Unlocking Earth Observation Data for AI Agents with Katzilla

Why Earth Observation Data Matters for AI Agents

As of April 24, 2026, climate modeling, precision agriculture, disaster response, and logistics optimization all depend on real-time and historical environmental data. The problem has never been a shortage of data — NASA's GES DISC alone hosts petabytes of it. The problem has been *access*: mismatched formats, scattered endpoints, and authentication overhead that burns engineering hours before a single byte of useful signal reaches your model.

Katzilla's Earth Observation category solves this. Today we're spotlighting five datasets you can hit right now through katzilla.dev/api/v1/earth-observation.

---

Datasets in Focus

1. MERRA-2 Land Surface Diagnostics (M2T1NXLND v5.12.4)

NASA's Modern-Era Retrospective Analysis for Research and Applications delivers 1-hourly, time-averaged land surface diagnostics at 0.625 × 0.5 degree resolution. Surface runoff, soil moisture layers, and latent heat flux are all here — invaluable for training hydrological models or feeding irrigation-optimization agents.

2. GPM MHS on METOP-B Brightness Temperature (GPM_1CMETOPBMHS)

The Global Precipitation Measurement mission's common-calibrated brightness temperature product gives you 17 km resolution microwave observations every 1.5 hours. Pair this with a precipitation forecasting model to get ahead of supply-chain disruptions caused by extreme weather.

3. MODIS/Aqua NRT Thermal Anomalies & Fire (5-Min L2 Swath, 5 km)

Near Real Time fire detection at 5 km resolution, refreshed every five minutes of overpass time. Wildfire risk agents can subscribe to this feed and trigger downstream alerts or rerouting logic the moment a thermal anomaly is confirmed.

4. SMEX03 Regional Ground Soil Moisture — Oklahoma v1

In-situ ground truth from the Soil Moisture Experiment campaign remains one of the best validation references for satellite-derived moisture estimates. Use it to calibrate ML models that blend remote sensing with local sensor networks.

5. RSS SSM/I Ocean Product Grids — DMSP F14 Daily NetCDF v7

Daily oceanic wind speed, water vapor, cloud liquid water, and rain rate on a global grid. Shipping-route optimization agents and offshore-energy planners have been heavy users of this dataset since Katzilla surfaced it through a normalized JSON interface.

---

Querying Earth Observation Data with the Katzilla SDK

Install the SDK and fire your first query in under two minutes:

npm install @katzilla/sdk

import { KatzillaClient } from '@katzilla/sdk';

const client = new KatzillaClient({ apiKey: process.env.KATZILLA_API_KEY });

async function getThermalAnomalies() {
  // Query MODIS/Aqua NRT fire detections for a bounding box in Oklahoma
  const response = await client.earthObservation.query({
    dataset: 'MODIS_AQUA_NRT_FIRE_L2',
    date: '2026-04-24',
    bbox: {
      minLon: -100.0,
      minLat: 33.5,
      maxLon: -94.4,
      maxLat: 37.0
    },
    params: ['fire_mask', 'brightness_temperature', 'frp'],
    format: 'json'
  });

  const hotspots = response.features.filter(f => f.properties.fire_mask === 'fire');
  console.log(`Active fire hotspots detected: ${hotspots.length}`);
  return hotspots;
}

getThermalAnomalies();

Swapping dataset to MERRA2_M2T1NXLND or GPM_1CMETOPBMHS follows the exact same pattern — no new authentication, no format translation, no custom parsers.

---

What to Build Next

• Wildfire alert agent: Poll MODIS_AQUA_NRT_FIRE_L2 every 30 minutes, cross-reference with MERRA2_M2T1NXLND soil dryness, and push Slack alerts when risk thresholds are crossed.
• Precision irrigation pipeline: Join SMEX03 ground moisture with RSS_SSMI_F14_DAILY atmospheric vapor to produce field-level watering recommendations.
• Shipping weather router: Feed GPM_1CMETOPBMHS brightness temperatures into a precipitation classifier and reroute container vessels in real time.

All 300+ Katzilla datasets share the same SDK interface. Start exploring at katzilla.dev/docs/earth-observation and ship your first Earth-aware agent today.