Nolet Home

Department of Geosciences 
320 Guyot Hall 
Princeton University 
Princeton, NJ 08544

Phone: (609) 258-4128
E-Mail:

My research interests - a brief timeline

numbers refer to the   list of publications

I have always been very interested in the role of the upper mantle of the Earth in shaping the surface as we observe it. Seismic surface waves are confined to the upper mantle, and therefore provide an excellent means of probing this region. Most of my early work dealt with the measurement and interpretation of Rayleigh and Love surface waves.

In the research I did for my PhD thesis in the 1970's, I developed methods to measure phase velocities of higher modes of Rayleigh waves, which penetrate much deeper than the fundamental mode [1-5,16]. 

Frustrated by having to spend six months digitizing photographic seismograms from the WWSSN network I developed an interest in digital instrumentation, and looked for ways to bring "long period seismology" (which we now call "broadband") to the field. In 1982, this led to the first portable broadband array of seismometers [19], which probed the upper mantle under western Europe between Goteborg in Sweden and Malaga in Spain [23,25, 31, 33, 38, 39]. This Network of Autonomously Recording Seismographs (NARS) was funded by the Netherlands Organization for the Advancement of Pure Research (NWO) as part of the European Geotraverse and was the first of its kind. It served as example for the PASSCAL instrument pool later developed by IRIS in the US. My interest in instrumentation led me in 1993 to devise a 9000 instrument "Lehmann telescope" for the IRIS 2000 proposal on the basis of low-cost seismometers developed for the Princeton Earth Physics Project, a high school educational initiative I did earlier with Bob Phinney. Although the Lehmann telsecope did not make it into the IRIS proposal, it did trigger much discussion and eventually led to the US Array project as part of  Earthscope  [109]. And currently, Jeff Babcock, Russ Davis (Scripps) and I are investigating  the use of autonomous floats to open up the oceans for teleseismic observations.

The methods used in the 1970's  were still assuming a nicely layered structure beneath the array. The shortcomings of this view became quickly apparent as the first results from NARS came in and showed significant lateral heterogeneity. For example, NARS was the first experiment to clearly show large topography on the 660 km discontinuity [65]. At first, it was not evident how one could tackle such lateral structure adequately with surface waves (I consider the two-station methods pretty inadequate!). I therefore switched my interest to body waves, and the method of seismic tomography which had been developed several years before by Dziewonski and by Aki. Earlier work in inversion theory [6,8,17] proved very useful to quickly improve the matrix solvers used at the time. At several conferences in 1983 I introduced Paige and Saunders' LSQR method - which had been published just a year earlier - into seismic tomography which for the first time allowed us to solve very large systems (millions of equations) without imposing an inherent scaling to the problem [27,41,51,58,77,81,82].  LSQR is now the method of choice for large scale tomographic inversions. My activities in seismic body wave tomography eventually led to Rob van der Hilst's discovery that slabs penetrate deeply into the lower mantle [52], and to the editing of a book on the methodology [36]. Efforts to constrain tomographic inversions by forcing slab images to satisfy heat advection and diffusion led to very crisp images of subduction zones and a clear indication that slabs thicken with age [73,99,100]. As a spinoff of this interest in body waves I developed an interest in ray tracing and chaos theory [61,67,71].

Body wave tomography is very powerful in imaging uppermantle structures in regions of strong seismicity such as subduction zones, but does rather poorly in resolving the structure of the Earth beneath its oldest cratons, or even beneath the Phanerozoic platforms, and is practically useless in resolving the upper mantle structure beneath the oceans (2/3 of the Earth's surface!). What was really needed was a way to reconciliate surface wave theory with the now obvious lateral heterogeneity of the Earth. I first developed an efficient way to invert surface wave waveforms at low frequency for single seismograms using nonlinear optimization [28,37,74].  Such inversions impose a average 1D structure along the source-station path, but at first it was not at all evident how to use such path constraints in a 3D tomographic inversion, because the 1D structure is generally non-unique and the a posteriori covariance matrix is not diagonal. Using the Hessian matrix of second derivatives, I derived independent constraints on 3D structure from the single seismogram inversions, which led to the very efficient "partitioned waveform inversion" method [49]. The method was applied by several of my students, and resulted in some interesting discoveries: the sharp and deep character of the suture zone that separates western from central Europe, known as the Tornquist-Tesseyre zone [62,63], the postulate that the Ukranian shield's root has been eroded by the action of volatiles during subduction of the proto-Atlantic [64], a remarkable image of the trailing edge of the Farallon plate resting at 600 km depth beneath the American west [83], and new S velocity models beneath North America [87] and beneath East Asia and the West Pacific [86].

The partitioned waveform inversion method still relies on the assumption of an undisturbed wavefield that propagates with a local phase velocity - violations of these assumptions become quickly apparent as the waveform fits deteriorate. We can escape this restriction using first order (Born) perturbation theory [35].
This would also allows us to include body waves: I earlier showed how body waves can be theoretically derived from a mode sum by imposing constructive interference of phases at the group arrival time [7,44].
This led to work by Thomas Meier and Henk Marquering [80,85,97] which formed the basis of our present research program in  finite frequency tomograpy . I teamed up with Ludovic Margerin to extend the first order approach to multiple scattering using radiative transfer theory, which we are currently applying to find the small scale structure in the lower mantle using  PKP waves . 

Finally, my interest in mantle structure inevitably led me to an interest in mantle plumes. An early study of the East African Rift led me to propose separate plume structures beneath the branches that run East and West of Lake Victoria [20], a view that finds support in results from more recent PASSCAL experiments by other groups. With Richard Allan and Jason Morgan we conducted our own PASSCAL experiment to image the shallow structure of the Iceland hotspot [93,107,108,110,111,112].

In order to keep this readable, I did not always name every student or collaborator in this overview. However, I'll be quick to admit they did most of the work, especially my current and former graduate students:  W. Spakman, R. Snieder, B. Dost, H. Paulssen, A. Zielhuis, T. Moser, S. van der Lee, H. Keers, T. Das, S. Lebedev, M. Deal, R. Allen, R. Montelli, Y. Zhou, K. Sigloch and postdocs: G. Shudofsky, R. Clouser, K. Vogfjord, C. Papazachos, Th. Meier, H. Marquering, Shu-Huei Hung, L. Margerin, B. Schlottmann, I. Tibuleac., F. Simons. At times, their interests would inspire me rather than the other way around, and their are quite a few publications with their name first that did not quite fit into this timeline.

Everyone who has followed my more recent work will clearly see how much I am indebted to  my Princeton colleagues, especially Tony Dahlen and Jason Morgan.