Artificial Island Construction Tracking in the South China Sea

Monitor dredging activity and base fortification in disputed waters at 5-day intervals. Klarety agents process Sentinel-2 multispectral imagery to detect sediment plumes, measure facility footprint expansion, and track material stockpile volumes. Each report includes annotated expansion maps and volume change estimates.

Start monitoring

Territorial expansion tracking from satellite imagery

Klarety AI

Facility expansion reports from one AOI prompt

Klarety agents process Sentinel-2 time-series and return dredging extent maps and facility footprint change reports.

Sentinel-2

Dredging plume and construction phase detection

Agents compute NDWI anomalies and RGB change signatures to classify dredging intensity and construction phase at 5-day revisit.

GIS Output

Expansion layers for geopolitical GIS monitoring

Output facility footprint change layers and stockpile volume estimates for territorial monitoring dataset integration.

Dredging plume and footprint expansion detection

Klarety agents compute NDWI time-series from Sentinel-2 to detect turbid dredging plumes and map newly reclaimed land area. Facility footprint polygons are compared quarter-over-quarter to measure expansion in hectares. SWIR change signatures classify new construction materials from bare reclaimed sand. Output includes expansion polygon overlays and area change tables.

Klarety AITrack island development in your AOI

analysis/scs_island_expansion.pyAgent code

python

import eeimport numpy as np
ee.Initialize()
# Fiery Cross Reef AOIaoi = ee.Geometry.Rectangle([113.8, 9.5, 114.1, 9.7])
def get_mosaic(start, end):    return (ee.ImageCollection('COPERNICUS/S2_SR_HARMONIZED')            .filterBounds(aoi)            .filterDate(start, end)            .filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', 15))            .median()            .select(['B3','B8','B11','B12']))
before = get_mosaic('2024-01-01', '2024-03-31')after  = get_mosaic('2026-01-01', '2026-03-31')
# NDWI: water = high positive, land/fill = negativedef ndwi(img):    return img.normalizedDifference(['B3', 'B8']).rename('ndwi')
ndwi_before = ndwi(before)ndwi_after  = ndwi(after)
# Land mask: NDWI < 0 = non-water (reclaimed land, structures)land_before = ndwi_before.lt(0).rename('land_before')land_after  = ndwi_after.lt(0).rename('land_after')
# New land: land after but not beforenew_land = land_after.And(land_before.Not()).rename('new_land')
# Turbid plume detection: high green reflectance in waterwater_mask = ndwi_after.gt(0.1)b3_after = after.select('B3')turbid = b3_after.updateMask(water_mask).gt(1200).rename('dredge_plume')
new_land_area = new_land.multiply(ee.Image.pixelArea()).reduceRegion(    ee.Reducer.sum(), aoi, 10, maxPixels=1e9).get('new_land').getInfo()
print(f"New reclaimed land area (2024-2026): {(new_land_area or 0)/1e4:.2f} hectares")

output/scs_expansion_report.mdOutput report