Ocean Sciences

Atlantic Canada’s Real-Time Water Level System Observations, Predictions, Forecasts and Datums on the Web

Atlantic Canada’s Real-Time Water Level System Observations, Predictions, Forecasts and Datums on the Web

Authors: Phillip MacAulay (CHS) macaulayp@mar.dfo-mpo.gc.ca Charles O’Reilly (CHS) oreillyc@mar.dfo-mpo.gc.ca Keith Thompson (Dalhousie) keith.thompson@dal.ca

)cos()sin(000

)sin()cos(000

00)cos()sin(0

00)sin()cos(0

00001

22

22

11

11

M

]01011[

]02021[

H

K Tttt KxMsKHIs 1)(

+

forecast forecast

3D Continuous datums

PWLNPWLN

Dalcoast II

observe, collect, predict

WebTideWebTide

LAT wrt MWLLAT wrt MWL

tt Hsx

zero phase shift, multi band pass, recursive filter

zero phase shift, multi band pass, recursive filter

RTWL System

Ocean Sciences

CHS Atlantic’s Permanent Water Level NetworkCHS Atlantic’s Permanent Water Level Network16 tide gauges, 000’s kms of coastline16 tide gauges, 000’s kms of coastline

Long Term Sea Level (GLOSS)Long Term Sea Level (GLOSS)

Tsunami GaugesTsunami GaugesStorm Surge Gauges

0990

0905

0755

17001805

2000

1430

06650835

0612

04910490

0365

0065

2145

1680

Presentation Subtext: Socio-economic and Hydrographic motivations for maintaining coastal water level monitoring programs and networksPresentation Subtext: Socio-economic and Hydrographic motivations for maintaining coastal water level monitoring programs and networks

Ocean Sciences

•New Versatile Sutron XPert Dataloggers•Backup Communications

GOES Satellite Systems (need 10 min permission) •High Frequency Water Level Measurements and Gauge Polling

1 minute averaged sampling, 10 minute gauge polling, average 5 min data latency

•Sutron XConnect real time data collection software

•Redundant Sensors Primary Rotary Encoder (Stable, Accurate, Reliable)Secondary Bubbler (Stable, Secondary Backup)Tertiary Pressure sensor (Extreme Water Levels, Fast Response Wells, Tertiary Backup)

Laser sensor (Stable, Highly Accurate, Direct Independent Measurement Testing for GLOSS sites)

•New Versatile Sutron XPert Dataloggers•Backup Communications

GOES Satellite Systems (need 10 min permission) •High Frequency Water Level Measurements and Gauge Polling

1 minute averaged sampling, 10 minute gauge polling, average 5 min data latency

•Sutron XConnect real time data collection software

•Redundant Sensors Primary Rotary Encoder (Stable, Accurate, Reliable)Secondary Bubbler (Stable, Secondary Backup)Tertiary Pressure sensor (Extreme Water Levels, Fast Response Wells, Tertiary Backup)

Laser sensor (Stable, Highly Accurate, Direct Independent Measurement Testing for GLOSS sites)

Patm

1) Legacy Sensor: Continuous chain float and pulley with optical encoder and tungsten

weight, specific gravity =18

2) Secondary Sensor: Continuous fast response bubbler

3) Backup Sensor:Submersible pressure sensor

4) Laser sensor:

000

o

o

o

o

pump

P=wgh

h

o

P=wgh

PWLN Equipment and Operational Upgrades PWLN Equipment and Operational Upgrades

Who is Crazy?

Ocean Sciences

Schematic of the Atlantic RTWL System A Collaborative Effort

Schematic of the Atlantic RTWL System A Collaborative Effort

Data Acquisition PC (XConnect)

Water Level Clients

PWLN Tide Gauges

DFO Science(Matlab, Modeling)

RTWLOracle Dbase

ISDM

XConnect Backup

Archival Data

CHS XP Network

(XConnect)

DFO web server,RTWL web pages

Real time data

Data Backup

GOES Satellite

link

GOES Satellite Download

Archival Data

CHSDFO ScienceISDMDalhousieClients

Dalhousie Dalcoast II

Ocean Sciences

RTWL Web PagesRTWL Web Pages

Kohila Thana, Science InformaticsKohila Thana, Science Informatics

Ocean Sciences

Examples of other Real or Quasi-Real Time SystemsExamples of other Real or Quasi-Real Time Systems

• National Oceanographic and Atmospheric Administration’s (NOAA) Tides and Currents web site ----Physical Oceanographic Real-Time System (PORTS)

• European Sea Level Service (ESEAS)

• Monitoring Network System for Systematic Sea Level Measurements in the Mediterranean and Black Sea (MedGLOSS)

• UK National Tidal and Sea Level Facility (NTSLF)

• National Oceanographic and Atmospheric Administration’s (NOAA) Tides and Currents web site ----Physical Oceanographic Real-Time System (PORTS)

• European Sea Level Service (ESEAS)

• Monitoring Network System for Systematic Sea Level Measurements in the Mediterranean and Black Sea (MedGLOSS)

• UK National Tidal and Sea Level Facility (NTSLF)

Ocean Sciences

Observations, Constituent Predictions, The ResidualObservations, Constituent Predictions, The Residual

• Weather (storm surges, positive and negative)

• Departures at tidal frequencies (variations in the tidal constituents)

• Everything else affecting water level variations (infra-gravity waves, harbour seiches, tsunamis, shelf waves …)

• Weather (storm surges, positive and negative)

• Departures at tidal frequencies (variations in the tidal constituents)

• Everything else affecting water level variations (infra-gravity waves, harbour seiches, tsunamis, shelf waves …)

The Residual: Observations – Predictions (non-tidal behavior)The Residual: Observations – Predictions (non-tidal behavior)

Timescales (minutes to months) Reason for continued monitoring of coastal water levels (Presentation Subtext)Immediate impact on coastal environments, populations, and infra-structureDanger to navigationValidate storm surge models

Timescales (minutes to months) Reason for continued monitoring of coastal water levels (Presentation Subtext)Immediate impact on coastal environments, populations, and infra-structureDanger to navigationValidate storm surge models

North Sydney, NSSt. Lawrence, NFLD

Ocean Sciences

Forecasting the ResidualForecasting the Residual• Effects of weather Dalcoast II -- 2D, barotropic, non-linear, lateral diffusion,

1/12o , forced by EC 3hr winds and pressure (48hr surge forecast, once per day)

• Everything else Spectral-nudging Kalman Filter (Thompson etal., 2006) -- zero phase shift, multi band pass, recursive filter

• Effects of weather Dalcoast II -- 2D, barotropic, non-linear, lateral diffusion,

1/12o , forced by EC 3hr winds and pressure (48hr surge forecast, once per day)

• Everything else Spectral-nudging Kalman Filter (Thompson etal., 2006) -- zero phase shift, multi band pass, recursive filter

Bobanovic and Thompson, 2001

)cos()sin(000

)sin()cos(000

00)cos()sin(0

00)sin()cos(0

00001

22

22

11

11

M

]01011[

]02021[

H

K Tttt KxMsKHIs 1)(

st denote an auxiliary state vector x1, x2, ….,xt equispaced scalar values to be filtered

Tt ffK ]0)(20)(21[ 2211

Build TermBuild Term Decay TermDecay Term

Proto-Forecast = Constituent Predictions + timeseries of 0-24 hr surge + 24-48 hr surgeForecast model residual error = Observations (up to last) – (Proto-Forecast) Residual error Nowcast = Filter (forecast model residual error), *Nowcast mode, build/modify state vector energyResidual error forecast = Filter, *Forecast mode, project existing state vector energy into the future (no decay)Final forecast = Proto-forecast + residual error nowcast and forecast

Proto-Forecast = Constituent Predictions + timeseries of 0-24 hr surge + 24-48 hr surgeForecast model residual error = Observations (up to last) – (Proto-Forecast) Residual error Nowcast = Filter (forecast model residual error), *Nowcast mode, build/modify state vector energyResidual error forecast = Filter, *Forecast mode, project existing state vector energy into the future (no decay)Final forecast = Proto-forecast + residual error nowcast and forecast

tt Hsx

.

1 tt Mss

Filter band pass width, gain, (spin up period)-1

Filter band pass width, gain, (spin up period)-1

Filtered valuesFiltered values ForecastForecast

NowcastNowcast Tide gauge locations

Bias term

Band Pass Frequencies

Ocean Sciences

Selection of Filter Frequencies and GainsSelection of Filter Frequencies and Gains

• From Theory: Diurnal O1, Semidiurnal M2 gain/filter band widths wide enough to capture other diurnal and semidiurnals

• From Recent Observations (analysis of forecast model residual error timeseries): Upper Tidal and Seiche Frequencies, gain/decay timescales ~ 1.5 to 2 wave periods

• From Theory: Diurnal O1, Semidiurnal M2 gain/filter band widths wide enough to capture other diurnal and semidiurnals

• From Recent Observations (analysis of forecast model residual error timeseries): Upper Tidal and Seiche Frequencies, gain/decay timescales ~ 1.5 to 2 wave periods

Forecast model residual error Seiche content

Seiche energy/frequenciesResidual tidal energy/frequencies

Ocean Sciences

Example ForecastExample Forecast

Past

Future

DatumsDatumsOperational code under testing,

Available on RTWL web pages soonOperational code under testing,

Available on RTWL web pages soon

Ocean Sciences

Continuous Datums for Atlantic CanadaContinuous Datums for Atlantic Canada

Discrete CalculationDiscrete Calculation

Gridded SurfacesGridded Surfaces

Dynamical ModelingDynamical Modeling

• What are they? Where are we starting from?

• Why do we need them?

• Where are we in Atlantic Canada?

• How should we now best proceed?

• Planning, Development Stage Although in principal prototypical versions have already used

06-07 Labrador route survey Webtide 07 Northumberland Strait Survey Webtide 07 Fundy Survey Webtide, Omnistar

• What are they? Where are we starting from?

• Why do we need them?

• Where are we in Atlantic Canada?

• How should we now best proceed?

• Planning, Development Stage Although in principal prototypical versions have already used

06-07 Labrador route survey Webtide 07 Northumberland Strait Survey Webtide 07 Fundy Survey Webtide, Omnistar

Ocean Sciences

Sea bed

Spatially variable tidal properties

2D Simplification Ellipsoid

Long Timescale WL MonitoringVertical Control Datum Management

Long Timescale WL MonitoringVertical Control Datum Management

• Hydrographic reason to maintain a minimal, effective water level monitoring network/program, maintain/track tidal datums wrt relative sea level rise

(Presentation Subtext)• Present and past approach, What are continuous Datums? Constant Datum Transform Assumption (CDTA)

• Hydrographic reason to maintain a minimal, effective water level monitoring network/program, maintain/track tidal datums wrt relative sea level rise

(Presentation Subtext)• Present and past approach, What are continuous Datums? Constant Datum Transform Assumption (CDTA)

+ BM

+ CD(xxxx)

+

+

Stn 0491 Lat, Long

DEV(xxxx)

MWL(xxxx)

Transforms

++ + +

+

+

Discrete Vertical Control,Tide stations

datum

target ambiguity

Growing systematic errors

Continuous Surface Honoring Discrete Control

mwl

Tid

al r

ange

Hat

Lat

Geodetic

+<30 cm

Ocean Sciences

Why do we need them?Why do we need them?• With GPS based vertical positioning, RTK or modeled (Omnistar), errors

introduced to charts by the CDTA can become significant part of vertical error.

• CDTA can be inappropriate some activities:

– route surveys

– surveys where tidal target surfaces change significantly over survey scale (estuaries, Bay of Fundy)

– Flood mapping of large areas

– RTK solutions and Anywhere, anytime water level

applications (WebTide), Continuous applications need

continuous datums

• With GPS based vertical positioning, RTK or modeled (Omnistar), errors introduced to charts by the CDTA can become significant part of vertical error.

• CDTA can be inappropriate some activities:

– route surveys

– surveys where tidal target surfaces change significantly over survey scale (estuaries, Bay of Fundy)

– Flood mapping of large areas

– RTK solutions and Anywhere, anytime water level

applications (WebTide), Continuous applications need

continuous datums

WebTideWebTide

Constituents derived from a Hydrodynamic Barotropic Ocean Model Assimilating Topex Posieden Altimeter dataConstituents derived from a Hydrodynamic Barotropic Ocean Model Assimilating Topex Posieden Altimeter data

Dupont, F., C.G. Hannah, D.A. Greenberg, J.Y. Cherniawsky and C.E. Naimie.  2002.  Modelling system for tides for the North-west Atlantic coastal ocean. [online]. [Accessed 21 April, 2008]. Available from World Wide Web: http://www.dfo-mpo.gc.ca/Library/265855.pdf

Dupont, F., C.G. Hannah, D.A. Greenberg, J.Y. Cherniawsky and C.E. Naimie.  2002.  Modelling system for tides for the North-west Atlantic coastal ocean. [online]. [Accessed 21 April, 2008]. Available from World Wide Web: http://www.dfo-mpo.gc.ca/Library/265855.pdf

Multi-constiuent, anywhere tidal prediction

Ocean Sciences

Distance weighted datum transform MWL to CD (Discrete calculation)

Distance weighted datum transform MWL to CD (Discrete calculation)

Exponent p determines local flatness and width of transition zoneExponent p determines local flatness and width of transition zone

Webtide provides anytime, anywhere tidal prediction wrt to a floating MWL We want tidal prediction wrt CD, need anywhere transform between MWL and CD

Webtide provides anytime, anywhere tidal prediction wrt to a floating MWL We want tidal prediction wrt CD, need anywhere transform between MWL and CD

Used in: • Labrador 06-07• Northumberland 07• Fundy 07

Used in: • Labrador 06-07• Northumberland 07• Fundy 07

1, jjji wwtT

j

pj

pj

j distfar

distfarw

)(

)(

j

j

j

j

i

coastline

Coastal datum control points

distj

far=max(distj)

Ti – transform at itj – transform at jwj – weight for tj

Location transform required

Herman VarmaHerman Varma

Ocean Sciences

How should we best proceed?How should we best proceed?

• Closely examine the programs and methods employed by others, NOAA/NOS, UK (Excellent Workshop Monday)

• Develop plans of approach (brainstorm, communicate needs/requirements

• Discuss options with Datum clients

• Develop project outlines/proposals

• Acquire necessary resources and get on with it

• Closely examine the programs and methods employed by others, NOAA/NOS, UK (Excellent Workshop Monday)

• Develop plans of approach (brainstorm, communicate needs/requirements

• Discuss options with Datum clients

• Develop project outlines/proposals

• Acquire necessary resources and get on with it

Ocean Sciences

How should we best proceed?Concept outline for an Atlantic Plan

How should we best proceed?Concept outline for an Atlantic Plan

1. Develop an initial theoretical target surface

example: LAT as Sum of WebTide constituent amplitudes

1. Develop an initial theoretical target surface

example: LAT as Sum of WebTide constituent amplitudes

WebTide is based around a floating MWL, thus a surface computed from it must then be referenced to observed mean water levels. MWL is better known at some shore based locations than others. Away from the coast we must rely on either satellite altimetry or Geodetic models.

WebTide is based around a floating MWL, thus a surface computed from it must then be referenced to observed mean water levels. MWL is better known at some shore based locations than others. Away from the coast we must rely on either satellite altimetry or Geodetic models.

Ocean Sciences

How should we best proceed?Concept outline for an Atlantic Plan

How should we best proceed?Concept outline for an Atlantic Plan

2. Compare existing shore-based Datum reference points to theoretical target surface. There will be misfit. Find and identify poor reference control, iteratively adjust target surface to more closely honor high quality shore-based reference control. Iterations between steps 1 and 2 may be necessary.

2. Compare existing shore-based Datum reference points to theoretical target surface. There will be misfit. Find and identify poor reference control, iteratively adjust target surface to more closely honor high quality shore-based reference control. Iterations between steps 1 and 2 may be necessary.

Ocean Sciences

How should we best proceed?Concept outline for an Atlantic Plan

How should we best proceed?Concept outline for an Atlantic Plan

3. Adjust or supersede the target surface where good surrounding shore-based reference Datums permit independent surface estimation. Sequester inshore areas requiring further data or modeling. (NULL datum concept)

4. Calculate the seamless Datum surface grid. Grid spacing could be variable based on the local rate of change of the Datum surface.

5. Agree on standardized methods for interpolation between datum surface grid points and thus calculation of all Datum transformations.

6. Transform Dissemination tools?

For Hydrographic work? Public? (VDatum?)

7. Border transitions? regional (Atlantic/Quebec), international (Atlantic/US)

3. Adjust or supersede the target surface where good surrounding shore-based reference Datums permit independent surface estimation. Sequester inshore areas requiring further data or modeling. (NULL datum concept)

4. Calculate the seamless Datum surface grid. Grid spacing could be variable based on the local rate of change of the Datum surface.

5. Agree on standardized methods for interpolation between datum surface grid points and thus calculation of all Datum transformations.

6. Transform Dissemination tools?

For Hydrographic work? Public? (VDatum?)

7. Border transitions? regional (Atlantic/Quebec), international (Atlantic/US)

Combination of the 3D Datum surfaces with WebTide will yield an anytime, anywhere, on Datum water level prediction/sounding reduction tool.

Combination of the 3D Datum surfaces with WebTide will yield an anytime, anywhere, on Datum water level prediction/sounding reduction tool.

Ocean Sciences

• Savi Narayanan, CHS directorate, direction and funding

• John Loder, expertise on tsunami gauge locations

• John O’Neill and Kohila Thana, system and web design

• Mike Ruxton, Craig Wright, Larry Norton, Mark McCracken, James Hanway, CHS Geomatics

• Steve Parsons, Herman Varma, Chris Coolen, Fred Carmichael, CHS Atlantic

• CHS Quebec, Pacific and Central and Arctic

• Carmen Reid, Canadian Coast Guard

• Savi Narayanan, CHS directorate, direction and funding

• John Loder, expertise on tsunami gauge locations

• John O’Neill and Kohila Thana, system and web design

• Mike Ruxton, Craig Wright, Larry Norton, Mark McCracken, James Hanway, CHS Geomatics

• Steve Parsons, Herman Varma, Chris Coolen, Fred Carmichael, CHS Atlantic

• CHS Quebec, Pacific and Central and Arctic

• Carmen Reid, Canadian Coast Guard

AcknowledgementsAcknowledgements

Ocean Sciences

Higher High Water High

Water

Lower Low Water

Low Water

MHHW

NOS Generation of Tidal Datum Fields NOS Generation of Tidal Datum Fields

• • MHW tidal datum fields (as well as MHHW, MLW, MLLW, MSL, MTL, DTL) from calibrated hydrodynamic models

• • Analysis of model-produced time series, then adjusted to provide a best fit to datums at NOS gauges.

Ocean Sciences

Vertical Datum Transformations in VDatumVertical Datum Transformations in VDatum

Orthometric DatumsNAVD 88 North American Vertical

Datum 1988NGVD 29 National Geodetic

Vertical Datum of 1929

3-D/Ellipsoid Datums

NAD 83 (NSRS) North American Datum 1983WGS 84(G873) World Geodetic System 1984 (G873)WGS 84(G730) World Geodetic System 1984 (G730)WGS 84(orig) World Geodetic System 1984

(original system -- 1984)WGS 72 World Geodetic System 1972ITRF International Terrestrial Reference Frame

1988-94, 1996-98, 2000SIO/MIT 92 Scripps Institution of Oceanography /

Mass. Inst. of Tech. 1992NEOS 90 National Earth Orientation Service 1990PNEOS 90 Preliminary Nat’l Earth Orientation Service 1990

Tidal Datums

MLLW Mean Lower Low WaterMLW Mean Low WaterLMSL Local Mean Sea LevelMTL Mean Tide LevelDTL Diurnal Tide LevelMHW Mean High WaterMHHW Mean Higher High Water

Ocean Sciences

Conversion value at any location is determined by interpolation using surrounding values. For geodetic grids, set of nine points are used with bi-quadratic interpolation.

For points in tidal grid, two methods are used. If four surrounding VDatum points are non-null, bilinear interpolation is used. If 1-3 points are null value, inverse-distance-squared weighting is used for non-null values.

VDatum Interpolation to the User Input LocationVDatum Interpolation to the User Input Location

Ocean Sciences

RTWL Initiatives 2006/07RTWL Initiatives 2006/07

• Link Matlab to RTWL database and web pages─ Employ Matlab’s powerful computational engine for RTQC─ Interactive web content using Matlab

• Real time quality control (RTQC), tsunami detection, and water level forecasting ─ Reliability of Nowcast data on the web─ Identification of tsunami signals─ Water level forecasts (storm surge model)

• Observation-based tsunami detection and warning enhancement (Sable/Hibernia)─ Early detection/verification─ Landfall wave height estimation, Zhigang Xu (IML)

• Link Matlab to RTWL database and web pages─ Employ Matlab’s powerful computational engine for RTQC─ Interactive web content using Matlab

• Real time quality control (RTQC), tsunami detection, and water level forecasting ─ Reliability of Nowcast data on the web─ Identification of tsunami signals─ Water level forecasts (storm surge model)

• Observation-based tsunami detection and warning enhancement (Sable/Hibernia)─ Early detection/verification─ Landfall wave height estimation, Zhigang Xu (IML)

Sable

Hibernia

Ocean Sciences

Storm SurgesStorm Surges

~1 meter storm surges during spring tides, Feb/01/2006Potential Flooding~1 meter storm surges during spring tides, Feb/01/2006Potential Flooding

Ocean Sciences

Simulated Tsunami ArrivalSimulated Tsunami ArrivalSimulated tsunami signal, Zhigang Xu (IML)Simulated tsunami signal, Zhigang Xu (IML)