Memo: Some advantages of SOFAR floats that move under their own power. Doug Webb made some engineering cosputations of the power required to move steered SOFAR floats a t small speed,s such 10 km/day for periods of up to several months. The power requirements are so small that the idea is a practical one. Vfuat advantages might they have? Various ideas like shooting them across acoustic tomographic ranges, or at intervals northvmrd from Bermuda to map the recirculation, come to mind. There appears to be another advantage: small nQ~bers of movmng floats seem to gather better information on the.mean velocity field than passive floats do. The idea seems obvious: (i) they visit more parts of the eddy field in the same ti~e; (ii) they are not so easily trapped in portions of trapped water moving with the eddies, and in the limit of high speed (iii) they would appear to give nearly pure horizontal instanteous means. over large distances. But high speeqs are too wasteful of power, and we can ask to have the advantages demonstrated for powered speeds not much different from those of the eddy structures themselves: what then? Lloyd Regier and I have made so~e rudimen~~ry model simulations to get a first look. Consider an eddy field - purely kinematic - given by a streamfunction of the following kind: It represents a ~quare-celled array of eddies, in alternating checkerboard fashion, p~op~gating in the x-direction with .. pha~e speed Cx and in the y-direction with phase speed cy• 2 Figure I shows a sketch of the eddy field at time t ~ 0, and also shows the trajectories of some floats that one would get were both phase velocity components zero. In this particular example the 'wavelength ( twice the "eddy size") is 200 km, and the amplitude of the current is 20 km/day. Some days are marked on the trajectories. Floats simply go round and round. However, if there is a non-vanishing phase velocity things are very different. Floats can get trapped and move with the phase velocity; others can be handed on from one eddy to another, and move in the opposite direction. Figure 2 x -6 km/day, Depending on shows some sample float trajectories computed with and amplitude of the current at 30 km/day. C where the floats are inserted, they can move for long distances eit!~er ea::-tor west - and they execute very complicated trajectories. Indeed it is qui te impossible to guess at the na ture of the velocity field from the inspection: the floats move in widely different directions, the curlicues are of different sizes, trajectories cross, etc. The trajectories are cOr.1putedfro~l1the forrmla: a. • -Q. In Figure 2, uo and o arc taken both as zero, otherTIise V they can be interpreted as the mean velocity TIithin the whole field ( or the time mean taken in the Eulerian sense at any point). - 3 Transforming the coordinate system to one moving with the phase veloci ty it is pos sible to.see th;tt!'oughly t::,ro-thirds of the water is trapped in closed eddies moving westward, and one third moves eastward meandering between the eddies. These proportions of course vary 'with the ratio a/cx• Indilfridual floats remain "stuck" in' certain regimes, and adequate sampling for determination of the var:,ous means would require very large nQmbers of floats - long times do not help for regular systems of this kind where trapping in one regime or the other is perpetual. More complicated randomised eddy fields might behave differently. One way to get better sa~pling is to power the float and steer it through the water at some fixed velocity uf,vf relative to the water. Under these circumstances the trajectory is computed using uo:= U + uf, vo; V + vf, where the bars denote a large-scale, long period mean over the whole field that is to be determined by measurement. In principle, if we make uf,vf large enough there will be no trapped floats, althought there will be trapped water particles. Un60rtunately this requires float relative velocities comparable to the amplitude a, and this could mean required float speeds of some 20 km/day for complete avoidance of float trapping. But some improvements of estimates of uo,vo 'to be obtained from movement of clusters of floats might be obtained with relatively small valu~s of Uf,Vfe Th8refore we make some simulation runs of floats with various values of uf'vf, for the case depicted in Figure 1. a 20km/day, cx - 5 km/day, ~imensions of instanteo~s eddy 100 km ( half wavelength). _.,-" ~..- f. ~ --.r' .:J ~ . i ~f':-i :i .--l~_--iI L __" I -J. _ I I . -+-.- ". ,.- I ! I ,.... I I I, ._1- __ .f--' ,,' I I "I " _ ! _L_ .. , _ I I.-I ! i i J': ..;. .~'. ',:r.'~., : .-:1 ' .I h.::-~. "7--- Q -.. C> ,- 'ri" ~::-:-r::-~-'1-.. ;:~:'! .•"..,.1-; ::! . -. r4-;:!j~r~:i:~1. . i .__.~---~._-_ ..- .._. ,. ~ ._--.,-. ... .•. 'I ""'?- ( + i ;:1: l j .,-~ -,- <::I "~=I...... .' ~:'l:i .....• ,.! lei f' ~~:;.:-.----t-=r~::r~~:i ,.{-:;.:-- ':' ' :<I '::r':'.1 ,. ':~.:~~:...:::t- - . c:> .~t~-~;~l-~10~f;~~~-:-I--"-~ ~[: ---.r ----,.. ----_._---~ ..t :-~;,.:. . , - --i~;..: . '-~'I' - : -1-: .: f . i -;~:._-::-' -'1 • I I ; I .... i , ~ ' - - i, ,- > - I • - - - f . .1 ." 'r'-. '" - i : . -;...-- -, ---j . ! .1 .j I Simulation runs: U and V are the drift velocity of the center of mass of a cluster of floats corrected for uf' vf, over the entire nUMber of days of drif~. Simulation No. floats Float days vf U A 6 120 30 0.92 0.66 B 6 i20 200 0.18 0.02 c 6 .120 0 -1.76 -0.85 D 6 120 10 E 1 200 10. 0.50 F 1 180 10 0.06 -1.43 G 4 400 10 0 .• 01 -1.0lf- H 1 240 10 1.50 0.48 Gf 1 100 10 -4.68 -0.22 I 7 560 10 0.08 0.50 V -2.,29 2.98 0.38 The plans of the initial deployment of these floats are given in Figure 3. In all runs except F and G the float direction was hel~ constant. In these runs direction was changed at fixed days. Thus in run F, the i~itial value of vf was positive; at 30 days reversed; at 60 days reversed again, and so on. The integral of vf over the 18Odtotal run was zero. Thus in this run the floaL was steered across the sa~e region more or less, six times. In run H, a similar type repeated scanning was done, but according to the schedule: Statring Day 0 30 60 120 150 180 210 240 uf vf 0 10 0 -10 0 -10 0 10 10 0 0 -10 0 -10 10 0 I ,'- ' •••..•...... ,': , --;"'Plans:;'or--.Si"mulatHm'- initial",pos i-: i ,. '. . " I J iti-ons-'-'" l' ; , 1tJ'; : • L . ... oj I . ;-'!. .,. - I If' \. Sh. ---,-'" I' ; : , : i , -i-! _.~... 1(:f~ j '--->'"I'~' ,- l> --:>- /f.. 0--1--'-'" _.. 5'0 E _ •.•...•• _.• ....-t..~. , --- ~ ". -I A:, B)'C) D ••• '. -.....-..._ ... --- .. -.. --'- ,-- .~ i~ •.. , ~ . ... >'J"'- F, H I, •• J .. -j l~, " I ' \,. I tt,-t .•"'...( :~, •..)' ...;i.•.••..: ~i-------_.-.----, , 12,.(; til 1 46 ({t- 9 ~ .-~ r.;. ~~ 1 ;,.9 ~' !, ': " ~-- , ! f I I : J ! i • .--,, ,, ' •• ~ •• I •• !.. •• J i ' ,k I (!(i .•'<t~' + ~ j ., j , ---"-----".--",''''''''- I ' 'r--" , I I', ! l' ':"'r •.. ,.+-l .t ..; "t_ ! .! \ ._T~j::::<. -+ i I _j.1 .... .•.. 1 . ••• 2() J--- i ,. '._~.....~.- /, '~ ,- i' , .; . i-! , I I ; \ " 1 : : '1 • : ,,,' i _ ';"! !~O~E;; ;.:In :runs ,F ~arid iH the 'value of ;Uf, vf ,was chaqged :~everal I: ;. ~,;. : !. L 1.:' .1',by 'i.n~ernal '.f<~oa:t program dl~ring the, run, !; !-i'. l:'j. " .-Jtj:.:'.L~.,~:.:....l. ......; .:jj :. . times . : I LJ '..~_:.~j~~~1~tL~L.i..t ~~I~ ~ 1_~~.L.~.. _... :_ .~: Comments: Obviouslyw~ need to do many more runs: but something emerges: Compare for example the U and V for cases Band C. Case C has passive floats. From the total 120 float days south( 6 floats each 20 days ) we seem to find a/westward mean drift of almost 2 km/day •. There is actually a ~ velocity eve~ywhere in the field. mean Eulerian Case B with the rapidly moving float ( 200 km/day) gives a much better estimate of zero. Powered floats can be trapped ( see the trapped float in G'). The cluster I of 7 floats moving northward for 80 days each, moved about 800 km north of the original deployment - much like a line launched near Bermuda would go across the recirculation region. It gives a pretty acceptable estimate of the mean. We will make a lot more runs, of clusters of 20-40 floats each, and try to make the case for the advantages of powered floats more definitely. Lloyd Regier and Henry Sto~mel April 20, 1978 17- s-. ~ "&F!:' .:--""'f-} "'""'l' ~ >b -k r.