Skip to content

Signal Processing

CSIR NET Earth Sciences Part C geophysics: signals, Fourier series, correlation, LTI systems, sampling, FFT, Z-transform, filters, deconvolution, fractals.

4
Topics
86
Min read

Chapter Overview

Signal Processing is the first unit of the Geophysics section of Part C in CSIR NET Earth, Atmospheric, Ocean and Planetary Sciences (subject code 702). Every geophysical measurement ends up as a signal: a seismogram in time, a gravity or magnetic profile in distance, a well log in depth, a tide record in hours. This unit is the toolkit for turning those records into information: describing them, comparing them, sampling them correctly, computing their spectra, filtering them and undoing the distortion of the recording chain.

The mathematics is the same as in electrical engineering and physics. What is particular to geophysics is the use: the convolutional model of a seismic trace, vibroseis correlation, minimum-phase wavelets, predictive deconvolution of multiples, and the power-law (fractal) behaviour of many Earth signals. Questions mix short derivations, hand calculations on three- or four-sample sequences, and statement evaluation.

What You’ll Learn

By the end of this chapter you will be able to:

  • Classify signals (continuous or discrete, periodic, energy or power) and test whether a sampled sinusoid is periodic
  • Expand periodic signals in Fourier series, use symmetry, Parseval and the Gibbs effect, and apply standard transform pairs
  • Compute auto- and cross-correlations by hand, find delays, and use the Wiener–Khinchin theorem and vibroseis correlation
  • Test systems for linearity, time invariance, causality and stability, and find outputs by convolution
  • Apply the sampling theorem, find aliases, and read DFT resolution, leakage and FFT operation counts
  • Use Z-transforms with their regions of convergence, place poles and zeros, and recognise minimum-phase wavelets
  • Specify filters in decibels, compare FIR and IIR and the three phase types, and design least-squares filters
  • Explain spiking and predictive deconvolution, and measure fractal dimension from box counts, dividers and spectral slopes

Prerequisites


Learning Path

The topics follow the order of the official syllabus lines:

graph TD
    A["1. Continuous and Discrete Signals,
Fourier Series and Transforms"] --> B["2. Auto- and Cross-Correlation
and LTI Systems"] A --> C["3. Sampling Theorem, FFT,
Z-Transform and Convolution"] B --> C B --> D["4. Filters, Deconvolution
and Fractal Analysis"] C --> D D --> E["Quick Revision"]

Topics in This Chapter

TopicWhat it coversTime
Continuous and Discrete Signals, Fourier Series and Transformssignal classes, energy and power, elementary signals, periodicity of discrete sinusoids, trigonometric and complex Fourier series, Dirichlet conditions, symmetry, coefficient decay, Gibbs, Parseval, transform pairs and properties, DTFT and DFT30 min
Auto- and Cross-Correlation and LTI Systemscorrelation definitions and properties, correlation coefficient, Wiener–Khinchin, vibroseis and Klauder wavelet, system properties, impulse response, convolution, causality and BIBO stability, frequency response, trace model, random inputs30 min
Sampling Theorem, FFT, Z-Transform and Convolutionideal sampling, Nyquist frequency, aliasing (time and space), quantisation, DFT resolution, zero padding, leakage and windows, radix-2 FFT, linear and circular convolution, Z-transform pairs, ROC, poles and zeros, minimum phase35 min
Filters, Deconvolution and Fractal Analysisfilter types, decibels, Butterworth, FIR and IIR, phase types, running average, Wiener normal equations, inverse filtering, spiking and predictive deconvolution, prewhitening, water reverberation, cepstrum, fractal dimension, box counting, dividers, Hurst exponent, spectral slope, \( D = 2b \)35 min
Quick Revisionevery formula and trap on one page10 min

Official Syllabus Lines and Where They Are Covered

Official topic lineCovered in
Continuous and discrete signals, Fourier seriesContinuous and Discrete Signals, Fourier Series and Transforms
Auto- and cross-correlation, LTI systemsAuto- and Cross-Correlation and LTI Systems
Sampling theorem, FFT, Z-transform, convolutionSampling Theorem, FFT, Z-Transform and Convolution; convolution is introduced with LTI systems in Auto- and Cross-Correlation and LTI Systems
Filters, deconvolution, fractal analysisFilters, Deconvolution and Fractal Analysis

How CSIR NET Asks It

Per the NTA information bulletin for December 2026, Earth Sciences Part C has 80 questions, of which 25 are to be attempted, at 4 marks each with −1.32 for a wrong answer, a maximum of 100 marks (full table on the Earth Sciences overview). If more than 25 Part C questions are answered, only the first 25 answered are evaluated. There is no official count of questions per unit. From the syllabus wording, expect these kinds of question:

Question typeTypical contentWhere to practise
Recognitionenergy or power signal; is a sequence periodic; which harmonics a waveform has; linear, time-invariant, causal, stable; FIR or IIR; phase typeLevel 1 and 2 on each page
One-step numericalfundamental period; Parseval power; Nyquist frequency; alias; DFT bin spacing; FFT operation count; Butterworth attenuation; box-counting dimensionLevel 2
Hand computation on short sequencesautocorrelation, cross-correlation lag, linear and circular convolution, inverse of a dipole, a 2-point Wiener filterLevel 3
Statement evaluationcorrelation properties; ROC and stability; minimum phase; deconvolution assumptions; Hurst exponent and spectral slopeLevel 3

Quick Reference

QuantityValue or formula
Fourier coefficient\( c_n = \frac{1}{T}\int_T x\,e^{-in\omega_0 t}dt \)
Parseval (series)\( P = a_0^2 + \frac{1}{2}\sum(a_n^2 + b_n^2) \)
Cross-correlation\( R_{xy}(\tau) = \int x(t)\,y(t + \tau)\,dt \leftrightarrow X^{*}Y \)
Wiener–KhinchinPSD = Fourier transform of autocorrelation
LTI output\( y = x * h \), \( Y = XH \); stable iff \( \sum\lvert h\rvert < \infty \)
Nyquist\( f_N = 1/(2\Delta t) \); alias \( \lvert f - kf_s\rvert \)
DFT spacing\( \Delta f = 1/(N\Delta t) \)
FFT cost\( (N/2)\log_2 N \) vs \( N^2 \)
Z-transform\( a^n u[n] \leftrightarrow 1/(1 - az^{-1}) \), \( \lvert z\rvert > \lvert a\rvert \)
Butterworth\( \lvert H\rvert^2 = 1/[1 + (f/f_c)^{2n}] \), \( 6n \) dB/octave
Fractal dimension\( D = \log N/\log(1/r) \); profile \( D = 2 - H = (5 - \beta)/2 \)

Start Learning

Begin with the first topic:

→ Continuous and Discrete Signals, Fourier Series and Transforms

Advertisement
Advertisement