Shipboard ADCP measures water and bottom velocities relative to the ship. Water velocity in earth coordinates, called "absolute velocity", is calculated by subtracting the ship's velocity (over the ground) from water velocity measured by ADCP which is called "relative velocity". Here, the ship's velocity is calculated from GPS positioning data.
Detailed correction method for the ship's velocity and ADCP relative velocity is as follows.
- Correction of the GPS data (ship's velocity)
Errors in GPS data need to be removed in advance in order to calculate high quality ship's velocity.GPS data used here is 1-Hz GPS NMEA data recorded in the ADCP logger.Bad position data are detected by the following criteria, when:
a) the estimated ship's velocity from the GPS data in every second exceeds 19 knots, and
b) the estimated ship's acceleration by GPS data in every second exceeds 0.5 m/s^2.
Bad data are removed and interpolated from nearby good data.Furthermore, high frequency fluctuations are removed by a 20-second running mean filter.
- Correction of the ADCP data
Errors (alignment of hull-mount transducer and roll/pitch error) in relative velocity were rejected by the following method:
1) Calculation of miss alignment using bottom track data
The miss alignment value is estimated from the ADCP data measured in bottom track mode, as follows:
* When bottom track data were not measured, preset alignment value which was obtained in the latest performance test cruise has been applied.
1-1) Ship's velocity correction and roll/pitch error correction
For the selected bottom track data, GPS data is replaced by the corrected data pre-processed in Section 2. Then, roll/pitch error is also corrected if ship's heading and roll/pitch data are measured by the inertial navigation system. These processing are performed by "VM-DAS" software (Teledyne R. D. Instruments).
* Please refer to "readme" of each cruise whether roll/pitch error correction is performed or not.
1-2) Estimation of miss alignment
Miss alignment value was estimated from bottom track velocity and GPS velocity, using formula of Joyce (1989). Then, input data were selected as follows: a) ship velocity and ship heading were continuously, b) the "Ping Correlation" of ADCP is 200 counts and over (MAX: 255 counts), and c) the "Echo Intensity" is 60 counts and over (MAX: 255 counts).
2) Correction of miss alignment and roll/pitch error for all period of data
Using the estimated miss alignment value, the miss alignment is corrected for all period of the ADCP data by VMDAS software. Then, replacing of GPS data and roll/pitch error correction are performed by the same processing as 1-1) in this section.
- Calculation of absolute velocity for each ping and making 5-minute time-average dataset
Dataset provided here is a 5-minute time average of absolute velocity, which is calculated from the screened pings of absolute velocity data.
Screening of ping velocity data is executed in the following sequence:
1) Removal of the error velocity near the sea surface (so-called ringing velocity)
Velocity data near the sea surface, which measured immediately after sending pulse, is often recorded as much faster than actual velocity by the influence of transducer resonance. Such error velocity near the sea surface, called "ringing velocity", is often measured with similar direction to ship's course.
So, for each ping data,
a) if the magnitude of calculated absolute velocity for a layer is faster than 1.0m/s and more than twice compared with the value of one layer below, and -
b) if the direction of the velocity is within ±45 degrees from the ship’s heading,
then, the velocity data between the surface and this layer are all removed as ringing velocity.
2) Removal of the "Bad Correlation" and "Bad Echo Intensity" data
Ping velocity data are used for the average calculation only when the "Ping Correlation" of ADCP is 120 counts and over (MAX: 255 counts) and the "Echo Intensity" is 25 counts and over (MAX: 255 counts).
3) Removal of data near the seafloor
The data measured near the seafloor is often recorded error velocity due to side rope. Thus, 85% or more water depth measurement data is removed. If the seafloor depth was not measured by ADCP, we apply the seafloor depth measured by MBES system
concurrently. If neither of the seafloor depths exist, data quality flag (4 : Questionable) are added to all vertical layer of the data during the period that sound reflection from the seafloor seems to be detected.
4) Removal of the random noise
In order to remove random noise, 2-sigma filter is used for ping velocity data in every 5-minute average section.
In calculating 5-minute average, if more than 90% of ping velocity data in an average section are screened out by the above filters, "Percent Good" is not more than 10%, then the average data is assumed unreliable and replaced it to be invalid (NaN).
In addition, data quality flag (4 : Questionable) are added when:
a) Absolute "Error Velocity" in 5 minutes is 12cm/s and over, or
b) "Percent Good" in 5 minutes is less than 50%.
The dataset is written in ODV (Ocean Data View) format. See Data Format for details.