Oil Storage Tank Fill Level and Inventory Monitoring

Track crude and refined product levels at terminals and strategic reserves. Klarety agents analyze Sentinel-1 VV backscatter to estimate tank fill levels, compare month-over-month changes across tank farms, and flag inventory builds or draws before official EIA reporting. 10-meter resolution enables site-level granularity.

Monitor reserves

Satellite inventory intelligence for energy traders

Klarety AI

Inventory estimates ahead of EIA releases

Klarety agents estimate tank fill levels from SAR backscatter and return site-level inventory reports before official data drops.

SAR - Sentinel-1

VV backscatter fill level estimation at 10m

Agents process descending orbit passes over tank farm AOIs and compute fill proxies from surface reflectivity changes.

GIS Output

Tank farm layers for energy GIS pipelines

Export per-tank fill estimates as attributed polygons for terminal capacity and pipeline flow GIS dataset integration.

SAR backscatter tank fill level estimation

Klarety agents process monthly Sentinel-1 descending passes over tank farm AOIs, compute mean VV backscatter per tank polygon, and convert to fill level proxies using a calibrated inverse relationship (fuller tanks return lower dB). Month-over-month delta flags builds and draws. Output is a per-tank fill table and time-series chart at 10m resolution.

Klarety AIMonitor tank inventory at your terminal

analysis/cushing_tank_fill_levels.pyAgent code

python

import eeimport numpy as npimport pandas as pd
ee.Initialize()
# Cushing Oklahoma tank farm AOIcushing_aoi = ee.Geometry.Rectangle([-97.02, 35.88, -96.62, 36.08])
def get_mean_vv(start, end):    """Monthly mean VV backscatter over tank farm area."""    s1 = (ee.ImageCollection('COPERNICUS/S1_GRD')          .filterBounds(cushing_aoi)          .filterDate(start, end)          .filter(ee.Filter.eq('instrumentMode', 'IW'))          .filter(ee.Filter.eq('orbitProperties_pass', 'DESCENDING'))          .select('VV')          .mean())    val = s1.reduceRegion(        ee.Reducer.mean(), cushing_aoi, 10    ).get('VV').getInfo()    return round(val, 3) if val else None
# Monthly backscatter time seriesmonths = [    ('2025-10-01', '2025-10-31', 'Oct 2025'),    ('2025-11-01', '2025-11-30', 'Nov 2025'),    ('2025-12-01', '2025-12-31', 'Dec 2025'),    ('2026-01-01', '2026-01-31', 'Jan 2026'),    ('2026-02-01', '2026-02-28', 'Feb 2026'),]
# Calibration: VV dB to fill level proxy# Empirical: -13.5 dB ~ 35% fill, -12.0 dB ~ 20% filldef vv_to_fill_pct(vv_db, vv_empty=-12.0, vv_full=-14.5):    fill = (vv_db - vv_empty) / (vv_full - vv_empty) * 100    return max(0, min(100, fill))
results = []for start, end, label in months:    vv = get_mean_vv(start, end)    if vv:        fill = vv_to_fill_pct(vv)        results.append({'month': label, 'vv_db': vv, 'fill_pct': round(fill, 1)})
df = pd.DataFrame(results)print(df.to_string(index=False))df.to_csv('/app/output/cushing_fill_levels.csv', index=False)

output/cushing_fill_report.mdOutput report