DEM Generation from InSAR Using Sentinel-1 & ESA SNAP

Interferometric Synthetic Aperture Radar (InSAR)-Based Terrain Reconstruction Workflow

---

Table of Contents

• Overview
• Project Objective
• Study Area
• Data Selection Strategy
• Data Source & Acquisition
• Image Selection Criteria
• Required Software
• Methodology
• Workflow Summary
• Results
• Challenges Encountered
• Reproducibility Notes
• Tools Used
• Future Improvements
• References

---

Overview

This project demonstrates the generation of a Digital Elevation Model (DEM) from Sentinel-1 Synthetic Aperture Radar (SAR) imagery using Interferometric SAR (InSAR) processing techniques in ESA SNAP.

By exploiting phase differences between two SAR acquisitions, terrain elevation was reconstructed for the Delamar Mountain Wilderness, Nevada, USA.

---

Project Objective

To generate an accurate, terrain-corrected DEM from Sentinel-1 SLC imagery through interferometric processing and phase unwrapping while minimizing coherence loss and atmospheric distortions.

---

Study Area

Delamar Mountain Wilderness, Nevada, USA

Nevada was selected due to its favorable environmental conditions for InSAR processing:

• Arid desert climate
• Sparse vegetation cover
• Minimal surface water bodies
• Low relative humidity (~30–38%)

These conditions improve interferometric coherence and reduce atmospheric interference.

Terrain Characteristics

Feature	Description
Elevation Range	790 – 1800 ft
Topography	Deep winding canyons
Landforms	Rocky peaks & steep cliffs
Slope Features	Long sloping bajadas

---

Data Selection Strategy

DEM generation using InSAR requires:

• Stable surface reflectivity between acquisitions
• Minimal vegetation-induced decorrelation
• Low atmospheric moisture

To reduce atmospheric distortion, imagery was selected from Nevada’s dry season:

Dry Season Window

May – September

Chosen Acquisition Months

• May
• August

---

Data Source & Acquisition

SAR imagery was obtained from the Copernicus Programme.

Dataset Specifications

Parameter	Value
Satellite	Sentinel-1
Product Type	SLC (Single Look Complex)
Acquisition Mode	IW (Interferometric Wide Swath)
Download Portal	Copernicus Data Hub

Additional image IDs and study boundaries are available in the helplist.txt file.

---

Image Selection Criteria

To ensure reliable interferometric DEM generation:

1. Orbit Direction Consistency

Both images must:

• Be acquired in Ascending orbit
• Maintain identical viewing geometry

2. Perpendicular Baseline Constraint

To ensure sufficient topographic sensitivity:

• Perpendicular Baseline > 150 m

Verification Tool

• ASF Vertex

---

Required Software

Software	Purpose
ESA SNAP	InSAR Processing
SNAPHU	Phase Unwrapping
Google Earth Pro	DEM Visualization

---

Methodology

1. Preprocessing

• Update and install SNAP plugins
• Reinstall SNAPHU after output path modification
• Import Sentinel-1 SLC scenes

2. Image Preparation

• Apply S-1 TOPS Split

  • Select subswath
  • Select polarization
  • Select burst covering AOI

• Apply Precise Orbit File

3. Co-registration

• Perform S-1 Back Geocoding
• Apply Enhanced Spectral Diversity

4. Interferogram Generation

• Perform Interferogram Formation
• Run TOPS Deburst

5. Filtering

• Apply Goldstein Phase Filtering
• Perform Multilooking
  • Range = 5
  • Azimuth = 1

6. Phase Unwrapping

• Export to SNAPHU with:

  • Cost Mode = TOPO
  • Row Overlap = 300
  • Column Overlap = 300
  • Tiling = 4×4

• Run SNAPHU Unwrapping

• Import unwrapped output back into SNAP

7. DEM Extraction

• Perform Phase to Elevation
• Save DEM output

8. Final Correction

• Perform Terrain Correction
• Export final DEM as KMZ

---

Workflow Summary

Import Data
   ↓
TOPS Split
   ↓
Apply Orbit File
   ↓
Back Geocoding
   ↓
Enhanced Spectral Diversity
   ↓
Interferogram Formation
   ↓
TOPS Deburst
   ↓
Goldstein Filtering
   ↓
Multilooking
   ↓
SNAPHU Export
   ↓
SNAPHU Unwrapping
   ↓
Phase to Elevation
   ↓
Terrain Correction
   ↓
Export DEM

---

Results

Generated DEM Preview

---

Challenges Encountered

High Computational Load

SNAPHU phase unwrapping was computationally expensive.

Solution Implemented

• Used 4×4 tiling during SNAPHU export.

Side Effect

• Introduced minor bounding line artifacts in the unwrapped interferogram.

---

Reproducibility Notes

To reproduce this workflow accurately:

• Use Sentinel-1 SLC IW imagery only
• Maintain same orbit direction for image pair
• Ensure baseline > 150 m
• Select dry-season acquisitions
• Preserve all listed SNAP/SNAPHU parameters
• Maintain identical filtering/multilooking settings

---

Improvements to DEM

Potential enhancements to improve DEM quality:

• Experiment with alternative filtering techniques
• Test different baseline thresholds
• Compare results using varied SNAPHU tiling configurations
• Integrate Ground Control Points (GCPs) for validation
• Quantitatively assess DEM accuracy using RMSE

---

⭐ If you found this project helpful, consider starring the repository!