Fig. 1.  A modeling example showing the higher-mode energy domination can occur  with a particular normal velocity structure, which can be a case with soil site investigation.
Fig. 2.  A modeling example showing the possibility of different dispersion images for different offset ranges.  This is because of the roles of relative energy partitioning
between different modes (called excitability) and different damping effects of the surface waeves.  
Fig. 3.  A possible future inversion approach based on the match of the measured and theoretical
dispersion images together without modal dispersion curve extraction.
Fig. 4.  Schematic of a future in-field real-time processing system.
Current Issues and Future Directions

The approach based on M0-curve extraction followed by inversion for a 1D Vs profile is deemed to be less robust than traditionally considered, as the multimodal energy
partitioning and such resultant complications as modal uncertainty and mixture commonly occurs.  Although this is espe¬cially the case with measurements over pavement, it
is now often observed during normal soil inves¬tigations even with the multichannel (MASW) approach (
Fig. 1).  Some investigators (e.g., O’Neil, 2003) indicated that an
unusually high Poisson’s ratio for the top layer (for example, 0.45 or higher), or a significant stiffness contrast at shallow (for example, 1 m or less) depths (shallow bedrock),
or a combination of both can give rise to such a phenomenon over a soil site.  This issue has to be continuously studied in the future through the comparative study of theory
and field measurements.  In addition, it is not properly accounted for in most current dispersion analysis where different modes get different energy at the time of generation
and this relative energy partitioning (also referred as excitability) is a function of frequency that in turn changes with different layer models (
Fig. 2).    
With the aforementioned complications in dispersion analysis, it seems that a
more competent inversion approach other than the M0-curve-based inversion
should be used in MASW approach to increase reliability of the result, something
similar to the apparent dispersion curve concept in the SASW method.  Study of
approaches such as dispersion-image (i.e., phase-velocity spectrum)-based or
multi¬channel-field-record-based inversion (
Fig. 3) that account for modal
complications as well as acquisition parameters should be continued with an
emphasis on improving practical aspects such as speed and automated
opera¬tion.  The surface-wave phenomenon over both soil and pavement may need
to be re-examined from a perspective that considers not only the conventional
elastic propagation but also new issues such as non-propagating resonance, leaky
waves, non-elastic deformations, etc., so that new inversion approaches in the
future can be equipped with a more robust forward-modeling scheme.

Current 2D Vs mapping with the multichannel approach that is based on the
perfectly-layered-earth model being followed by the independent dispersion
analysis of individual records collected over different surface locations has great
room for improvement by considering lateral variations of the earth and
simultaneous analysis of multiple records (Xia et al., 2000).  The improvement will
be facilitated and accelerated by the possibly then improved definition of the
surface-wave phenomenon previously mentioned.  The passive method urgently
needs to be studied, especially in the dominant modal nature of the
traffic-generated surface waves as well as in dispersion analysis techniques.  This
will require intensive field studies in a combined manner with active surveys and
cross checking by other methods such as borehole measurements.

In spite of aforementioned complications that, once resolved by research, will make
the method applicable for a broader range of geotechnical engineering projects, the
current state of active MASW method has reached the level of successful
applications for many projects.  With this stage of the method achieved, it seems
that practice of the in-field real-time processing will soon be observed as overall
system of data acquisition and processing is getting more automated (
Fig. 4).