trajectory v3.1 file update · v3.1 format has possibility of improving velocity calculations and...
TRANSCRIPT
Trajectory V3.1 File Update
Megan Scanderbeg, J. Gilson, N. Zilberman
Argo Steering Team Meeting
March 2016
Papers using Argo trajectory files
0
2
4
6
8
10
12
14
16
200
3
200
4
200
5
200
6
200
7
200
8
200
9
2010
2011
2012
2013
2014
2015
99+ papers
4 papers focused on global velocity fields
Most are regional studies
10+ papers used YoMaHa
3+ papers used ANDRO
1+ paper used G-YoMaHa
3+ assimilated into models
Number of ‘R’ Trajectory files on the GDAC by DAC and version
DAC Total Version 3.1 Version 2 Float type in V3.1
AOML 5577 0 5577
BODC 420 0 420
Coriolis 2121 789 1332 PROVOR, ARVOR
CSIO 324 10 314 PROVOR, ARVOR
CSIRO 569 375 194 APEX w/ ARGOS
INCOIS 360 57 303 PROVOR, ARVOR
JMA 1392 0 1392
KMA 193 0 193
KORDI 119 0 119
MEDS 398 96 302 APEX, NOVA
NMDIS 19 15 4 ARVOR
Progress 12%
Outline
Trajectory Cookbook updates
Scientific motivation for moving to V3.1 files
Flexibility of V3.1 format
Need to finalize and document b-traj files
Delayed mode quality control for trajectory files
Trajectory Cookbook Updates New, updated tables for all float types except
PROVOR/ARVOR which include all MCs, not just cycle timing ones
Added table on Deep NINJA
Removed all references to STANDARD_FORMAT_ID since this variable was removed at ADMT-16 in November
Moved APEX estimation methods to Annex
Better suited to delayed mode rather than real time
Updated relative measurement codes
Allows for inclusion of more technical data
Float cycle timing information Sending back crucial timing information
Not sending back crucial timing information
APEX APF9i with firmware revision 072314 or later
ARVOR/PROVOR
NAVIS
NINJA from 2002-2007 and Deep NINJA
NOVA
SOLO-II/S2A
APEX APF8, APF9a and t, APF9i with firmware revision before 072314
NEMO
NINJA from 2008
SOLO
These models are mostly older and often not deployed anymore, but still lots of data on GDACs. Don’t know what is happening with APEX and APF11, but Dana Swift is now involved and knows crucial cycle timing information.
Additional timing information Argo’s data policy calls for data that floats send
back to be available in 24 hours
All times are not available unless using V3.1
More timing information can improve velocity estimates
With Iridium, more timing information is becoming available
Where will this be stored?
ARVOR
NAVIS
Nova
SOLO-II
APEX
APF9i
0
200
400
600
800
1000
1200
200
5
200
6
200
7
200
8
200
9
2010
2011
2012
2013
2014
2015
Iridium
Argos
Deployments by year
A user’s prospective of Argo trajectory files
Calculating velocities at 1000m in Southern Pacific
A typical year (2013) of Argo coverage for the 140°E to 70°E and 20°S to 40°S region
Doing extrapolation based on background velocity and inertial current (Park et al, 2004) to estimate location of rise and fall of Argos floats
Using GPS locations for Iridium floats
Profile
pressure
Parking
pressure
Surface
Cycle N
DST DET DDET
AST
AET
TST
TET
Depth
Cycle N-1
Argos/GPS
locations
PET
Extrapolation method relies on knowing AET/TST and
TET/DST and surface locations
• Blue times in V3.1 files only
• TET, PST and PET can help with drift velocity
calculations
PST
Floats in Southern Pacific 140°E to 70°W; 20°S to 40°S
DAC # of floats with at least one profile in region
AOML 927
Coriolis 40
CSIO 3
CSIRO 112
JMA 31
Total 1113
KORDI 1
MEDS 2
Total 1116 365
331
93
56
20 5
SOLO
APEX w/ Argos
SOLO-II
APEX w/Iridium
PROVOR
NAVIS
Used 873 out of 1116 floats
Majority (124) rejected due to <3 cycles close to each other at correct drift pressure
73 APEX Argos floats rejected due to bad AET/TST, bad TET estimation, bad file format
713 have Argos; 160 have Iridium
In East Australian Current (31°S,154°E)
APEX APF8 float
Magenta dots are
actual surface fixes
Blue dots are extrapolated
positions at thirty equal
time intervals from surface
arrival to estimated
transmission end time
TET
Estimated TET
using process in
Trajectory Cookbook
Extrapolation
method from Park,
et al 2004
Estimated additional
10 km traveled on
surface by float –
surface velocity
increases by a
factor of 2
Meridio
nal dis
pla
cem
ent
in k
m
zonal displacement in km
Magenta dots are actual
surface fixes
Blue dots are
extrapolated positions at
thirty equal time intervals
from estimated surface
arrival to estimated
transmission end time
In East
Australian
Current
(15°S,150°E)
Dmoded SOLO float
Extrapolation method
from Park, et al 2004
Estimated additional
25 km traveled on
surface by float –
changes subsurface
velocity calculation
by 25%
Meridio
nal dis
pla
cem
ent
in k
m
zonal displacement in km
Flexibility of V3.1 format V3.1 format identifies where each measurement
occurred within cycle, even if no time is available
Allows for additional timing estimates to be inserted in delayed mode
Allows some information previously stored in tech files to be in traj files, but need to focus on keeping scientifically useful data, not everything float sends back
Need to think about when to include information in trajectory file or another file – users may need to get comfortable using traj files
B-trajectory files
Format needs to be finalized and well documented
Mathieu lists 672 floats as having DOXY
Time consuming and needs an expert to liaise between Bio-Argo community and ADMT
Separate b-Trajectory Cookbook will likely be needed
Agreed on how to store surface in-air measurements needed for calibration of oxygen sensors
Agreed on how to store RAFOS timing information
Location of all
b-profiles on GDACs
462 floats
Delayed mode trajectory files John Gilson has 1317 ‘D’ trajectory files
Some are for SOLO floats: dmoded as they die
Some are for active SOLO-II floats so that additional information float sends back can be sent out in real time
Working with Annie Wong to start discussions on delayed mode trajectory files with current dmode operators & considering a workshop in 2017
Some dmode procedures will apply to all float types and some will be float type specific
Some dmode operators may choose to use ANDRO V3.1 trajectory files for dmode
Conclusions Conversion to V3.1 format is starting; DACs need help with
this transition
V3.1 format has possibility of improving velocity calculations and opening up other areas of study if more timing information is included
Need to focus on putting scientifically useful information in V3.1 traj files even though it is flexible enough to handle many types of data
Could use a b-trajectory file expert to help describe and document a b-trajectory cookbook
Need to start developing a dmode quality control process for trajectory files
Possible dmode procedures Applying PRES/PSAL corrections from profile dmode
to trajectory files
Quality controlling cycle number
Quality controlling surface times and positions
Estimate additional timing information
Check that measurement codes are applied correctly
Apply the grounded flag
Calculate the best drift PRES/TEMP/etc.
Recent papers where Argo trajectory files used to calculate velocities • Gray & Riser, 2014 – global mean velocity field • Ollitrault & Colin de Verdiere, 2014 – ANDRO • Ollitrault & Rannou , 2013 – ANDRO • Katsumata, 2010 – Gridding of YoMaHa
• Castellanos et al, 2015 – Tropical Atlantic • Chiswell et al, 2015 – near New Zealand (used YoMaHa) • Cravatte et al, 2015 – South Pacific • Liu et al, 2015 – Luzon Straight • Ohde et al, 2015 – North Atlantic • Park et al, 2015 – East Sea • Ponsoni et al, 2015 – East Madagascar (used ANDRO) • Bowen et al, 2014 – near New Zealand • Carrier et al, 2014 – assimilating traj file into NCOM-4DVAR • Chiswell et al, 2014 – assimilating traj file into BRAN; compared to YoMaHa • Dong et al, 2014 – South Atlantic (used G-YoMaHa) • Giglio, 2014 (thesis) – North Pacific • Koenig et al, 2014 – surface velocities in Drake Passage • Kosempa et al, 2014 – Southern ocean (compared to YoMaHa) • Perez et al, 2014 – Atlantic (used YoMaHa) • Schmid, 2014 – South Atlantic • Zilberman et al, 2014 – South Pacific
Where to keep profile times As a new parameter in the core profile files?
Works best if times accompany all profile measurements How would dmode be done on times in conjunction with
trajectory files?
In V3.1 trajectory files? Works best if a sparse snapshot of times Already included for SOLO-II, PROVOR/ARVOR
In the B files? Might require a B file even if it is a core float Not favored
Keep this data at the DACs and not in distributed files? Is this information useful? Or only used by DM
operators for specific calculations? If more generally useful we need a cost/benefit assessment…