![](https://csdnimg.cn/release/download_crawler_static/4206824/bg11.jpg)
4 1 Introduction
to obtain φ(x). Although there are many tables and analytic methods of obtaining Fourier
transforms and their inverses, numerical estimates of φ(x) may be of interest, such as where
there is no analytic inverse, or where we wish to estimate φ(x) from spectral data collected at
discrete frequencies.
It is an intriguing question as to why linearity appears in many interesting geophysical
problems. One answer is that many physical systems encountered in practice are accompanied
by only small departures from equilibrium. An important example is seismic wave propagation,
where the stresses associated with elastic wave fields are often very small relative to the
elastic moduli that characterize the medium. This situation leads to small strains and to a
very nearly linear stress–strain relationship. Because of this, seismic wave-field problems in
many useful circumstances obey superposition and scaling. Other fields such as gravity and
magnetism, at the strengths typically encountered in geophysics, also show effectively linear
physics.
Because many important inverse problems are linear, and because linear theory assists in
solving nonlinear problems, Chapters 2 through 8 of this book cover theory and methods
for the solution of linear inverse problems. Nonlinear mathematical models arise when the
parameters of interest have an inherently nonlinear relationship to the observables. This sit-
uation commonly occurs, for example, in electromagnetic field problems where we wish to
relate geometric model parameters such as layer thicknesses to observed field properties. We
discuss methods for nonlinear parameter estimation and inverse problems in Chapters 9 and
10, respectively.
1.2 EXAMPLES OF PARAMETER ESTIMATION PROBLEMS
■ Example 1.1 A canonical parameter estimation problem is the fitting of a function, defined
by a collection of parameters, to a data set. In cases where this function fitting procedure
can be cast as a linear inverse problem, the procedure is referred to as linear regression.
An ancient example of linear regression is the characterization of a ballistic trajectory. In
a basic take on this problem, the data, y, are altitude observations of a ballistic body at a
set of times t. See Figure 1.1. We wish to solve for a model, m, that contains the initial
altitude (m
1
), initial vertical velocity (m
2
), and effective gravitational acceleration (m
3
)
experienced by the body during its trajectory. This and related problems are naturally of
practical interest in rocketry and warfare, but are also of fundamental geophysical interest,
for example, in absolute gravity meters capable of estimating g from the acceleration of a
falling object in a vacuum to accuracies approaching one part in 10
9
[86].
The mathematical model is a quadratic function in the (t, y) plane
y(t) = m
1
+ m
2
t −(1/2)m
3
t
2
(1.13)
that we expect to apply at all times of interest not just at the times t
i
when we happen to
have observations. The data will consist of m observations of the height of the body y
i
at
corresponding times t
i
. Assuming that the t
i
are measured precisely, and applying (1.13)
to each observation, we obtain a system of equations with m rows and n = 3 columns that