Locust Swarm and Agricultural Pest Early Detection

Identify locust swarm activity from satellite vegetation damage signatures before ground teams confirm. Klarety agents process high-frequency Sentinel-2 time-series to detect anomalous vegetation loss consistent with pest defoliation, then generate region-level alert packages for spray campaign coordination.

Set up alerts

Early pest detection from vegetation damage signatures

Klarety AI

Pest alert packages from satellite vegetation data

Klarety agents detect anomalous vegetation loss in Sentinel-2 time-series and return region-level pest alert reports.

NDVI - Sentinel-2

High-frequency defoliation signature detection

Agents compare rolling NDVI composites to flag sudden vegetation loss inconsistent with seasonal growth patterns.

GIS Output

Defoliation vectors for agricultural response GIS

Export defoliation event polygons with severity scores for spray campaign planning and FAO early warning system integration.

Rapid NDVI defoliation detection for locust alerts

Klarety agents compute 10-day NDVI composites from Sentinel-2, detect pixels with NDVI drops exceeding 2 standard deviations of inter-annual variability in a 2-week window, and classify events by spatial pattern — locust swarm signatures show clustered, irregular defoliation distinct from drought patterns. Output is an alert polygon GeoJSON with affected area, severity, and onset date.

Klarety AISet up locust early warning for your region

analysis/locust_defoliation_alert.pyAgent code

python

import eeimport numpy as npfrom datetime import datetime, timedelta
ee.Initialize()
# East Africa locust monitoring zoneaoi = ee.Geometry.Rectangle([40.0, 2.0, 44.0, 6.0])
def ndvi_composite(start, end):    return (ee.ImageCollection('COPERNICUS/S2_SR_HARMONIZED')            .filterBounds(aoi)            .filterDate(start, end)            .filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', 25))            .map(lambda img: img.normalizedDifference(['B8','B4']).rename('ndvi'))            .mean())
# Build 5-year inter-annual variability baseline (same 10-day window)baseline_years = [2020, 2021, 2022, 2023, 2024]target_window = ('04-01', '04-10')  # 10-day window to monitor
baseline_imgs = [ndvi_composite(f'{y}-{target_window[0]}', f'{y}-{target_window[1]}')                 for y in baseline_years]baseline_coll = ee.ImageCollection(baseline_imgs)ndvi_baseline_mean = baseline_coll.mean().rename('ndvi_mean')ndvi_baseline_std  = baseline_coll.reduce(ee.Reducer.stdDev()).rename('ndvi_std')
# Current 10-day windowcurrent = ndvi_composite('2026-04-01', '2026-04-10')
# Anomaly z-scorez_score = ndvi_baseline_mean.subtract(current).divide(ndvi_baseline_std.add(0.01))
# Alert thresholdsalert_mask  = z_score.gt(2.0).rename('defoliation_alert')severe_mask = z_score.gt(3.0).rename('severe_defoliation')
alert_area = alert_mask.multiply(ee.Image.pixelArea()).reduceRegion(    ee.Reducer.sum(), aoi, 10, maxPixels=1e9).get('defoliation_alert').getInfo()
severe_area = severe_mask.multiply(ee.Image.pixelArea()).reduceRegion(    ee.Reducer.sum(), aoi, 10, maxPixels=1e9).get('severe_defoliation').getInfo()
print(f"Alert area (z>2.0): {(alert_area or 0)/10000:.0f} ha")print(f"Severe area (z>3.0): {(severe_area or 0)/10000:.0f} ha")print(f"Alert onset: Apr 1-10, 2026")

output/locust_alert_report.mdOutput report