RECENT MEASUREMENT
AND ANALYSIS TECHNIQUES
DEVELOPED AT N.I.O.
by
M. J. Tucker
Conference ori Ocean Wave Spectra
Eaa ton,
Maryland
May 1-4,
1961
RCT
SUE7 AND ANALYSIS TECiNIQUEs
DEVLQP
LT N.I.O.M.J. Tucker
Ntional Ins bitute of Ocuoaphy
Worniley, Surrey, 1gland.
Introduction
N.I.O. and its
predecessor, the Oceanographic Group of the Admiralty Research Laboratory, have devoted considerable ef f ort over a period of sixteen years to waveresearch.
In
the field of technique, our efforts to record and aza1yse waves for research purposes have on the whole been successful: though "he who never reade s mistake never made anything", end we have made our share ofmistakes.
Our efforts to provide wave recorders for civil engineers haveon
the whole been less successful,though this is a difficult problem and the number of really successful wave reocrdor dvaloped for this application anywhere
in
the world,can be
ooited on the fingers of one hand, leaving perhaps three or four spare fingere. very civil engineeringinstallation
presents different problems, and one may devote a lot of effort toover-ociing
these inone case, only to find a new
and equally difficult set facing one inthe
next.
A further difficulty is that the staff on the site, particularly when this is overseas and one cannot spe*k to there personally, are often synathetio to the equiçnent, and.get
annoyed and. lose interest at the first diffIculty. Probabi,ymost installations use bottcxn-pressuro
meter
type wave recorders, and in inoderete winds producing short waves, those give apparently falserecordings
owing to the attenuation of the shorter waves with depth. Correction using the apparent period and. attenuation factorsfrczn
tables is littlehelp:
the apparent period of the waves on the surface may be four seconds, end cii the recording six seconds.Use f the
factor corresponding to either of these periods will give the rong answer. Th,e
reaction of the
engineer on the site, 'suo doesn't believe in instruments anjway, is to say "It's no good", arid loso interest.&ver the years we have
built up a
number of ideas on how to set about measuring and analysing waves, and theso will be set out below. They owe a lot, of course, to other people's ideas and experience.inoiples governing the development of 'save iae3suring instrumente
In the desi of instruments, two factors of fiidmental importance sometimes
get
overlooked.
Thesearo that the
user must have confidence in the accuracy of the measurements, and that the easier the analysis of the results, the more likely they are to got used. In the oase of wave-recording, both thase factors point to the measurement of absolute height (or pressure) because recording of a tidal cycle willthon give a convincing check on
the
calibration of the equipment, and there is no rieoesaity forany correction of the results
for the variation of the response of theinstrument itself with frequency, though hydro&ynomio response factors will usually still need correction. Such an instrument also has the advantage that if it is
sificiontly accurate, the long 'savez
(of several minutes' period) mey be filtered fran its output and recorded separately. It is not, of course, practicable to record absoluto height on the deep sea whore no fixed pintfonn is available as a reference.Fixed aerisitivity is deireble, since thon
the recorda are not
wasted ifcoone
forgets to noto the setting
of
the itivLtycontrol, end the r300rcls do not go of?
ecale by mistake.
This may necezsitcte high resolution recording to
obtain adequate
precision over thepossible range of
vo-heigJt to be mussurod, but thisrange is
not usually v'ry grec t.In a shore-based installation, the
biggest single item of
epanditure is often the oost of the cable,and if it ha3
tohe specially manufactured, the
time reqir'et
for this can be the factor
determining the earliest pozsible d.ate of installation.It is therefore an advantage if the
thatrznent requires
cnly a simple co-axial or2-core oable,
Such cables aro relatively cheap and
can often beobtained "off the
ahe].f",
Those last
two considera tions indicate a measuringhead
ihioh gives a frequencydepe4ent on the
wave-height, that is, anf.m. system.
Prequonoy can be rnoasured
and recorded withextremely
high precision; two ocres suffic'e tooar' both the power supply to the
measuring
headand
the output signal backto the shore;
in addition
the characteristics of the cable do notaffect the oalibration,
and the output can be tolometoredby radio if necessary.
For roi.rtine wave-recording,
self-contained wsvo recorders, that is, inatrnts
laid on the
sea-bed for several weeks, or possibly months, at a timo, end recording ithin themselves, hava been f o.nid to be unsatisfactory.ly too often it te found
on recovery that sc*xeththg has gone wrong with
the device and the records ars lost.Such a loes may spoil a year's set
of records. For SOEDO research purposes where Thewave recorder is only in op-atìon for
3 short time
before it is recovered ath the recorda exsmined, this is not30
important, but is still a siìifioant factor.For fixad instalistions,
instrtzuants
which measurethe
variations in pressure as the waves peas overhead are still th3 rost practicalin most oiro'ixistanocs, in
spite of the severe disadvantage f their insblity to measure short wsvea
..etiafao-tori].y. This is partly
raJC&U.$O of the great oxperae of insta).ling structures
capable of supporting on ìnctr.aentp 8ciflg
bhrDugh the water curfooe, endpartly
duo to the leak of
roci),y sctisfotory s'rfaoci-}ìeight recarder
capable of handling
the range of lev1a required, which rca:c be sa.'enty feet in an exposed situation
(waves plus tides). Foi' domo otvil. cngine:ring tpplications, the
insensitivity to short waves may
be an advantage, aince it i
sciotimaa only
the longer waves which
are importent.
For near-shore reoord.ing, then,
it eraa that on sbolute pressure gauge, mo'.tted a alose to the surface as practicable, with an f.m.output, and oonncoted by cable to
a recorder
ashorc,i3 likelyto be the roost satisfactoxy instrtxnont for
general use.
As far as the author ±s aware, Snodgress, Munk end
Tucker
(1958)2t the Scripps
Tnstitution of Oceanography were the first to develop
such e system.
For recording 'nves on deep water, teeny people have
tried using the still water
below the action of the waves
as a
reference, Such3StCflLS typically have either
a
surface-height measuring device held steady by
a drogue
in deep vater, or arrange fora buoy on the surface to raise and lacier a pressure matar or integrating
current-meter suspended in the deep water.
In order ta obtain reasonable
responso to the
longer waves, such a system
requrea a
sUkpL3nBiÒn at least 100
s.
Long.
'Ñ have
-3-fod such systems to be
atifctory.
The equiçznent is usually clumsy and
difficult to handle in stormy 1eb
4i.ìi
xm&y bQi.hject to sppx'eoiable
stretch;
difforeritial currents, wind dr3g, or drag through the cable cornisoting the
instri.znent to the ship cn cause the viro to tilt from thu vrtiol, introducing
drift into the record and unknown factors into the
responce.The most aatisfaotorj doap-su v.evo recorders so
fardeveloped are those which
measure the motions of a buoy.
In their simplest form these contain a vertical
gooelsroter measuring the oomponent of acceleration perpendicular to the vater
surface, or in the direction of the resultant acceleration, vthich is, in principle,
the sama thing.
Better accuracy con be obtained if the aoco1eroneter is motmted on
roocpe so that it measures the oomporiunt of acceleration in the triia vertical.
The silo can either be integrated truce before recording to give wave elevation,
or if the poier spectrum is being obtained, the acceleration spectrun can be
deter-mined and ccnverted to the elevation spectrum by multiplying by 1/cr'.
By also
measuring pitch and roll, a useful amount of information can be obtained about the
directional speotrus of the waves (see Cartvuright'z paper presented to this oonf er-.
ence).
Suchsystems are eminently praotiooble, and can be considered 85 fully
establishßd.To obtain
moreifozation about the directional speothri is d.ifficult, and
N.I.O. hes not yet succeeded in producing any
equipment which willdo this.
The
problem is, of course, receiving a
'oat deal of attention by workers in several
countries.
The f,m. pressure recorder
This is a pro asure-reatar typo wave recorder crinected to the shore by a cable.
A diagaei of the measuring head is ahocn in figure 1
The pressure acts on a
disphragn carrying one plato of a parallel ploto capacitor.
TheparntiOE, of the
capacitor plates at zero pressure is 0.015 inch and is reduced to 0.0015
inch at apressure corresponding to 200 lt. of water.
This capacitor is the timing ospacitor
of an L.0. oscillator, ïhoso frequency is thus a measure of pressure.
The measuring-head circuit is shown in figure 2.
The paver supplies are
feddown a twin-core or co-axial cable.
The oscillator is not deooi.led, so that the
supply current fluctuates and approximately 100 mV of sial is devaloped.,aoross the
680 load resistance at the shore end (if the cable is short enough to produce little
attonuatixt).
The oscillator has bean desi-ìod to give maximum stability against changes in
terxorature and supply voltage.
Tho overall temperature cocf'iciont of the
measur-ing hcad at o'bnospheria
pressure is approximately I pt.in iO
of full scolo
(corresponding to approximately 0.25 inch of rater) per °C.
It ha
not boon possibl3
to measure it at other pressures, since we have not so far boen able to maintain
sufficiently constant pressures.
Supply voltage changes produce a negligible effect.
The calibration o'o of frequency against pressure is not linear, but over a normsl
working range it may be regarded as linear for most practical purposes.
The froquency-iueaauring arrangement used for routine wave rocording is shown in
In this inatnxnent en air-filled rubber
ag is fixed to the sea-bed, or suspondßd
a fixed distance above it by a float.
The sir iii the bag takes up the pressure of
the water around it, which v-aries as the waves pass overhead.
The bag is oonneotod
by s hosepipe to a pressure recorder ashore.
The principle is simple and attractive, not least because no power aplies ero
n006ssary.
However, a ntxnber of unexpected snags arise in praotioo, mainly in
octineotion with the design of the sec unit and with the arrangements for f illin,g it
with the correct amount of oir.
(e might cite the operators in the Persian G&2)Ì,
who, when the temperature is 120°F in the shade, find that the effort required to
ptp
upthe sea unit using a hand pup is more than they can manage.
N.I.O. la
a lowly overcoming the problema, but can't o la im tha t the ins truent is I 0()
satisfactory yet.
Por unattended intermittent c)er3tion In remoto places, a synchronous-motor
chart drive is usod, driven from a transistor oscillator, arxd
witohed by a
spring-drive proanng olook.
Wave-moe suing bucys
A desoripticxi of the prinoiplo and practice of wave measurement using the N,I.O.
bucys is given by Longuet-Higgins, Cartwright and Smith, and an example of their
use
in a practical problem
is given by Cartwright, in papers presented to this oorEerenee,
ily a very
briefdescription will therefore be given her..
The self-contained buoy is ca st in alimiinitzn, is circular,
about aix feet in
diarmetez, and one foot deep.
It contains batteries, a vertical
accelerometer,pitch and
roll gyros, electronics,a multi-channel photographic
recorder and eprcgrarxming clock.
It is lined up in the wind direoton by means of a
nall drcg
attached to it by a rope bridle. Ithas
been usedsatisfactorily in winds up to
Force7.
Tboui this buoy is a fully praotioal device and
has been useds great deal, it
is rather large and heavy to handle, and in cQnmon with all se f-contained devices,
since the use of transistors presents some problems (for emple, the iiedanoea in
the wave/tide filter are too high for
'ensistor circuits).
o advantage of this inatrncnt is its flexibility.
An R.C. oscillator using
thermistors in the frequency-determining nctwork h3 boon connected in parallel with
the pressure oscillator.
Thus, e frequency dependent on temperature also appears
across the 68c load resistor and is arranged to be a l
frequency so that it may be
filtered out and measured.
Long waves may bu acasured by connecting a suitable
filter and output stago to the output of the diodo-punp frequency meter:
such a
system is 8t present in use and appears to be satisfactory.
For research purposes,
the output ay be recorded digitally using a counter-typo froqucncy metor feeding a
tape-punch.
o minor disadvantage of the present design le that it is rather too heavy for
ease of' handling during calibration.
-5-.
soffers fri the disadvantage,
alreay
d.icuosed, that occasionally
.t i
foimd OEZ recovery that snething has not ceratcd correctly,and no
records have been obtaird A naller and lig)tcr buoy has therefore been macle which is connected by a cable to the ship where the outputs are recorded. In this buoy, a single gyroscope with a vertical axis carries light vertical accelerometer, and has pitch end. roll pick-sits gimbal bearings.
This
and itsassociated
elootrics are contained in a watertightcan fixed in
the middle ofa circular
woodendisc four
feet indiameter
ami four inches
thick.
The light multicore
cable which conneots it to the ship la kept slack, but the tenaion is sutficient tocontrol the aliment
of the buoy.These buoys allow the zero order, first and second angular harmonics of the wave ape otrun to be determined.
Principles of recording and
analysis
The
analysis of
wave records falls into threeoateories
-(i) Por moat routine recordings, only a height and period are required frun each record, and these ars then subject to statistical analysis.
Por researoh purposes and for the development of prediction formulae, the enerr-.apeotrxn of a non-directional
reoording may
be required.Por other research purposes such
83 the measurement of directional spectra
and the sttxy of ship motion, cross power spectra between records may berequired.
For the
simple
analysis, it is possible to devise systeme which record mean waveheight and
period directly: foremple,
that described by ßnodgrassend
Putz(1958)
forettaahing to the
output of an electrical wave mater or that developed byWcmeletelder (de acribod in
Ferguson, Weinalafelder and Santema,
1957)
for amechanical system.
Such inatrunents are usually
arranged to talca en oocesionalordinary wave record.
We havenot, however,
so far felt the need for such systxna.We believe that full
wave
reoorda should inany case be
takenat
freqtnt intervals,pertly because they usually show any fault which has
develqod in
the system, andpartly because
we often subBequently select a series cf these records and oiipute their frequanoy spectra for reeearoh purposes. Having got ther000rds, a manual
analysis
baeed onsound
statistical principles has been devised which is quiok end simple (see bolow).Por obtaining the power spactrun of a non-directional
recording,
thephoto-electric analogus analyser
(Tucker 1956) is still mostly used.. This has the advantage that elmoet any fcam ofrecord can be converted into a form suitable for
analysis on it by one of anuiber of
simple, ifacnetimes rather laborious, teohniqucs
and it gives a more detailed analysis
than the usual digital procedure. It has two major limitations. Firstly, the records must not contain more than about 200 wavesof the highest frequency of interest, which limita the
statistical
aocuracy. Seoondly, it in limited in accuracy of calibration to about ±1C end in relativeaccuracy within
e spectrun to about ±5%.
These
record lengths and accuracies sofficefor many ptooees, but
when better accuracy is required, digitalmethods are
used.Por this, the records
can either be recorded directly on 5-holetape, or recordad in
grsphio form and digitized on theequipment described below.
For obtaining direoticial
spectra and for some specisl
purposes, sh 85 the
study af ship motion, the
cross power spectra between pairs of
reoords are required.
When thia problem first
serious),y arose, N.T.O. had no analogue equipment capable
of
performing such analysis.The photo-electrio corralator (Tuoer
1952) would go
part of trie way, but its accursoy is poor, it will not handle
output oorrelo'amslong reoorda,
and itsrequire Fourier analysis
to give the cross-epetra.
Development
cf a speoíal analogue
nichine would be expensive,
and
the resulting
irìatrxnent wo&4
not be very flexible.
Suitable digital cctnputer proarns wero, however, already
available, and ere
comparatively easy to raodffy
as required.
Thus, everything
pointed to the use of digital
techniques for this purpose, and these hava, in
fact, been adopted.Having started
on the use
of high-speed digital computers1 N.I.0. has been
steadily extending its
range of proams
and of ancillary equirent.Many of these
techniques
are not, of course, N.I.0.aoriginal conception, but we hope that before
lg we shall
have an t*iusual
range of facilities
available. Sinpia mea8lxrement cf wave reoords
Though this system is not yet fully standardised
oven at N.I.O., it is thought
to be worth presenting
here in thehope
that it may be disoussodat the meeting.
Theinstructions are as follows, and are illustrated
in figure ii.,
Measure off a 10-minute
length of the record,
and oonsder on3.y
waves in this
mt
a rva i.Draw in a mean water-level
line by eye.
Coa.nit
the nber of crests N0.
A crest is defined sa e point where the water level
la mentarily oo13tant,
falling to either aide.
Sotse crests may be belcr4v
mean water level.
Coimt the riaber of tunca
N5 that the reo1
croases
the zero line nioving in en
upward direction.
Measure the height A of the
highest crest and the
height B of the seoond highest
crest, measuring frun the
zero line.
Measure the depth C of the
lost trough and the depth
D of the second lowest
trough, measuring from the
zero line and
taking both qintitiea as poeitve.Record H,
A+C
B + DT0
period of cro 3t5
600
T5 -r- period of zero crossings
rge
The theoretical basis for this system of measurement is
g4van by Ca1-twL'ight and
Longuet-Higgina
(1956)
and byCartwright
(1958),
(See also Putz, 19
) and is
The statistical distribution of wave-heighta is governed by the
r.m.a. wave
height Hr.m.s
and by a speotra1-idth parameter e
From the measurements, the best estimate of
eis
i
-Hr.m.s
is the r.m.s. deviation of the surface from
mean water level, and is estimated
as follows.
where
O1og6
These are the best estimates to a good dcgree of approximation.
The statistical errors in these estimates are less than might be oeoted, and
are not mth worse than that of the mean of the highest
-ç waves in the reoords.
The
fulae fr them are cc2nplicatod, but in
a typical case where N0
100,
e0.8, the
proportional standard error in the estimate of Hr.m.s. frc2n H1 is approximately 13%
and frcin E2 about 10%.
In practice, for many civil engineering purposes, the relevant wave-height ia
and the relevant period T.
Digital recording. and d±itisaticn of analojue records
N.IP0. novi has a digital recorder which will record, an input voltage as
s
patten-of holes on 5-hole teleprinter tape.
The equipment can record one, two or four
channels at a maximin rate of five readings per second with
a resolution of 0.1% of
full scale.
Each channel has a reoording potenticiiìeter fitted with
a ooddßiso on the pen
drive shaft.
The patton on the disc corresponding to the angular position of the
shaft is read photo-electrically, stored in relays and then pimehed onto the tape
using three successiva rows of holes.
The code used is Watts Rfleoted Deoiral with
a parity digit added.
This code is a form of cyolic progressive binary-coded
decimal and is used for instrtutental reesoris.
After every four readings, an
identi-ficeton "o'zrna" is pmched.
The instrtuent wa manufactured by liilgor axi Watts
Limited.
This digitizer can be used either to record the output of a wave meter directly,
or it can be fItted to th3 output of a photo-electric curve-follower.
This curve
fol1.ier is an improved version of that described by Tucker and Collins (1947),
azwill follow line records and convert them Into a varying electric voltage.
It must
have records in the form of lìrvs on plain paper or on transparent fiLn.
Records in
other forms can often be traced onto ohart suitable for feeding into the machine.
-7-I-h1 (2o)
(i + 0.289 6'
- 0.2l.7
a2)
--i-E2 (20) 2 (1 - 0,211 &
- 0.103
2) FrOEn H1Hrm s.
Frcxn FI2 :FIrme
Sctne reoords, notably those from the self-contained N.I.0. buoy, are more easily digitisad on semi-manual digitisers (medo by Southern Instrunents or
Benson-Lehner)
towhich we
have access atthe
Royal Aircraft Establishment. These records, cn 70 nia, film, have several tracas which may cross endwhich subsequently have
to becross-correlated, so
that matching cf time-scales is important. In the R.A.L inatrtznents a cross-wire is moved onto the record, and depressing a pedal then oausss the reading to be punched onto a computar card. The readings from all the traces et a partioit1.rtime are punched in ti.u:'n onto the
seme card:
the
record ismoved forward, the card
replaced by a new one, and the process repeated.
A system for recording digitally the output of an F.M. measuring head. is in process of being developed. In prinoiple it is similar to that used by Munk and Snodgrasa et Scripps Institution of Oceanography. The number of cyclos in the siial is counted for, say, I second on an electronic counter, end the state of the
decades is then punched onto 5-hole tape.
Such a
system is capable of very high resolution, and using the F.M. pressure recorder described above arid digitalfilter-ing techniques,
it
should allci long waves to be filtered out of the ordinary wavere cord.
r000ssingof digital records
The digitiser recording on 5-hole tapa is
unfortunately not too
reliable, and mecea onthe
average one error in about 200 readings. The code is alao not one cf the usual digital computer oodos. Thefirst stage in hsnd.Jing a
tepe f r this instrument is therefore to feed it throu,gh a computar p rrmed to change the code and look for errors. It first makes sure that the pattern of holes forras alegitimate codo, than decodes them and thon takes first differences. if these exoeed a certain value (fed in with the prograrrgno), it regards the reading
as en
error. It also makes sure that the correctnumber of roia of holes occur
between"OLias&'.
The record is thon punched out in standard coda with zeros where the
errore occurred. The serial number et the errors is also
given, so that the
operator can go through the tape and punch in corrected or interpolated readings by
hand.
The prograninos for
correlation andFourier
tranaformation ere besad on the well-laiown principles first sot out by Tukey, end will not be described here.Ac kncv].o dgements
The techniques described above hava been developed by quite a numerous team at
NI.O.
These inoltide (in alphabetical order) L.A. Baxter, R. Bowers, Miss D. Catton, D. E. Cartwright, L.A. Thaper, U. Harris, M. S. Longue t-Higgina, F.L Pierce and
Bof
cronacaCartvright, D.E. and Longuot-Higgins,
u.S.1956
"The statistical distribution
of the maxima of a rDnd1l function".
Proc. Roy. Soc. A,
Vol. 237,
p.212.Cart-drigììt, D.E.
1958"On estimating
the moan oner of seaevea from the
hig1ost waves in a record".
Proc. Roy. Soc. A, Vol. 2L7, pp. 22-48.
Snodgraos, F., Munk, TI. and Tucker, M.J.
1958
"Offahoro recording of
Li-fraquency
ocean
Trans. Am.
Goophys.
Union, Vol.39,
pp, iii-1a.Putz,
R.R. 195) "StatisticalAn3lysis of
3VORecords".
Proc.
Lth Conf, on
Coastalßngg., pp.
13-2L,-.Snodgrass, F.E. and Putz, R.R.
1958
"A wave height end frequency meter".Proc.
6th Conf, onCoastal
ìgg., pp.209-22L
Ferguson, H.A., Wumoisfeldor, P.J. and Santeros, P. 1957 19th
Internsti.al
Navigation Coness, London, pp. 207-231.Tker, M.J.
1952 "A photoelectric correlation motor".J. Soi. Instrun., Vol. 29, pp. 326-330.
Tuckor, M.J. 1956 "The
N.I.O.
VíavoAnalyser".
Proc. Ist Conf. on
Coastal
Bigg. Instrtiucnts, pp. 129-133.Tixkor,
M.J. and
Collina, G. 1917 "A photo-electric cure-fo1lower".(i) Cross-scotion of F.M. prossura
r000rdcr measuring hoad.
F.M. prosso recorder
mosuring-head
circuit.
The oscillator froqinny ja
approximately 500 ko/o
st enosphoricpressura, end 150 ko/a at a proasurc
oorre spending to a depth of 200 ft.
The froqncy measuring circuit and. wave/tide filter for routine
wave recording.
(4.)
To illustrato the simple measurement of
a wave reoord
(only five minutea of thar000x'd is shown haro).
The points marked v;ith a dash aro wave croata, and
with
3circle aro zero crosainga in an
ward.s direction.
Interior is
evacuated
Terminal
box
filled
with
compound
Oscillator coil
/
Support for electronics
ll/4 inches
Plate of capacitor
Mond waphrogm
Gunmetal case
irjmierjiu
t
C
o
C
o
o'
o a
O2,F
5pF
47Ort
122r
0C44
L1/
(JL,'
-e-6V
68 r.
.
I Frommeoauring
h cad'1H
B A REAmpi itier
r
8,.jF IiOMf
LCrystal
Oscillator
Mixer
Low pass
Filter $
Amplifier
Diodepumpio
fiApprox Ql V D.C.
per foot
SWITCH POSITIONS
RecorderA
Waves a tides
B
Waves only
Stated
and OtIt putC