Geophysical 2D Data Processing Platform

2D geophysical potential-field data processing platform — 28 function modules covering gridding, enhancement, field separation, spectral analysis, normalization, coordinate transformation, and layer-stack processing for gravity and magnetic data.

Overview

A comprehensive batch-processing GUI for 2D gridded potential-field data. All operators are pure-function +ops package members callable headlessly or through the GUI. Features automatic input-mode switching, multi-output support, cross-session persistence, and unified frequency-domain wavenumber conventions.

Design Principles

• Pure-function operators: Each +ops function takes (G, params) → returns G. Callable headlessly in pipelines.
• Auto input-mode switching: grid / scatter / dualGrid / stack — the GUI rebuilds input panels accordingly.
• Multi-output: Multi-scale separation, normalization-to-reference etc. produce multiple outputs; preview dropdown lets you switch.
• Cross-session persistence: All inputs auto-saved to appState.json, auto-restored on next launch.

Features (28 Modules)

Gridding

• Scatter interpolation (14 methods): 5 MATLAB built-in + 9 Surfer methods (Kriging, IDW, minimum curvature, RBF, modified Shepard, natural neighbor, triangulation, moving average, local polynomial)

Grid Math

• Gaussian smoothing: NaN-aware 2D separable Gaussian kernel convolution
• Detrend: Least-squares polynomial surface removal (1st/2nd order)

Potential-Field Enhancement

• Upward continuation: Frequency-domain exp(-|k|·h) operator
• Vertical derivative: |k|^n enhancement, n=1 or 2
• Multi-method enhancement (8 methods): Horizontal derivative X/Y, Total Horizontal Derivative (THD), Analytic Signal (AS), Tilt angle, TDX (Tilt horizontal gradient), Reduction to Pole (RTP), Downward continuation
• Euler deconvolution: Sliding-window source depth inversion (SI 0–3, Reid et al. 1990)

Field Separation (9 methods)

• Difference of Gaussian (DoG), Upward continuation, Matched filter, Interpolation cutting, Wavelet (DWT sym4), Dual-tree complex wavelet (DTCWT), Variational mode decomposition (VMD 2D), CEEMDAN, Least-squares (polynomial fit)
• Multi-scale separation: Two-stage cascade + parallel stripping with configurable scale lists, optional parfor acceleration

Grid Tools

• 2D Pull-back correction: Directional leaky integration with damping
• Directional derivative: Frequency-domain |k|^n in x/y/z/total gradient
• FK wavenumber-domain filtering: Radial bandpass with cosine taper (physical wavenumber rad/m)
• Radial power spectrum: Spector & Grant (1970) depth estimation (ln P vs k)
• Grid integration: 1D cumulative / 2D window (rectangular/trapezoidal/Simpson)
• Grid expand: Edge padding with sync of geo-headers
• Format conversion: ASC ↔ GRD ↔ MAT ↔ XYZ

Normalization

• Self-normalization: Linear (min-max) / Robust (percentile) / Anchor (piecewise with auto inflection detection)
• GRD threshold analysis: Histogram + polynomial fit + derivative root-finding for auto-anchor detection
• Normalize to reference: Map distribution B → distribution A with piecewise linear mapping

Grid Operations (dual-input)

• Time-lapse differencing: B − A with grid geometry validation
• Spectral norm matching: Optimal scaling factor minimizing ||A − αB||₂

Layer Stack Processing

• Layer difference: Adjacent depth slices → ∂f/∂z approximation
• Along-layer extraction: Vertical linear interpolation at horizon depth
• Layer-stack averaging: Thickness-average / weighted average between top/bottom horizons

Coordinate Transformation

• Grid rotation: Rigid rotation about data center with re-interpolation
• 7-parameter coordinate transform: Bursa-Wolf + Gauss-Kruger projection (via Par.txt)

Utilities

• White clip: Polygon (inpolygon) masking, keep interior or exterior
• Grid expand: padarray with geo-header sync

System Requirements

Component	Requirement
OS	Windows 10/11 (64-bit)
Runtime	MATLAB Runtime R2026a
Surfer (optional)	Golden Software Surfer 13+ for advanced gridding
Memory	8 GB recommended

Screenshots

Complete GUI layout — function tree (left), parameter panel (center), preview area (right)

Heatmap preview with interactive colormap selection (19 colormaps)

28-module processing pipeline — from gridding to layer-stack export

Euler deconvolution source model illustration

Documentation

Document	Content
用户手册 (User Manual)	Full mathematical principles, formulas, and parameter reference for all 28 modules (§4), Headless CLI API, wavenumber conventions, error handling
操作手册 (Operations Guide)	Quick-start (5 min), UI walkthrough, step-by-step operations for 4 input modes, typical workflow scenarios, parameter cheat sheet, glossary, troubleshooting
安装说明 (Installation Guide)	System requirements, license activation, toolbox dependencies, Surfer COM setup
代码说明 (Developer Guide)	Code organization, package structure, operator contracts, command-line usage

Workflow

Scatter → Grid → Smooth → Upward Continuation → Enhancement (THD/Tilt) → Euler Deconvolution
                                  │
                                  ├→ FK Filter → Spectral Analysis
                                  │
                                  └→ Field Separation → Multi-scale → Layer Stack

Headless CLI Usage

addpath('Geophysical2dDataProcessingPlatform')

% Read → Process → Write pipeline
G = core.gridRead('bouguer.asc');
G = ops.opUpward(G, struct('height', 1000));
G = ops.opGridDeriv(G, struct('direction', 'z', 'order', 1));
core.gridWrite(G, 'processed.asc');

% Batch processing
core.batchRun('./input/', '*.asc', './output/', @ops.opSmooth, struct('sigma', 3));

% Dual input
core.batchRunDual('./time1/', './time2/', '*.asc', './diff/', @ops.opTimeLapse, struct());

% Layer stack
core.batchRunStack('./slices/', '*.asc', './layers/', @ops.opLayerDiff, struct());

Output Formats

Format	Extension	Description
Esri ASCII	.asc	ArcInfo text grid, N→S row order
Surfer DSAA	.grd	Surfer 6 text grid, S→N row order
MAT	.mat	Grid struct (z, ncols, nrows, xll, yll, cellsize, nodata, name)
XYZ	.xyz	3-column X Y Z, NaN rows skipped

Wavenumber Convention

All frequency-domain operators share a unified wavenumber computation (+ops/wavenum.m):

$$k_x = \frac{2\pi}{n_x \cdot d} \cdot \text{fftAxis}(n_x), \quad k_y = \frac{2\pi}{n_y \cdot d} \cdot \text{fftAxis}(n_y), \quad |k| = \sqrt{k_x^2 + k_y^2}$$

The radial wavenumber already includes 2π (rad/m). Wavelength-to-wavenumber conversion: k = 2π / λ.

Parameter Cheat Sheet

Operator	Key Parameter	Starting Point
Scatter interpolation	method / cellsize / radius	kriging / 0(auto) / 0
Gaussian smooth	sigma	1–3
Upward continuation	height (m)	500–2000
Field enhancement	method / inc / dec	THD
Euler deconvolution	si / window	1 / 10
Field separation	method / scale	DoG / 5
FK filter	klow / khigh / taper (rad/m)	λ-dependent; taper=0.2
Radial spectrum	nBins	50
Normalization	method / outMin / outMax	linear / -1 / 1

---

24 colormaps · NaN-safe frequency-domain operators · Batch failure isolation · Full cross-session state persistence.