Transcript Characteristics of Sea Level Records

Characteristics of Sea Level Records
Philip L. Woodworth
Permanent Service for Mean Sea Level
With thanks to Simon Williams
Ostende GLOSS Course 2006
Definitions:
• The expression ‘sea level’ means different things
to different people.
• Sea Level for us means:
The height of the sea measured relative to a mark
on the nearby land called the Tide Gauge
Benchmark
Sea levels measured by tide gauges at the Newlyn Lighthouse tidal
observatory are measured relative to the TGBM in the floor of the
building
Sea Level Measurements
• Sea level is measured by a tide gauge
• Values of sea level (or sea surface height) are
either spot-measurements at regular time
intervals, or averages (called integrations) over
the time intervals.
• For most tidal, storm surge or Mean Sea Level
work, time intervals of 5, 6, or 15 minutes are
adequate.
• For tsunami work, time intervals of a minute
or less are usually needed.
A Sea Level Record will show:
• The ocean tide
• On timescales of minutes to hours, there could
be signals due to seiches or tsunamis
• On timescales of hours to days, there could be
signals due to storm surges
• On longer timescales, there could be seasonal
and longer changes and even, at the longest
timescales, signals due to climate change
Tides
• Periodic movements which are directly
related in amplitude and phase to some
periodic geophysical force
• The dominant geophysical forcing function
is the variation of the gravitational field on
the surface of the earth, caused by regular
movements of the moon-earth and earthsun systems.
Tidal signals in March 1981 at
five sites which have very
different tidal regimes
Karumba : diurnal
Musay’id : mixed
Kilindini : semidiurnal
Bermuda : semidiurnal
Courtown : shallow sea
distortion
Lunar characteristics
responsible for these tidal
patterns
Frequency distribution of hourly tidal levels
Newlyn, semidiurnal
Karumba, diurnal
The Tide
To learn about tides – read chapters 2-5 of
David Pugh’s book
Main Points about the Tide to make here:
• Tidal characteristics at any location can be
determined well from a year of good tide
gauge data
• The knowledge obtained can be used to
produce Tables of Predicted Tide Levels (‘Tide
Tables’) for local use
• Tidal and Non-tidal components of the sea
level record can be separated easily using
Harmonic Tidal Analysis of a tide gauge data
set.
Non-Tidal Changes
• Storm surges caused by changes in air pressure
and winds
• Seiches due to resonant behavior of harbours
and bays (caused by many things, often wind)
• Seasonal, interannual changes due to changes
in water density
• Tsunami caused by earthquakes
• And many more non-tidal sources.
Mean Sea Level
• Daily MSL is calculated by applying a suitable
filter (see IOC Manual III for details) to the
measured tide gauge data (i.e. not the tidal or
non-tidal parts separately, but the measured
values)
• Monthly MSL is then defined as the arithmetic
average of the daily MSL values in that month
• Annual MSL is defined as the average of the
daily MSL values in a year
Typical analysis of monthly mean sea levels
PA
PA
Z 0 t   Z 0  at  N  S a  S sa  b0 PA  b1
 b2
 et 
x
y
Long term mean
and trend
Periodic tidal
terms (nodal,
annual and
semi-annual)
Meteorological effects
residuals
Permanent Service for Mean Sea Level
• Time series of MSL are collected by the
PSMSL
• Long records show signals due to seasonal
processes, El Nino, and sea and land level
changes
• One goal in ODINAFRICA is to eventually
have long records of MSL from more African
stations
Sea Level Changes in Last 100 Years
Past 100 years
• Most PSMSL records show
evidence for rising sea levels
during the past century
• IPCC Third Assessment
Report concluded that there
has been a global rise of
approximately 10-20 cm
during the past 100 years
Types of Sea Level Records
Real-time. Data are often transmitted to a warning centre
which can look out for flooding due to storm surges or
tsunamis. Delay (latency) has to be as short as possible so
there is no time for detailed inspection of the record
• Delayed Mode. Delay is not a major issue. Data are
archived by a centre for subsequent detailed analysis and
removal of errors. Such highest-quality data are needed for
scientific research, for production of a range of products
(e.g. tide tables), for computation of MSL etc.
There are almost 2 different communities of people
involved: Operational agencies, Scientists/Engineers
Why do we care about tidal analysis if we just want to
provide real-time data and monthly MSL (neither of
which require tidal analysis in theory)?
Two main reasons:
1. The separation of the sea level record into tidal and nontidal components is needed if one needs to produce tide
tables or tidal predictions
2. The non-tidal signals (seiches, tsunamis) become clearly
identified
3. Tidal analysis also enables errors in the sea level time
series to be identified  highest-quality data for
subsequent analysis
How does a tidal analysis work?
• The tide is parameterised in terms of
harmonics with periods specified by the orbits
of the Moon and Sun but with unknown
amplitudes and phase i.e.
Pi t   Z 0 
Tide
N
H
j 1
j


f j cos  j t  g j  V j  u j 
The unknown parameters are Z0 and the (Hj, gj).
The fitting is adjusted so that the sum of the squares of
the difference between the observed and computed tidal
levels is minimized. The residuals to the fit are
considered to be the ‘non-tidal’ terms
An example – maybe not the best but one I had available – from
Port Stanley Nov-Dec 2004
Measured
Tide
Non-tidal
In the non-tidal record:
1. Variations of a few days, amplitude of a few cm (this is southern
summer so no big storm surges)
2. A lot of high-frequency noise of a few cm due to harbour seiches
3. On 27 December arrival of the Sumatra tsunami (15 cm or so)
 None of this is evident from looking at the total measured record.
•
The non-tidal component is often called the Residual (or
sometimes the ‘tidal residual’).
•
It will be immediately clear (especially with some
experience) by looking at the residuals if there is:
1. A spike or jump in the data due to instrumental faults
2. A timing error due to a section of data being given the
wrong time
3. And many other errors  see the IOC Manuals
These errors can then be fixed in the data set. The final data
set is called the Quality Controlled Delayed-Mode data
set.
Such tidal analysis can be undertaken using one of the
software packages available for the GLOSS and
ODINAFRICA programmes, see
http://www.pol.ac.uk/psmsl/training
In this course we will be using the TIDE TASKS FOR
WINDOWS package.