Muutke küpsiste eelistusi

Neural Networks in Atmospheric Remote Sensing Unabridged edition [Kõva köide]

,
  • Formaat: Hardback, 234 pages
  • Ilmumisaeg: 30-Jun-2009
  • Kirjastus: Artech House Publishers
  • ISBN-10: 1596933720
  • ISBN-13: 9781596933729
Teised raamatud teemal:
Neural Networks in Atmospheric Remote Sensing Unabridged editionSuurem pilt
  • Formaat: Hardback, 234 pages
  • Ilmumisaeg: 30-Jun-2009
  • Kirjastus: Artech House Publishers
  • ISBN-10: 1596933720
  • ISBN-13: 9781596933729
Teised raamatud teemal:
Püsilink: https://www.kriso.ee/db/9781596933729.html
Märksõnad:
Blackwell (senior technical staff at the Massachusetts Institute of Technology's Lincoln Laboratory) and Chen (senior engineer at Signal Systems Corporation) present an applications-oriented treatment of neural network methodologies for use in atmospheric remote sensing. Their focus is on the retrieval of atmospheric parameters, such as the Earth's temperature, water vapor profiles, and precipitation rate, but the methodologies can also be applied to other problems where function approximation is required. They begin with simple, theoretical examples demonstrating how performance is affected by basic neural network attributes such as model selection, initialization, and training methodology and then build on those to describe applications common in atmospheric remote sensing. The examples are often accompanied by MATLAB software codes, available on the accompanying CD-ROM, which can be used for larger and more complex problems. Annotation ©2009 Book News, Inc., Portland, OR (booknews.com)
Preface xiii
Introduction
1(6)
Present Challenges
1(1)
Solutions Based on Neural Networks
2(1)
Mathematical Notation
3(4)
References
5(2)
Physical Background of Atmospheric Remote Sensing
7(30)
Overview of the Composition and Thermal Structure of the Earth's Atmosphere
7(5)
Chemical Composition of the Atmosphere
8(1)
Vertical Distribution of Pressure and Density
9(1)
Thermal Structure of the Atmosphere
10(1)
Cloud Microphysics
11(1)
Electromagnetic Wave Propagation
12(4)
Maxwell's Equations and the Wave Equation
12(1)
Polarization
13(2)
Reflection and Transmission at a Planar Boundary
15(1)
Absorption of Electromagnetic Waves by Atmospheric Gases
16(3)
Mechanisms of Molecular Absorption
17(1)
Line Shapes
17(1)
Absorption Coefficients and Transmission Functions
17(1)
The Atmospheric Absorption Spectra
18(1)
Scattering of Electromagnetic Waves by Atmospheric Particles
19(3)
Mie Scattering
19(2)
The Rayleigh Approximation
21(1)
Comparison of Scattering and Absorption by Hydrometeors
22(1)
Radiative Transfer in a Nonscattering Planar-Stratified Atmosphere
22(8)
Equilibrium Radiation: Planck and Kirchhoff's Laws
24(1)
Radiative Transfer Due to Emission and Absorption
24(1)
Integral Form of the Radiative Transfer Equation
25(2)
Weighting Function
27(3)
Passive Spectrometer Systems
30(3)
Optical Spectrometers
31(1)
Microwave Spectrometers
32(1)
Summary
33(4)
References
35(2)
An Overview of Inversion Problems in Atmospheric Remote Sensing
37(18)
Mathematical Notation
38(1)
Optimality
38(1)
Methods That Exploit Statistical Dependence
39(6)
The Bayesian Approach
39(2)
Linear and Nonlinear Regression Methods
41(4)
Physical Inversion Methods
45(3)
The Linear Case
45(1)
The Nonlinear Case
46(2)
Hybrid Inversion Methods
48(1)
Improved Retrieval Accuracy
48(1)
Improved Retrieval Efficiency
49(1)
Error Analysis
49(2)
Analytical Analysis
49(1)
Perturbation Analysis
50(1)
Summary
51(4)
References
52(3)
Signal Processing and Data Representation
55(18)
Analysis of the Information Content of Hyperspectral Data
56(3)
Shannon Information Content
56(2)
Degrees of Freedom
58(1)
Principal Components Analysis (PCA)
59(10)
Nonlinear PCA
61(1)
Linear PCA
61(2)
Principal Components Transforms
63(1)
The Projected PC Transform
64(3)
Evaluation of Radiance Compression Performance Using Two Different Metrics
67(2)
Representation of Nonlinear Features
69(1)
Summary
70(3)
References
71(2)
Introduction to Multilayer Perceptron Neural Networks
73(24)
A Brief Overview of Machine Learning
74(8)
Supervised and Unsupervised Learning
74(1)
Classification and Regression
74(1)
Kernel Methods
75(1)
Support Vector Machines
76(2)
Feedforward Neural Networks
78(4)
Feedforward Multilayer Perceptron Neural Networks
82(3)
Network Topology
82(2)
Network Training
84(1)
Simple Examples
85(9)
Single-Input Networks
85(8)
Two-Input Networks
93(1)
Summary
94(1)
Exercises
95(2)
References
96(1)
A Practical Guide to Neural Network Training
97(18)
Data Set Assembly and Organization
97(3)
Data Set Integrity
98(1)
The Importance of an Extensive and Comprehensive Data Set
98(1)
Data Set Partitioning
98(2)
Model Selection
100(1)
Number of Inputs
100(1)
Number of Hidden Layers and Nodes
100(1)
Adaptive Model Building Techniques
101(1)
Network Initialization
101(1)
Network Training
102(3)
Calculation of the Error Gradient Using Backpropagation
102(2)
First-Order Optimization: Gradient Descent
104(1)
Second-Order Optimization: Levenberg-Marquardt
104(1)
Underfitting and Overfitting
105(2)
Regularization Techniques
107(4)
Treatment of Noisy Data
108(2)
Weight Decay
110(1)
Performance Evaluation
111(1)
Summary
112(3)
References
114(1)
Pre- and Post-Processing of Atmospheric Data
115(22)
Mathematical Overview
116(1)
Data Compression
117(1)
Filtering of Interfering Signals
118(6)
The Wiener Filter
119(1)
Stochastic Cloud Clearing
120(4)
Data Warping
124(10)
Function of Time of Day
125(4)
Function of Geolocation
129(2)
Function of Time of Year
131(3)
Summary
134(3)
References
135(2)
Neural Network Jacobian Analysis
137(12)
Calculation of the Neural Network Jacobian
138(1)
Neural Network Error Analysis Using the Jacobian
139(4)
The Network Weight Jacobian
139(1)
The Network Input Jacobian
140(1)
Use of the Jacobian to Assess Noise Contribution
141(2)
Retrieval System Optimization Using the Jacobian
143(3)
Noise Smoothing Versus Atmospheric Smoothing
144(1)
Optimization Approach
145(1)
Optimization Results
146(1)
Summary
146(3)
References
148(1)
Neural Network Retrieval of Precipitation from Passive Microwave Observations
149(30)
Structure of the Algorithm
149(4)
Physical Basis of Preprocessing
150(3)
Physical Basis of Post-Processing
153(1)
Signal Processing Components
153(12)
Limb-and-Surface Corrections
153(2)
Precipitation Detection
155(4)
Cloud Clearing by Regional Laplacian Interpolation
159(4)
Temperature-Profile and Water-Vapor-Profile Principal Components
163(1)
Image Sharpening
164(1)
Development of the Algorithm
165(3)
Retrieval Performance Evaluation
168(7)
Image Comparisons of NEXRAD and AMSU/HSB
168(1)
Numerical Comparisons of NEXRAD and AMSU/HSB Retrievals
169(4)
Global Retrievals of Rain and Snow
173(2)
Summary
175(4)
References
176(3)
Neural Network Retrieval of Atmospheric Profiles from Microwave and Hyperspectral Infrared Observations
179(26)
The PPC/NN Algorithm
180(1)
Network Topology
181(1)
Network Training
181(1)
Retrieval Performance Comparisons with Simulated Clear-Air AIRS Radiances
181(7)
Simulation of AIRS Radiances
182(1)
An Iterated Minimum-Variance Technique for the Retrieval of Atmospheric Profiles
183(1)
Retrieval Performance Comparisons
184(1)
Discussion
185(3)
Validation of the PPC/NN Algorithm with AIRS/AMSU Observations of Partially Cloudy Scenes over Land and Ocean
188(13)
Cloud Clearing of AIRS Radiances
188(1)
AIRS/AMSU/ECMWF Data Set
188(1)
AIRS/AMSU Channel Selection
189(1)
PPC/NN Retrieval Enhancements for Variable Sensor Scan Angle and Surface Pressure
189(1)
Retrieval Performance
190(4)
Retrieval Performance Sensitivity Analyses
194(4)
Discussion and Future Work
198(3)
Summary and Conclusions
201(4)
References
202(3)
Discussion of Future Work
205(6)
Bayesian Approaches for Neural Network Training and Error Characterization
205(1)
Soft Computing: Neuro-Fuzzy Systems
206(1)
Nonstationarity Considerations: Neural Network Applications for Climate Studies
207(4)
References
209(2)
About the Authors 211(2)
Index 213
William J. Blackwell is on the technical staff at the MIT Lincoln Laboratory and is currently a science team member involved with atmospheric sounding systems aboard NPOESS and NASA EOS/NPP Missions. Frederick W. Chen was most recently a technical staff member at the MIT Lincoln Laboratory, where he worked on problems in satellite-based atmospheric remote sensing using microwave and infrared data. David H. Staelin is a professor of electrical engineering in the Research Laboratory of Electronics at MIT.