Ghawar SAR time-lapse — Sentinel-1, 2015 → 2025

Area 1 (north)

Area 2 (south)

scroll = zoom · drag = pan · dbl-click = reset

Loading…

Date

Frame —

Acquired —

0 / 0

Speed

Before (A)

Date—

After (B)

Date—

Diff threshold

Sensitivity 25

Surface darkened (sand burial)

Surface brightened (sand cleared / new structure)

Diff is B − A on grayscale amplitude. A larger threshold suppresses speckle noise but also weak real changes. Both areas (north and south) are diffed together.

Decadal darkening trend (2015 → 2025)

Per-pixel OLS slope on temporally despeckled SAR amplitude, significance from FDR-controlled t-test (α = 0.05) with autocorrelation adjustment. Only pixels that survive FDR are shown.
Red = surface darkened (smoothing signature — possibly sand burial, but see caveat). Blue = surface brightened.

Overlay opacity

Hotspots

Connected FDR-significant pixels with slope < −1.0 DN/yr, area ≥ 50 px (5 000 m²). Sorted by area × |mean slope|. Click to zoom.

Important caveat. SAR amplitude darkening means surface smoothing, which can be sand burial or vegetation loss or new asphalt or moisture change. Calling this map "sand encroachment" is unverified — disambiguation requires a Sentinel-2 optical cross-check (not done). The input PNGs are amplitude-stretched (not σ⁰ in dB), so slope units are arbitrary DN/yr — useful for relative ranking, not absolute physical inference.

Trend projection — illustrative only

For pixels with an FDR-significant darkening trend, the slider shows which would cross a typical sand-sheet amplitude floor if the decadal trend continued linearly. Colour ramps by time-to- threshold: red = sooner, orange = mid, yellow = later.

Look ahead 5 yr

+1 yr (2026) +25 yr (2050)

Why this is "illustrative", not "predicted".
1. Linear extrapolation. The model can't see a dune front advancing into a currently-stable pixel — only pixels already darkening light up. Real desertification has a spatial-kinematic component this model ignores.
2. The "burial floor" (25 DN) is a visual guess on uncalibrated 8-bit amplitude. Without σ⁰ in dB, it's not a physical threshold.
3. No uncertainty propagation. The colours treat time-to-threshold as a point estimate; the actual confidence interval is wide for slopes near zero.
4. Sand burial is episodic (driven by shamal storm pulses), not linear. Forecasts past ≈ 5 years are unreliable.
Use this map for screening — pointing a field crew at the most plausible sites. Don't use the colours as dates.

Period	Cadence	Reason
2015 – Dec 2021	~6–7 days	Both Sentinel-1A and Sentinel-1B operational
Dec 2021 – Apr 2025	~12 days	Sentinel-1B failed; S1A flying solo
Post-2025 (not in this stack)	~5 days again	Sentinel-1C launched Dec 2024 — joined S1A

Ghawar SAR backscatter time-series — Sentinel-1, May 2015 → April 2025

About this AOI

Time-series shape

How to read the signal