Voici une image du hotspotting qui devait permettre de réaliser
des
miracles
:
Pretty, isn't it?
Et voici la version étendue de l'article
(bien plus court) que nous avons publié dans la revue Nature
du 12 Novembre 1998 (cet article a été soumis in extenso
le 23 avril 1998 !) . Le titre était alors :
LIFE AND DEATH
of HOTSPOTTING
|
By Daniel Aslanian*, Louis Géli, Jean-Louis
Olivet
Ifremer, Marine Geosciences Department, BP 70, 29280
Plouzan?, France
* To whom correspondence should be addressed (aslanian@ifremer.fr).
In may 1997, Nature published an article by Wessel and Kroenke1
presenting an age-independent geometric technique which was supposed to
"refine absolute plate motions" and "relocate extinct hotpsots".
According to the authors, this technique has the potential to estimate
the pseudo-age for all seamounts in a seamount chain and link cogenetic
seamounts produced by hotspot ; in addition, they claim that its
application points to a recent change in the Pacific Plate motion and relocates
the Louisville hotspot to the Hollister ridge, south of Eltanin FZ.
Three months later, the same authors proposed to Geophysical Research
Letters a note2 (published in february
1998) based on the analysis of synthetic and actual seamount data, showing
that the interpretation of raw images computed by their technique is not
straigthforward. This note however does not explicitely point out all the
insufficiencies of the method and does not cast doubt on the conclusions
resulting from its application. A careful "non-blind" examination of this
technique shows that it cannot be used in practise without age information
and so, it does not present any progress relative to the classical backtracking
method ; in addition, the practical conclusions (such as the location of
the Louisville hotpsot) are not correct3,4.
I. Dissection of the method
Using synthetic data to explore the nature of the image of cumulative
volcano amplitude (CVA), W&K2 demonstrate
that "factors such as across-trail scatter of seamount locations, inaccurate
stage poles and migration of hotspots exert strong influences on the location
and appearances of CVA maxima". Nonetheless, doing
so, they consider their technique as a black box - which takes a set of
seamounts as input and produces a CVA image as output - and analyze the
effects that perturbations on the quality of the input may exert on the
quality of the output. Here, we instead prefer look at the "inside of the
black box", evaluate its practical application using real data and show
that the limits of this technique could have been revealed at the beginning.
Tautology.
The hot-spotting method is based1
on "the fact that seafloor beneath hot-spot -produced seamounts have
crustal flowlines that intersect at actual hot spot locations provided
there is more than one stage rotation and that hot spots are stationary".
Then, because the set stage pole used for computing flowlines is calculated
in order to fit the trend of the Hawa*an-Emperor seamounts chain, the clear
X on the image of cumulative volcano amplitude (CVA) which marks the location
of the Hawaian hotspot is a normal consequence of geometry (Fig.1).
To pretend Ñ as W&K do Ñ that this X "illustrates
the power of the technique" is merely a tautology.
.
Figure 1 : Dissection of the Hot-spotting technique.
Crustal flowlines are drawn for a selection of alignments. We chose
those volcanoes in order to explain some of the brightest CVA maxima observed
on Fig. 6 of W&K1 (dotted thick lines; A-G) and keep the lisibility
of the picture. Note that: the "Raratonga CVA maximum" (A) is mainly the
result of the intersection of the Marshall-Gilberts Islands and Samoa crustal
flowlines ; "Mehetia CVA maximum" (B) and "Macdonald CVA maximum" (C) are
underlined both by the flowlines from the Phoenix Islands; "Fatuhiva CVA
maximum" (D) occurs at the intersection of flowlines from Northern Line
Islands, Central Basin Ridge and Marquesas alignment; the Emperor Chain
flowlines produce two CVA maximum artefacs (E and F), with respectivly
the Northern Line Islands and the Central Basin Ridge flowlines. Oceanic
plateaux (in orange) and Volcanoes (in red) were digitilised from the gravity
map of Sandwell & Smith8
Interferences.
In their self-criticism, W&K2
note "that the large, broad CVA intersections at Raratonga may in part
be caused by interference by other maxima associated with large seamounts
in the western Pacific". However, they underestimate the effect of
those interferences. If, instead of plotting the CVA image, we plot the
crustal flowline of each volcano (Fig. 1), we observe
that the bigest and brightest CVA maxima shown in their image, in the vicinity
of Raratonga and Rurutu, is obtained by the intersection of flowlines issued
mainly from the following different alignements : 1) the Samoa alignment,
which presents an age progression from 10 to 0 Ma5;
2) the Marshall-Gilbert Islands (for Raratonga), which does not present
any age progression6; and 3) the Mid-Cretaceous
ridges of the Phoenix Fracture Zone7. However,
age and/or direction constraints demonstrate that there is no possible
genetic relationship between Rarotonga and any of these three alignments,
and that these alignments cannot be genetically related. A similar analysis
also applies to the Marquesas Islands which are located at the intersection
of flowlines issued from seamouts of the Hawaian-Emperor chain, the Line
Islands chain, the Central Basin Ridge and the Marquisas alignement (Fig.
1).
Consequently, in order to avoid problems of interferences and image
intersections where no hotspots are present (see the CVA maxima close to
the Line Islands on figure 1), one must, for all practical
purposes, stop plotting flowlines at the age of volcanoes. This requirement
strongly reduces the practical interest of the method! Furthermore, even
so, it is not possible to avoid all the interferences : for instance, the
Phoenix Islands crustal flowlines play a strong role in both Mehetia and
MacDonald CVA maxima althought this alignment obviously cannot be linked
with both hot-spots.
Hotspot
and non-hotspot alignments.
The hotspotting technique could have had some power for discriminating
the hot-spot produced seamounts if - as claimed by W&K2
- "seamounts not created at a hotspot have flowlines that do not converge
at a single point.". However, this statement is not correct, since
non-hotspot alignments may converge at a single point of maximum focus
(Fig. 1). For instance, the flowlines issued from the
synchronous volcanoes of Marshall-Gilbert alignment6,
the volcanoes of the Central Basin Ridge or the Phoenix FZ respectivly
converge in one point of maximum focus. Any alignment created for instance
within a fracture zone that is parallel to a small circle described by
Hawaii-Emperor chain will experience the same fate.
Reactivation
of volcanism.
Some Pacific alignments - such as Line9,10
and Cook-Austral Islands11 - present two
or more periods of volcanic activity that are separated by more than 30
Ma, and it is not possible to discriminate the resulting diagenetical volcanoes,
otherwise than by measuring their age. Because the hot-spotting technique
is unable to take into account the possibility of volcanic reactivation
over a given alignment, these volcanoes will contribute to the same CVA,
simply because they are on the same trend.
More generally, since hotspotting is a simple,
geometric technique, it cannot take into account the numerous "complexities
in the tectonic development of the Mesozoic Pacific". One can guess
that if the problem of deciphering the absolute motions of plates were
relevant to geometry only, it would have been already solved.
2. Practical application of
the method
Although they recognize that the interpretation of raw CVA images
is not as straightforward as initially thought, W&K2
still keep claiming that the hot-spotting concept can be used to refine
absolute plate motions "by seeking pertubations to the initial stage
poles that will focus the CVA image". One of the problems is to define
the criteria that can be used practically for focusing the image. Because
of the numerous complexities of Pacific volcanism, a global, statistical
approach extended to all Pacific seamounts (such as the minimization of
the overall standard deviation of the CVA image) has intrinsically no geological
significance. Defining criteria based on statistics restricted to Hawaii
and Louisville seamounts is not either an easy task, since it is not possible
to modify one stage pole and keep a good fit on both alignments without
modifying the other poles. In addition, volcanism along the Louisville
seamount chain declined sharply after 25 Ma and was almost totally interrupted
after about 11 Ma12. Therefore, the quality
of the CVA image on Louisville only depends on volcanoes older than 11
Ma : changes in stage poles for Late Miocene and Pliocene periods will
merely shift the location of the "X" of the Louisville CVA trails, but
will not affect the focus of the Louisville CVA image.
Computing the absolute motions of the Pacific Plate using the conventional
backtracking approach requires at least two different alignments of the
same age that are as far away as possible and the checking of the resulting
pole on all other synchroneous Pacific alignments. The pole of W&K1
fits the recent bend of the Hawaian trend but fails to describe most Pacific
alignments of age < 5 Ma (Fig. 2). Using the lineations
of the geoid of age < 20 Ma, Wessel et al.13
computed a pole that fits with the recent alignments, but does not describe
the Hawaian bend. In order to reconcile the change in the Pacific plate
absolute motion that probably occured in the last 5 Ma14
with the trend of all other Pacific alignments, we propose a pole15
that yields a bend that is less important than what is marked in the Hawaiian
topography, suggesting that the actual bend in the topography is likely
to be related to the presence of near-by surface features10 (such as the
Molokai Fracture Zone).
Figure 2
Small circles computed with W&K's most recent pole1 (25¡N,
27¡W). This pole fails to describe most of the alignments of age<
5Ma (underlined in yellow).
Finally, the pole proposed by W&K1
relocates the present Louisville hotspot to the Hollister Ridge, south
of the Eltanin FZ, which is not consistent with the geochemical constraints4.
The pole that we propose using conventional backtracking locates the Louisville
hotspot at 50.9S, 137.6W, in the front of a volcano located near 50.5S,
139.2W 12 which is known to have a Louisville
isotopic signature16,17.
3. Conclusion
W&K2 pretend that the hot-spotting
is more objective than backtracking. We have seen that hot-spotting is
flawn
by intrinsic and extrinsic errors and that its application is not
age independent in practise. But in addition, it does not present any
advance relative to classical backtracking. Indeed, using the few existing
dated volcanoes, it is possible to "pseudo-date" most volcanoes from the
Pacific by bactkracking and link the cogenetical alignments. It is true
that computers are faster and allow to handle more data than paper and
scissors, and that the hotspotting technique can employ all Pacific seamounts
(more than 10000) that can be characterized remotely by satellite altimetry,
"instead of being limited to a few hundred seamounts with dates of highly
variable quality". However, the blind use of computers - "by throwing
away half of the data"18,19
- fails to take into account the complexity of nature and leads to erroneous
conclusions. Rapidity is not necessarily synonym of good science and one
should better remind the words of Claudio Monteverdi, whose music, written
back in the 16th century, is still alive : "I know that it is possible
to compose fast but fast and well do not go together well".
Acknowledgements
We thank Pierre David, chairman of Ifremer, and Xavier Le Pichon
for their support. The figures were drawn using the GMT software20.
References
1. Wessel, P. & Kroenke, L.
A geometric technique for relocating hotspots and refining absolute plate
motions. Nature, 387, 365-369 (1997).
2. Wessel, P. & Kroenke, L.
Factors influencing the ocations of hotspots determined by the hot-spotting
technique, Geophys. Res. Let. , 25, 555-558 (1998).
3. Vlastelic, I., Dosso, L., Etoubleau,
J. , Joron, J.-L., Bougault, H., & G?li, L., New geochemical data from
the Pacific-Antarctic Ridge and the Hollister Ridge, Terra Cognita, 9,
515 (abstract) (1997).
4. Vlastelic, I., Dosso, L., Guillou,
H., G?li, L., Bougault, H., Etoubleau, J. & Joron, J.-L., Geochemistry
of the Hollister Ridge : relation with the Louisville hotspot and the Pacific-Antarctic
Ridge. Earth Planet. Sci. Let., submitted.
5 Natland, J.H., The progression
of volcanism in the Samoan linear volcanic chain, Am. J. Sci., 280-A, 709-735
(1980).
6 Lincoln, J.M., Pringle, M.S.
& Premoli Silva, I., Early and late cretaceous volcanism and reef-building
in the Marshall Island, in the Mesozoic Pacific: Geology, Tectonics, and
volcanism, Am. Geophys. U. Monograph , 77, 279-305 (1993).
7 Nakanishi, M., Topographic expression
of five Fracture Zones in the northwestern Pacific ocean, in the Mesozoic
Pacific: Geology, Tectonics, and volcanism, Am. Geophys. U. Monograph
, 77, 121-136 (1993).
8 Sandwell, D.T. & Smith, W.H.F.,
Global marine gravity from ERS-1, Geosat and Seast reveals new tectonic
fabric, Eos Trans Am. Geophys. Union, 73, 133 (1992).
9 Winterer, E.L., Bathymetry and
regional setting of the Line islands Chain, Initial Rep. Deep Sea Dril.
Proj., 33, 731-747 (1976).
10 Epp, D., Possible perturbations
to hotspots trace and implications for the origin and structure of the
Line Islands, J. Gesophys. Res., 89, 11273-11286 (1984).
11 McNutt M. K., Caress, D. W.,
Reynolds, J., Jordahl, K. A. & Duncan R.A., Failure of pume theory
to explain midplate volcanism in the Southern Austral Islands, Nature,
389, 389-392 (1997).
12 Lonsdale, P. Geography and
History of the Louisville hotspot chain in the Southwest Pacific. J. Geophys.
Res 93, 3078-3104 (1988).
13 Wessel, P., Kroenke, L.W. &
Bercovici, D., Pacific Plate motion and undulations in geoid and bathymetry,
Earth Planet. Sci. Lett., 140, 53-66 (1996).
14 Cox, A. & Engebretson,
D. Change in motion of the Pacific plate at 5 Ma BP. Nature 313, 472-474
(1985).
15 Geli, L., Aslanian, D. &
Olivet, J.-L., Vlastelic, I., Dosso, L., Guillou, H. Bougault, H., The
Hollister Ridge, Louisville hotspot and the hot-spotting technique, Science,
submitted.
16 Hawkins W., Lonsdale, P., &
Batiza, R., Petrologic evolution of the Louisville Seamount Chain, Am.
Geophys. U. Monograph , 43, 235-254 (1987).
17. Cheng, Q., Park, K.-H., Macdougall,
J. D., Zindler, A., Lugmair, G. W., Staudigel, H., Hawkins, J., & Lonsdale,
P., Isotopic evidence for a hotspot origin of the Louisville seamount chain,
Am. Geophys. U. Monograph , 43, 283-296 (1987).
18 Stein, S., hot-spotting in
the Pacific, Nature, 387, 345-346 (1997).
19 Schneider, D., Hot Spotting,
Scientific American, 22-24 (April 1997).
20 Wessel, P. & Smith, W.H.F.,
EOS 72, 441-446, 1992.
Back :
to the beginning
to part 1: "Dissection of the method",
to part 2: " Practical application of the method"
to the conclusion
Noter que dans leur Réponse,
qui n'en est pas une, Wessel & Kroenke font appel à six
références : trois ont pour
auteurs Wessel et/ou Kroenke, les trois autres
sont citées à contre-sens !
Derniere mise a jour : Septembre 1999
Marine Geosciences