IEEE Geoscience and Remote Sensing Magazine - June 2020 - 85

Physical Model

ContinuousMedium Model

Incoherent
Scattering Model

DiscreteMedium Model

Coherent
Scattering Model

Vegetation
Scattering Model
Phenomenological
Model

Imaging Simulation

IRF

Balance Between Efficiency and Accuracy
Accuracy

Wide Applicability

RVoG

Forest-Height
Inversion Model

Scattering
Calculation

Efficiency

Raw Signal

Model/Algorithm

Scene
Generation

MPA

Suitable for Scenes With Large Slope

S-RVoG
RVoG + VTD
TD-RVoG

Account for Natural and Human Decorrelation Factors
Account for Scatterers' Motion

DEM
Differencing

With Coherence Optimization
Without Coherence Optimization

Inversion Algorithm
Three-Stage
Inversion

With Coherence Optimization
Without Coherence Optimization

FIGURE 1. An outline of the various models and methods introduced in this article. The models and algorithms used in the case study are
highlighted in red. IRF: impulse response function; MPA: mapping and projection algorithm; RVoG: random volume over ground; S-RVoG:
sloped random volume over ground; TD-RVoG: temporal decorrelation random volume over ground; VTD: volume temporal decorrelation;
DEM: digital elevation map.

within the vegetation canopy. Nevertheless, the incoherent
scattering model has critical limitations in interferometric phase simulation [25], [26]. Incoherent models are not
able to calculate the accurate scattering phase and are not
suitable for InSAR simulations. They do not consider the
coherent effects of electromagnetic waves scattered by different vegetation components. When the vegetation contains dense scatterers and/or the radar frequency is low,
the coherence effect between the scatterers is nonnegligible, and the simulation accuracy of incoherent models
is greatly degraded.
COHERENT MODEL
To address the need for modeling and simulation of interferometric SAR, coherent scattering models have been developed in recent years. A comparison of the incoherent and
coherent models is summarized in Table 1. Coherent models often employ deterministic vegetation models. To have
a realistic description, two types of vegetation models are
often used: structural and growth-rule-based models [27].
In the structural model, vegetation structure is characterized by a level of detail whereby the fine features of distinct vegetation structures can be preserved to the desired
JUNE 2020

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

TABLE 1. THE INCOHERENT COMPARED WITH THE COHERENT
MODEL.
INCOHERENT MODEL

COHERENT MODEL

ADVANTAGES Easy to implement
High efficiency

Realistic
Can simulate complex scene

LIMITATIONS

High computation cost

Difficult to get the
probability-density
functions and the
pair-distribution
Can simulate only
simple scenes

level of accuracy [27]. The first level consists of the primary
components of the vegetation structure, such as stems or
trunks, which are, in general, the major scatterers. The
next level includes the branches and specifies how many
are grown from a single stem or trunk, along with their
orientation angles and branching angles. The third level
includes leaves and describes how many are grown from
each terminal branch or stem, their orientations, and their
connections to the terminal branches or stems, providing
a more details about the vegetation structure. Since some
85

IEEE Geoscience and Remote Sensing Magazine - June 2020

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2020