Drift Characteristics of Paroscientific pressure sensors … by Randolph Watts, Maureen Kennelly, Karen Tracey, and Kathleen Donohue (University of Rhode.

Slides:



Advertisements
Similar presentations
It is very difficult to measure the small change in volume of the mercury. If the mercury had the shape of a sphere, the change in diameter would be very.
Advertisements

Magdalena Andres 1, Jae-Hun Park 1, Mark Wimbush 1, Kyung Il Chang 2, Hiroshi Ichikawa 3 1 Graduate School of Oceanography, University of Rhode Island.
Preliminary Results from ATDD’s Soil Moisture/Temperature Testbed Soil Moisture and Soil Temperature Observations and Applications: A Joint U.S. Climate.
Propagation of Error Ch En 475 Unit Operations. Quantifying variables (i.e. answering a question with a number) 1. Directly measure the variable. - referred.
C. A. Collins 1, R. Castro Valdez 2, A.S. Mascarenhas 2, and T. Margolina 1 Correspondence: Curtis A. Collins, Department of Oceanography, Naval Postgraduate.
Horizontal Pressure Gradients Pressure changes provide the push that drive ocean currents Balance between pressure & Coriolis forces gives us geostrophic.
Determination of Station Depths Relative to NGVD29 Methods and results Jan 24, 2003 Charles Seaton.
Chapter 12 - Forecasting Forecasting is important in the business decision-making process in which a current choice or decision has future implications:
Horizontal Pressure Gradients Pressure changes provide the push that drive ocean currents Balance between pressure & Coriolis forces gives us geostrophic.
Chapter 2 Graphs, Charts, and Tables – Describing Your Data
Chapter 2 Describing Data Sets
Regression line – Fitting a line to data If the scatter plot shows a clear linear pattern: a straight line through the points can describe the overall.
Generalized Surface Circulation
Long-Term Ambient Noise Statistics in the Gulf of Mexico Mark A. Snyder & Peter A. Orlin Naval Oceanographic Office Stennis Space Center, MS Anthony I.
Principles of Sea Level Measurement Long-term tide gauge records  What is a tide station?  How is sea level measured relative to the land?  What types.
Anatomy of Anomalies Total Field Magnetics and Ground Penetrating Radar at a Potential Archaeological Site.
Barometric Altimetry Using the Vernier LabPro. Purpose of Report Improved altitude determination for Balloon Fest and other activities Improved altitude.
Classification of Instruments :
Dr Mark Cresswell Statistical Forecasting [Part 1] 69EG6517 – Impacts & Models of Climate Change.
V. Rouillard  Introduction to measurement and statistical analysis ASSESSING EXPERIMENTAL DATA : ERRORS Remember: no measurement is perfect – errors.
MOTION.
A Look at Two Tsunamison the BC Coast with the Neptune-Canada Tsunami-meters A Look at Two Tsunamis on the BC Coast with the Neptune-Canada Tsunami-meters.
Chapter 3 – Descriptive Statistics
BPS - 3rd Ed. Chapter 211 Inference for Regression.
Measuring pressure. Conductivity, Temperature and Depth (CTD) Despite the name, all CTDs actually measure pressure, which is not quite the same thing.
Quantitative Skills 1: Graphing
“ New Ocean Circulation Patterns from Combined Drifter and Satellite Data ” Peter Niiler Scripps Institution of Oceanography with original material from.
Lecture 2 Graphs, Charts, and Tables Describing Your Data
“ Combining Ocean Velocity Observations and Altimeter Data for OGCM Verification ” Peter Niiler Scripps Institution of Oceanography with original material.
Model LSW formation rate (2 yr averages) estimated from: (red) CFC-12 inventories, (black) mixed layer depth and (green) volume transport residual. Also.
Renellys C. Perez 12, Silvia L. Garzoli 2, Ricardo P. Matano 3, Christopher S. Meinen 2 1 UM/CIMAS, 2 NOAA/AOML, 3 OSU/COAS.
Unit One Notes: Graphing How do we graph data?. Name the different types of graphs (charts).
ISPOL Ocean Turbulence Project Miles McPhee McPhee Research Co. Naches WA USA.
Business Statistics: A Decision-Making Approach, 6e © 2005 Prentice-Hall, Inc. Chap 2-1 Business Statistics: A Decision-Making Approach 6 th Edition Chapter.
Sundermeyer MAR 550 Spring Laboratory in Oceanography: Data and Methods MAR550, Spring 2013 Miles A. Sundermeyer Observations vs. Models.
Measuring and Significant Digits. Parallax Error Parallax is the apparent shift in position of an object caused by the observer’s movement relative to.
Correlation & Regression Chapter 15. Correlation It is a statistical technique that is used to measure and describe a relationship between two variables.
Chap 2-1 A Course In Business Statistics, 4th © 2006 Prentice-Hall, Inc. A Course in Business Statistics 4 th Edition Chapter 2 Graphs, Charts, and Tables.
Propagation of Error Ch En 475 Unit Operations. Quantifying variables (i.e. answering a question with a number) 1. Directly measure the variable. - referred.
Application of Radial and Elliptical Surface Current Measurements to Better Resolve Coastal Features  Robert K. Forney, Hugh Roarty, Scott Glenn 
11/23/2015Slide 1 Using a combination of tables and plots from SPSS plus spreadsheets from Excel, we will show the linkage between correlation and linear.
Applied Quantitative Analysis and Practices
An example of vertical profiles of temperature, salinity and density.
Correlation – Recap Correlation provides an estimate of how well change in ‘ x ’ causes change in ‘ y ’. The relationship has a magnitude (the r value)
Level of No Motion (LNM)
CHAPTER 5: Data sampling design, data quality, calibration, presentation Sampling Physical variables are continuous, but samples (water samples) and digital.
Geopotential and isobaric surfaces
CfE Advanced Higher Physics
Statistics Presentation Ch En 475 Unit Operations.
 p and  surfaces are parallel =>  =  (p) Given a barotropic and hydrostatic conditions, is geostrophic current. For a barotropic flow, we have and.
1 Rosalia Daví 1 Václav Vavryčuk 2 Elli-Maria Charalampidou 2 Grzegorz Kwiatek 1 Institute of Geophysics, Academy of Sciences, Praha 2 GFZ German Research.
Uncertainty2 Types of Uncertainties Random Uncertainties: result from the randomness of measuring instruments. They can be dealt with by making repeated.
SIO 218A, Spring 2014 Petia Yanchulova. cPIES IES Data Processing Manual, URI, 2007.
Instruments. In Situ In situ instruments measure what is occurring in their immediate proximity. E.g., a thermometer or a wind vane. Remote sensing uses.
Southern Ocean Observations of AWI South Atlantic Workshop, San Ceferino, 9 May 2007 Andreas Macrander Alfred-Wegener-Institut für Polar- und Meeresforschung,
Measurement Counting yields exact numbers Measuring yields inexact values Always some error Your job is to minimize it.
Andrea Kaiser-Weiss, Melbourne Joint GHRSST Workshop, 6 th March 2012 Experiences with SST profiles from near-surface Argo measurements A. Kaiser-Weiss.
VLVNT’08 - ToulonM. Ardid – Positioning System of the ANTARES Neutrino Telescope Positioning system of the ANTARES Neutrino Telescope Miguel Ardid (IGIC.
BPS - 5th Ed. Chapter 231 Inference for Regression.
Review Design of experiments, histograms, average and standard deviation, normal approximation, measurement error, and probability.
Describing Data Week 1 The W’s (Where do the Numbers come from?) Who: Who was measured? By Whom: Who did the measuring What: What was measured? Where:
Electrode Spread (Array type)
For a barotropic flow, we have is geostrophic current.
Section 11.1 Day 3.
For a barotropic flow, we have is geostrophic current.
Mark A. Bourassa and Qi Shi
GG 450 February 19, 2008 Magnetic Anomalies.
Statistical Methods For Engineers
Statistics Review ChE 477 Winter 2018 Dr. Harding.
Week 5: Thermal wind, dynamic height and Ekman flow
Presentation transcript:

Drift Characteristics of Paroscientific pressure sensors … by Randolph Watts, Maureen Kennelly, Karen Tracey, and Kathleen Donohue (University of Rhode Island) PIES + current meter & CPIES arrays Paroscientific Digiquartz sensors 4 arrays, 7 settings leveling / dedrifting method Characterizing the observed drifts

Drift Characteristics of Paroscientific pressure sensors Aim to show… –(exponential+linear) curve is well suited to represent the drift –Geostrophic leveling & drift detection is reliable to ~0.01 dbar new drift curves differ from previous method that used just P data, although both use (exp + lin) fit –Averaging records from two or more ‘same-site’ sensors produces an average drift, but not zero drift –Pre-stressing reduces |drift|

current and pressure recording inverted echo sounder Measures bottom current. (50 m off bottom) Emits 12kHz sound pulses. Measures round trip travel times of acoustic pulses to sea surface and back. Measures bottom pressure (and temperature). (includes acoustic release + relocation radio+strobe light) CPIES:

CPIES array yields… A CPIES array yields daily maps of upper and deep circulation. Look-up tables interpret acoustic travel times as geopotential height (0 referenced to 5000 dbar). 2-D arrays of CPIES estimate horizontal gradients of geopotential to calculate geostrophic velocities. Velocity profiles are referenced by measured near- bottom currents. Bottom pressures are leveled using near-bottom currents to map the geostrophic streamfunction.

Digiquartz sensor - Paroscientific website Digiquartz frequency increases with pressure-applied load: –Accuracy 100 ppm; drift ~ 10 ppm; stability & resolution ~ <1-2 ppm PIES measures frequency with 4MHz temperature-compensated crystal –freq. spec: 10 ppb accuracy to serve as a stable reference

URI PIES & CPIES deployment sites 4 recent arrays with current meters; 7 settings

URI PIES & CPIES deployment sites 4 recent arrays with current meters; 7 settings Gulf of Mexico Drake Passage Kuroshio, East China Sea Kuroshio Extension

arrays of PIES & CMs … combined to level the pressure sensors geostrophically Drake Passage Gulf of Mexico Kuroshio Extension

Detide detide Tidal response analysis (Munk and Cartwright, 1977) determines the tidal constituents for each instrument. Tides are then removed from the pressure records.

Kuroshio Extension deep streamfunction maps LP filter (1-mo); July-Aug 2004 These are used to level and dedrift the P(t) records as follows.

Leveling (and drift) The streamfunction from CMs and the pressure from PIES measure the same geostrophic pressure field. – The two fields should only differ by a site- dependent leveling constant.

drift-curve fit to P(t) - P cm (t) Examples from KESS sites 31-d lowpass filter p & p cm Difference (p-p cm ) is due to mapping-error and drift

drift-curve fit to P(t) - P cm (t) Examples from KESS sites 31-d lowpass filter p & p cm Difference (p-p cm ) is due to mapping-error and drift Fit exponential+linear to the difference Residual rms<0.01 dbar Resid slope<0.005dbar/yr

drift-curve fit to P(t) - P cm (t) Examples from KESS sites 31-d lowpass filter p & p cm Difference (p-p cm ) is due to mapping-error and drift Fit exponential+linear to the difference Residual rms<0.01 dbar Resid slope<0.005dbar/yr Old method fit (exp+lin) to hourly data (dotted line), w/o geostrophic leveling could differ by dbar

Digiquartz Pressure Drift (1) How well-suited is the (exponential + linear) curve-fit to the ‘processes’ causing observed drifts? –Test by comparing ‘same-site’ differences of raw records against the sum of fitted drift curves –Note ‘same-site’ pairs could be 0.1 to 0.4 km apart

Consistency of method at ‘same-site’ pairs in GoMex (rows 1, 2) sites with two PIES each, separately dedrifted (row 3) The difference between raw records (blue) agrees with difference between drift curves (red); rms < dbar (model 46K’s)

Consistency of (exp+lin) at ‘same-site’ pairs in Kuroshio (rows 1, 2) Two sites with two PIES each, separately dedrifted (row 3) The difference between raw records (blue) agrees with difference between drift curves (red); rms=0.004, dbar (model 410K’s)

The (exp + lin) curve suits the drift process well Next simply characterize the drift of each of many records by its (exp + lin) curve…

144 drift curves from 92 sensors +0.1 dbar/yr ref slope

Digiquartz Pressure Drift (2) 6000 psi ~ 4000 dbar sensors, model 46K psi ~ 6800 dbar sensors, model 410K Do drifts scale with FS range? What sign and magnitude of drift?

Model 46K and 410K sensor drift 46K’s - all pre-stressed410K’s - mixed pre-stressing

Model 46K (stretched) and 410K sensor drift 46K’s - all pre-stressed 410K’s - mixed pre-stressing 46K’s, vertical scale X (10/6) -Slightly less drift, but the improvement may have arisen from pre-stressing

Typical |drift| ~ 0.4 dbar Upward and downward drifts would arise from two different processes Exponential and linear drifts would arise from different processes So there must be at least 3-4 substantial contributions to drift Yeow!

Digiquartz Pressure Drift (3) Since the largest part of the drift decays exponentially with time, can pre-stressing the sensors decrease subsequent drift? –We usually pre-stress for many weeks or months –4000 psi for 6000 psi FS (~2800 dbar for ~4000 dbar FS) –6000 psi for psi FS (~4100 dbar for ~6800 dbar FS) How much might it help to pre-stress at nearly the same pressure as the subsequent deployment? Does sensor improve with age?

How effective is pre-stressing? (410K’s) Without pre-stressing, all six |drifts| ~ -.4 dbar in a year With pre-stressing many |drifts| <0.1 dbar (+/-) in a year, but many others drifted up or down ~0.4 dbar in 1-2 yrs pre-stressing seems highly advisable, but does not guarantee small drift

How effective is pre-stressing? (410K’s) Without pre-stressing, all six |drifts| ~ -.4 dbar in a year With pre-stressing many |drifts| <0.1 dbar (+/-) in a year, but many others drifted up or down ~0.4 dbar in 1-2 yrs pre-stressing seems highly advisable, but does not guarantee small drift Next represent drift as (end-start)…

Does it help to pre-stress at ~ deployment P? Answer maybe… excepting outliers

Does drift depend on prior deployment depth? difference (P new - P prior ) Answer is similarly murky… Slightly less drift with `same’ new and old deployment depths.

Does age of sensor reduce drift rate? (1-year) total drift vs. age

Digiquartz Pressure Drift (4) Does a given sensor drift ‘predictably’ from one deployment to the next?

Digiquartz Pressure Drift (4) Does a given sensor drift ‘predictably’ from one deployment to the next? Answer… the drift is not necessarily replicated in magnitude or direction! But we had many variables, and have not yet sorted out all effects. A few sensors DID replicate drift.

Digiquartz Pressure Drift (5) If you average two “same-site” pressure records, how well can this approximate a drift-free record? –Test by comparing near-neighbor averages of raw records against the accurately dedrifted curves Near-neighbors could be 0.1 to 0.4km apart

These 46K sites average very well because they had small drifts that fortuitously opposed each other.

These 410K site-pairs did not average to small drift.

Drift Characteristics of Paroscientific pressure sensors Summary … –(exponential+linear) fitted curve is well suited to represent the drift, to which 3 or more processes contribute –Geostrophic leveling & drift detection is reliable to ~0.01 dbar New drift curves differ from previous method of fitting data, although both use (exp + lin) fit –Averaging records from two or more ‘same-site’ sensors produces an average drift, but not zero drift –Pre-stressing reduces |drift| (usually) < 0.10 dbar / yr Small net drift helps reduce uncertainty in fitted drift curve –Aged sensors improve like good wine –Drift of a given sensor is not predictable from one deployment to the next Choose a low range sensor (when possible) to achieve smaller drift

FINI

Might add histogram of decay time scale

144 drift curves from 92 sensors +0.1 dbar/yr ref slope

IES deployment sites

CPIES array yields… A CPIES array yields daily maps of upper and deep circulation. Look-up tables interpret acoustic travel times as geopotential height (0 referenced to 5000 dbar). 2-D arrays of CPIES estimate horizontal gradients of geopotential to calculate geostrophic velocities. Velocity profiles are referenced by measured near- bottom currents. Bottom pressures are leveled using near-bottom currents to map the geostrophic streamfunction.

Leveling flowchart Detide pressure Dedrift pressure Current at CMs Pre-stress pressure gauges Leveled pressure constants Mean pressureMean current LEVELING PRESSURE GAUGE WITH DEEP CURRENT METERS

Pre-stress gauges Experience has shown that pressure drift is greatly reduced by preconditioning. Sensors are subjected to pressures of 3000 dbar for 1-2 months in the lab.

Mapping flowchart Mean p and current maps Mean p at PIES sitesMean current at CM sites p anomaly Current anomaly p and current anomaly maps Daily p and current maps Leveled p Current at CM sites

Leveling (and drift) The streamfunction from CMs and the pressure from PIES measure the same geostrophic pressure field. – The two fields should only differ by a site-dependent leveling constant. Other differences arise from error in OI mapped streamfunction and drift in the pressure sensor. The sensor drift is detected by the difference from the temporal record of geostrophic pressures The drift is represented by a decaying exponential plus linear curve, least-squares fitted to this difference

Consistency of (exp+lin) at ‘same-site’ pair in Kuroshio(2) (rows 1, 2) Another site with two PIES, separately dedrifted (row 3) The difference between raw records (blue) agrees with difference between drift curves (red); rms <0.005 dbar (model 410K)

Is drift smaller in 2nd/3rd deployment? 1st 2nd 3rd 46K’s410K’s

Subsequent depth shallower/ same/ deeper shallower same depth +/- 500m deeper 410K’s drifts are mixed Shallower 46K’s drift to higher P or “flat” Deeper 46K’s drift to lower P or nearly “flat” Same-depth 46K’s - slightly better? -0.1 dbar/yr

Does drift depend on prior deployment depth? difference (P new - P prior ) Slightly less drift with `same’ new and old deployment depths.

Repeatable drifts? 46K Twelve representative repeats of 46K’s having same Bliley and smallest depth differences between deployments. The repeatability of 46K’s looks promising. However next look at 410K’s - not as reproducible

Repeatable drifts? 410K These are 2nd deployments of 410K’s. Not as reproducible as for the previous set. Other factors changed –Deployment depth –Time interval between