of numbers
("datastream") up to a GOES satellite. The datastream is then relayed to Wallups
Island, Virginia, and then to AOML (see Figure 4). Individual sensor values are then
parsed from the datastream and presented as near real-time values on the CREWS Web
Page (http://www.coral.noaa.gov/crews), with the disclaimer that the data should only
be considered as provisional since they have not yet been reviewed. The disclaimer is
142
Hendee, Stabenau, Florit, Manzello and Jeffris