12
Metadata Limitations and the effect on XBT depth adjustment A. Gronell and R. Cowley CSIRO Marine and Atmospheric Research

Metadata Limitations and the effect on XBT depth adjustment

Embed Size (px)

DESCRIPTION

Metadata Limitations and the effect on XBT depth adjustment. A. Gronell and R. Cowley CSIRO Marine and Atmospheric Research. Why we’re interested:. Depth correction:. We need entire database to same vertical scale - PowerPoint PPT Presentation

Citation preview

Metadata Limitations and the effect on XBT depth adjustment

A. Gronell and R. CowleyCSIRO Marine and Atmospheric Research

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

Why we’re interested:

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

Depth correction:

We need entire database to same vertical scale

• We do this by converting all t-7, deep blue, t-6 and t-4 profiles to conform to the Hanawa ‘85 equations

• High or full vert resolution depths are recalculated and replaced.

• Low vert resolution data depths are multiplied by 1.0336

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

Depth correction procedures:

• High vertical resolution profiles with ‘known’ probe type : ‘fixes’ profiles with truncated or rounded depths.

• Replace depths with calculated depths. All depths now to .01 resolution

• Low resolution profiles that either need correction (DPC$01) or are of unknown probe type and fit the criteria for correction:

• Multiply depths by 1.0336

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

• Inadequate Metadata is currently the biggest hurdle

• GTSPP best copy datasets for IO were downloaded and run through Depth correction software

• Anything that was not straight-forward was revisited

• 37678 profiles were screened

Issues affecting depth correction decisions

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

Metadata issues – con.

• Profiles from GTSPP best copy – 37378 profiles

No probe type info 18303 Action depends on date and max depth

PEQ$ missing but determined from elsewhere and added

11442 Action depends on probe type determined

Missing DPC$ field 156 Action depends on date and max depth

• Profiles with DPC$03 or no DPC$ – 18933 profiles

<200m OR >1000m OR post 1996 3960 Do not change depths

>200m AND <1000m AND pre 1996 14973 Change depths

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

Metadata problems found in GTSPP best copy dataset

• CTDs that were identified as XBTs – this is a problem if the DPC$ field is 03 and they fit our other criteria

• T10 profiles that go to 1500m (actually t-5 probes)

• MBTs misidentified (probably XBTs) – max depth for MBTs is ~250m

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

MBTs that go deeper than 300m

Red profiles are post 1975

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

MBT distribution in the Indian Ocean through time

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

Other issues

• Number of depths wrong – more or fewer in the parameter field

• Additional or missing depths – in full resolution profiles, depths duplicated or missing

• Deep_depth field value different from the actual deepest depth

• Single point profiles

• Profiles with exactly .6m resolution (depths assigned and not calculated?) – results in error of ~30m at 800m

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

Conclusions:

• Codes are confusing – there are too many places to find probe type information

• PEQ$• PRT$• FRE$• PTY$• PFR$

• We should be able to reconcile all these into a sensible PEQ$, even if it’s unknown.

• DPC$ need clarification – • 03 = not enough information, do not correct – but we know

most of these should be corrected.• 02 = known probe type, no correction – what does this

mean? No correction done or no correction necessary?

CSIRO. Insert presentation title, do not remove CSIRO from start of footer

Conclusions (con):

• To effectively calculate the corrections needed to fix the fall rate bias of XBTs, we need the best quality baseline data – all using the same base-line fall rate. We therefore need smart methods to get all profiles to the Hanawa equation.

• We can only do this if we have adequate metadata – otherwise, we’re just making educated guesses.