Minimal, self-contained InSAR processing examples for learning and teaching. Each example implements a complete interferometric pipeline from scratch using only basic scientific Python libraries.
NISAR/nisar.py — NISAR Unwrapped Interferogram in 76 Lines of Python (38s)
A compact NISAR L-band interferogram processor and unwrapper using numpy, h5py, scipy and opencv (+ matplotlib for plotting). Flat linear script, every step visible top-to-bottom with no function jumps. 76 lines of Python, 38s on Apple M4.
NISAR/nisar_numpy.py — NISAR Unwrapped Interferogram in 200 Lines of Python (54s)
NISAR L-band interferogram processor and unwrapper in 200 lines of Python using only numpy and h5py (+ matplotlib for plotting). Pure signal processing in radar coordinates, 54s on Apple M4.
Both produce unwrapped interferograms matching NASA ASF product NISAR_L2_PR_GUNW_005_172_A_008_006_2000_SH_20251122T024618_20251122T024652_20251204T024618_20251204T024653_X05007_N_F_J_001.
Data: Uses NISAR RSLC HDF5 files downloaded by the notebook example NISAR L-Band HH/HV RGB composite, HH interferogram, and unwrapped phase from the InSAR.dev processing ecosystem.
Sentinel-1/s1.py — Sentinel-1 TOPS Interferogram in 243 Lines of Python (80s)
A self-contained Sentinel-1 C-band TOPS burst interferogram processor with orbital processing, TOPS deramping, amplitude cross-correlation coregistration, differential reramp, flat-earth removal, Gaussian and Goldstein filtering, multilooking, branch-cut unwrapper, and geocoding. Uses numpy, scipy, tifffile, cv2, and ortools (+ matplotlib for plotting). 243 lines of Python, 80s on Apple M4.
Uses dataset downloaded in the InSAR.dev example Erzincan Elevation, Türkiye (2019) — DEM generation from a single Sentinel-1 interferometric pair. Reproduces the ESA tutorial DEM generation with Sentinel-1 IW. The processing generates a complicated elevation map and it uses a production-ready phase unwrapper written initially for the InSAR.dev project years ago; most of the time is spent on unwrapping, 80s total.