QC summit - IMOS.org.auimos.org.au/fileadmin/user_upload/shared/ANMN/QC_summit_2014/D1... · QC...
Transcript of QC summit - IMOS.org.auimos.org.au/fileadmin/user_upload/shared/ANMN/QC_summit_2014/D1... · QC...
QC summit 5-7th Nov 2014
Developments in Argo Delayed Mode QC:
catering for complicated sensor
and float types
Standard Argo Float & Mission
surface
1000 m
2000 m
drift for 9 days Measure: Pressure Temperature Salinity
Expanded Argo Fleet
> 20 different float models by 12 different manufacturers
PLATFORM_TYPE MANUFACTURER
PALACE WRC
APEX WRC/TWR
APEX_EM WRC/TWR
APEX_D TWR
PROVOR MARTEC/NKE
ARVOR NKE
ARVOR_C NKE
SOLO_II SIO_IDG
SOLO_W WHOI
NEMO OPTIMARE
NAVIS SBE
NINJA TSK
NOVA METOCEAN
… …
Argo Enhancements
EM Argo vertical velocity
Deep Argo
Ice Argo Regional Seas / Coastal Argo
Near Surface SST/SSS
Bio Argo
New Argo Data Format V3.1
Metafiles contain information about the float: • Float characteristics • Float deployment and mission information • Float sensor information • Float calibration information • Configuration parameters
Historically, metafiles have not been consistently populated or standardised between DACs.
• Fields are incorrectly filled or empty • Non-uniform entries, i.e. “SBE41CP_1769” or “1769” • Not recording enough detail to track float failure modes, i.e. battery type or controller board type
Metadata files
Why is metadata important?
SENSOR
ACOUSTIC_PRECIPITATION
ACOUSTIC_GEOLOCATION
CTD_CNDC
CTD_PRES
CTD_TEMP
EM
FLUOROMETER_CDOM
FLUOROMETER_CHLA
IDO_DOXY
OPTODE_DOXY
RADIOMETER_DOWN_IRR
RADIOMETER_PAR
SCATTEROMETER_BBP
SCATTEROMETER_TURBIDITY
SPECTROPHOTOMETER_NITRATE
STS_CNDC
TRANSISTOR_PH
TRANSMISSOMETER_CP
…
SENSOR_MAKER
AMETEK
AANDERA
DRUCK
FSI
KISTLER
PAINE
SEASCAN
SBE
WETLABS
MBARI
SATLANTIC
…
SENSOR_MODEL
SBE41
SBE41CP
SBE41CP_V1
SBE41CP_V1.2
SBE41CP_V3.0c
SBE61CP
SBE63_OPTODE
SBE43I
AANDERAA_OPTODE_3830
AANDERAA_OPTODE_4330
DRUCK
PAINE_1500PSIA
SUNA
SATLANTICOCR507
FLNTU
ECO_PUCK
ECO_[FL]_[BB]_[BB2]_[TRIPLET]_[CD]
…
• We need to keep track of an increasing diversity of sensors and parameters..
SCIENTIFIC_CALIB_COMMENT =
PRES Pressures adjusted using PRES_SurfaceOffsetTruncatedplus5dbar_dBAR;
Pressure drift corrected; Manufacturers sensor accuracy
TEMP No significant temperature drift detected; SBE sensor accuracy
PSAL Salinity drift - correction applied using OW piecewise-fit based on deepest-
possible theta levels and Argo and CTD reference datasets.
SCIENTIFIC_CALIB_EQUATION =
PRES PRES_ADJUSTED = PRES - [PRES_SurfaceOffsetTruncatedplus5dbar_dBAR
- 5]
TEMP no change
PSAL PSAL_ADJUSTED = sal(CNDC,TEMP,PRES_ADJUSTED); PSAL_ADJ
corrects conductivity cell thermal mass (CTM), Johnson et al, 2007, JAOT
SCIENTIFIC_CALIB_COEFFICIENT =
PRES PRES_SurfaceOffsetTruncatedplus5dbar_dBAR in TECH file for N-1 profile
TEMP no change
PSAL same as for PRES_ADJUSTED; CTL: alpha=0.021, tau=21;
Calibration Fields must be filled in
• For floats with multiple missions;
– The configuration from the first cycle is set to 1.
– Each subsequent configuration change has a new mission number, i.e. from 1 to N.
Mission Parameters
CONFIG_MISSION_NUMBER
• Argo floats usually have one mission for the life of the float; i.e. CONFIG_MISSION_NUMBER = 0 (launch/pre-deployment info) and = 1 (basic mission). • Some floats now perform multiple missions, i.e. combination shallow/deep profiles.
Missions and configuration parameters
Core Argo CONFIG_AscentSpeed_mm/s CONFIG_ParkPressure_dbar CONFIG_ProfilePressure_dbar CONFIG_DownTime_hours CONFIG_IceDetection_degC
Bio Argo CONFIG_SensorInPumpedStream_LOGICAL CONFIG_OptodeMaxTransmittedTemperature_mdegC CONFIG_<short sensor name>DepthZone<N>AscentSamplingPeriod_seconds CONFIG_CroverBeamAttenuationWavelength_nm CONFIG_FlbbSampling_NUMBER
Approved Units List
COUNT NUMBER byte days hours DDMMYYYY dbar …
• Accommodates multi-profile floats [N_PROF > 1].
• Some floats make measurements at different vertical levels in the same cycle; i.e. a primary CTD profile of P, T and S at 2 db intervals from 0-1000 db and a secondary profile of low res oxy data at N discrete levels.
• PRES is the link between parameters in the core- and b- files.
Example: high res primary CTD profile with low res secondary oxygen profile:
Multi-profile files
In the core-Argo profile file, N_PROF = 2, N_LEVELS = 500 PRES = [2, 4, 6, ……………………...............1000] = [250, 500, 750, 1000, FillValue, …………] TEMP = [T2, T4, T6, ……………………….T1000] = [FillValue, ………………………………..] PSAL = [S2, S4, S6, ………………………..S1000] = [FillValue, ………………………………..]
In the b-Argo profile file, N_PROF = 2, N_LEVELS = 500. PRES = [2, 4, 6, ……………………...............1000] = [250, 500, 750, 1000, FillValue, ………....] NITRATE = [FillValue, ………………………………..] = [N250, N500, N750, N1000, FillValue, …] [Other intermediate nitrate wavelength variables at same levels as NITRATE...]
Processing Bio Argo variables: Oxygen, Chl a and Backscattering
Available at: http://www.argodatamgt.org/Documentation
Processing documents for CHLOROPHYLL A and BACKSCATTERING are in draft form and will be discussed at BioArgo workshop 3-4th Nov 2014, Canada and should be available on the ADMT website next year.
Does not address oxygen QC but seeks to ensure that all countries are documenting data and metadata correctly.
Argo Data Structure • netcdf files - traj, meta and tech files
Example profile file: R5901156_001.nc
latitude, longitude, cycle number, calibration date, scientific calibration comment etc
Raw Adjusted Raw_QC Adjusted_QC Adjusted_ERROR
PRES PRES_ADJUSTED PRES_QC PRES_ADJUSTED_QC PRES_ADJUSTED_ERROR
TEMP TEMP_ADJUSTED TEMP_QC TEMP_ADJUSTED_QC TEMP_ADJUSTED_ERROR
PSAL PSAL_ADJUSTED PSAL_QC PSAL_ADJUSTED_QC PSAL_ADJUSTED_ERROR
QC Flag Meaning
0 No QC was performed
1 Good data
2 Probably good data
3 Probably bad data that are potentially
correctable
4 Bad data
5 Value changed
6 Not used
7 Not used
8 Interpolated value
9 Missing value
Overall Profile QC flags:
POSITION_QC
PROFILE_PRES_QC
PROFILE_TEMP_QC
PROFILE_PSAL_QC
Profile
QC Flag
Meaning
“ ” No QC was performed
A N = 100%; all profile levels
contain good data
B 75% <= N < 100%
C 50% <= N < 75%
D 25% <= N < 50%
E 0% < N < 25%
F N = 0%; No profile levels have
good data
• profile files – “R” or “D” files
DMQC processing Point-wise pressure check
and QC flagging
• can view primary and secondary profiles at the same time • QC individual values and bands of bad data • can transfer primary QC to secondary if depths match
TS plot and Nearby Argo comparison
TEM
P
PSAL PSAL
DOXY
PR
ES
PR
ES New features
include: visualisation for Bio Argo parameters: DOXY PDEN OXSAT PTMP Select nearby Argo profiles for comparison: • lat/lon range • time window ± number of days
• spikes
• bad values
• biofouling • realistic data
range
• early indicator
of drift
New interface for multi-profile floats
• allows you to select and locate individual profiles • can turn off neighbouring profiles with bad data
Manual profile inspection
Section plots - bad data values, grounding, front crossings
Salt hooks
• salty by approx 0.005
• 2 deepest measurements reported at
almost identical pressures, (end of
descent and first sample of profile)
• inadequate flushing, contamination
of high salinity water from park depth
Manual profile inspection
Density inversions and point-
wise inspection of
temperature and salinity data
(Now includes conductivity
ratio)
Profile by profile flagging
- spikes, flag inversions, salt
hooks, grounded profiles
From Barker et al, JAOT, 2011
1. PRES_ADJ (cycle i) = PRES (cycle i) – SP (cycle i+1)
Surface Pressure Correction
2. Once SPO applied to PRES_ADJ → recalculate PSAL_ADJ using new PRES_ADJ
Druck microleak
Normal diaphragm
Cracks in glass to metal seals
result in slow oil leak,
depressed diaphragm,
negative SPO, eventual
electrical short
• error in salinity as a result of the thermal
inertia of the conductivity cell, see Johnson et
al, 2007
• occurs primarily in regions of strong thermal
gradients, i.e strong thermocline to well mixed
surface layer
• software calculates new PSAL_ADJ to
correct for thermal lag spikes primarily in
regions of large gradients, i.e. base of the
thermocline
Cell Thermal Mass Error
Salinity Drift Assessment
• Comparison with 3 climatologies
• CARS / WOA98, Gouretski and Koltermann,
new CARS (inc. Argo)
• Comparison with nearby Argo data
Assess salinity drift on deep potential temperature
surface – where water masses are most stable
Salinity Drift Correction • Use software written by Owens and Wong to apply a salinity drift correction.
• Calculates a delta S correction based on a linear fit to climatology (historical CTD
data + nearby Argo data).