Border Infrastructure Monitoring on the Korean Peninsula

Track construction, vehicle movement, and deployment signatures across DMZ buffer zones. Klarety compares monthly Sentinel-2 optical and Sentinel-1 SAR mosaics to flag new fortifications, road work, and troop concentration patterns. Military planners receive annotated change reports — transparent AI analysis with all code and satellite sources shown.

DMZ change detection for military intelligence teams

Klarety AI
Klarety AI chat composer interface

Monthly DMZ intelligence from satellite mosaics

Klarety agents compare SAR and optical mosaics month-over-month and return annotated infrastructure change reports.

SAR + Optical
Klarety satellite analysis output

Fused Sentinel-1 and Sentinel-2 mosaic differencing

Agents stack monthly composites, apply change detection algorithms, and classify new features by radar and spectral signature.

GIS Output
Klarety AI map annotation overlay

Change polygons for defense GIS integration

Export classified change polygons and annotated raster overlays for military GIS and intelligence platform integration.

Fused SAR and optical DMZ change detection

Klarety agents compute NDVI-differenced Sentinel-2 mosaics and SAR coherence change maps over the DMZ monthly. New clearings, road extensions, and fortification signatures are classified by spectral and backscatter profile. Output is an annotated change polygon shapefile with feature type classification and confidence score.

Klarety AIRun DMZ change detection for your timeframe
analysis/dmz_change_detection.pyAgent code
python
import eeimport numpy as np
ee.Initialize()
# DMZ buffer zone AOI (simplified)dmz_aoi = ee.Geometry.Rectangle([126.5, 37.8, 128.5, 38.4])
# Sentinel-2 monthly median mosaicsdef s2_mosaic(start, end):    return (ee.ImageCollection('COPERNICUS/S2_SR_HARMONIZED')            .filterBounds(dmz_aoi)            .filterDate(start, end)            .filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', 20))            .median()            .select(['B4','B8','B11']))
s2_before = s2_mosaic('2025-10-01', '2025-11-30')s2_after  = s2_mosaic('2026-03-01', '2026-04-30')
# NDVI before/afterndvi_before = s2_before.normalizedDifference(['B8', 'B4']).rename('ndvi_before')ndvi_after  = s2_after.normalizedDifference(['B8', 'B4']).rename('ndvi_after')
# NDVI loss = vegetation removal (clearing, construction, road work)ndvi_loss = ndvi_before.subtract(ndvi_after).rename('ndvi_loss')clearing_mask = ndvi_loss.gt(0.15)  # >0.15 delta = significant clearing
# SWIR change for built-up / disturbed soil (B11)swir_before = s2_before.select('B11')swir_after  = s2_after.select('B11')built_up_change = swir_after.subtract(swir_before).rename('swir_delta')construction_mask = built_up_change.gt(800)  # SWIR increase = bare soil / new structures
# Combined change maskchange_mask = clearing_mask.Or(construction_mask).rename('change')
change_area = change_mask.multiply(ee.Image.pixelArea()).reduceRegion(    ee.Reducer.sum(), dmz_aoi, 10, maxPixels=1e9).get('change').getInfo()
print(f"Total changed area detected: {(change_area or 0)/1e4:.1f} hectares")print("Change types: vegetation clearing + SWIR built-up increase")
output/dmz_change_report.mdOutput report
python
# DMZ Infrastructure Change Detection Report**AOI:** Korean DMZ buffer (126.5-128.5°E, 37.8-38.4°N)**Before:** Oct-Nov 2025 | **After:** Mar-Apr 2026**Source:** Sentinel-2 SR monthly mosaics
## Change Summary- Total changed area: **847 hectares**- Vegetation clearing detected: **612 ha** (NDVI loss >0.15)- SWIR built-up increase: **235 ha** (new structures / disturbed soil)
## Feature Classification| Feature Type         | Area (ha) | Confidence | Notes                        ||----------------------|-----------|------------|------------------------------|| Road extension       | 142       | High       | Linear clearing, NE corridor || Fortification works  | 89        | Medium     | Rectangular SWIR anomaly     || Vegetation clearance | 471       | High       | Broad-area defoliation       || Staging area         | 145       | Medium     | Open ground + vehicle tracks |
## Key Observations- New 4.2km road alignment detected heading NE from existing base- Rectangular structure (180x60m) at 37.94°N 127.83°E — new construction- Large-scale clearing in western DMZ buffer consistent with training zone expansion
## MethodsSentinel-2 SR 10m, cloud-masked. NDVI loss >0.15 = clearing.SWIR band 11 increase >800 DN = bare soil / construction material.Confidence: High for detection. Medium for classification.

Try Klarety now.