What we made

An APP (no UI) that shows how Kyrgyzstan’s glaciers are melting now - and how they may change next. Built for water, energy, and safety planning.

Why it matters

Glaciers power drinking water, farming, and hydropower. We knew they’re melting - now we quantify how fast and where first for Ala-Archa and Bishkek.

The science (simple)

We use SAR (radar from space) that works through clouds and at night. We process Sentinel-1 time series to get change maps, trends, and confidence bands. A trained LSTM (Long Short-Term Memory) model adds short-term forecasts.

How it works

Ingest Sentinel-1 scenes → clean, align, terrain-correct.
Compute backscatter/coherence change and basic masks (elevation/aspect).
Output maps, charts, % loss, and a GLOF watchlist.

Architecture (no frontend)

Python-only toolkit: a set of scripts/modules, each responsible for its own step (download, preprocess, change metrics, ML forecast, export).
Run via CLI with a simple config (AOI, dates).
Outputs: GeoTIFF/GeoJSON/CSV/PNG ready for QGIS and reports.
No web frontend—fast, reproducible, offline-friendly.

Results: Golubin Glacier (Ala-Archa Gorge)

2016–2025: glacier tongue shrank by ~1–2% (tongue section only).
Forecast to 2050 (pessimistic): up to ~30% loss vs 2015.
Note: real loss is higher due to thinning and side retreat. We’ll refine with more SAR and improved ML.

Local impact

Water & energy: plan for seasonal shortages; smarter hydropower ops.
Hazards: target glacial lakes / GLOF watch zones.
Scales up: same pipeline runs on other Ala-Archa glaciers and nearby basins.

Tech stack

Python (NumPy, Pandas, SciPy, rasterio, Matplotlib), Sentinel-1 SAR, QGIS, LSTM ML, and the Tengri Analysis & Processing scripts/modules (no frontend).

upload_image.03:10:37.252749 upload_image.03:12:36.775997

2025 NASA Space Apps Challenge

TengriSpacers