Overview
On April 9, 2026, Tropical Cyclone Maila—an unusually intense and slow-moving Category 4 storm (Australian scale)—unleashed torrential rains over the Gazelle district of East New Britain, Papua New Guinea. The saturated terrain on the steep slopes of the Baining Mountains gave way, triggering multiple deadly landslides. This tutorial walks you through the process of using satellite imagery and precipitation data to analyze such events, using the PNG landslides as a real-world example. You will learn how to identify landslide scars from Landsat 9 images, interpret cyclone rainfall patterns from NASA’s Global Precipitation Measurement (GPM) mission, and understand the role of terrain and storm dynamics in landslide risk.
Prerequisites
To follow this guide, you should have:
- Basic familiarity with satellite remote sensing concepts (e.g., bands, resolution, false color).
- Access to the USGS EarthExplorer or NASA Earth Observatory to view Landsat 9 imagery.
- An understanding of tropical cyclone structure (eyewall, rainbands) and intensity scales.
- Optional: A GIS software (e.g., QGIS) to overlay layers, though visual inspection works as well.
Step-by-Step Instructions
1. Gather Cyclone Context
Start by collecting storm data. Cyclone Maila formed in the South Pacific, where the Coriolis effect is weak near the equator—a region typically at low risk for cyclones. However, record-warm sea surface temperatures and favorable atmospheric conditions allowed Maila to intensify to Category 4 (Australian) / Category 3 (Saffir-Simpson). Its slow forward speed (stalling near New Britain) caused rainbands to repeatedly strike the same area over several days. Check sources like the Australian Bureau of Meteorology or the Joint Typhoon Warning Center for track and intensity data.
2. Acquire Pre- and Post-Event Satellite Imagery
Use Landsat 9’s OLI (Operational Land Imager) to capture high-resolution (30 m) images. For the PNG case:
- Before image: September 24, 2025 — shows intact forest, no landslide scarring.
- After image: April 20, 2026 — captured during a break in the clouds, revealing fresh landslide scars.
How to download: Go to USGS EarthExplorer, set the area of interest (Gazelle district, East New Britain), select Landsat 9 Collection 2 Level-1, and filter dates. Download the “natural color” composite (Bands 4, 3, 2) for true-color interpretation, or use a false-color composite (Bands 5, 4, 3) to highlight vegetation health—landslide scars appear bright due to exposed soil.
3. Identify Landslide Scars
In the post-event image, look for linear, light-brown swaths that cut through the dark green canopy. They typically extend from ridgetops downslope toward valleys. In the PNG image, the scars run north toward the Toriu River. Compare with the pre-event image: the same areas should show uniform forest. Key indicators:
- Color contrast: Bare soil is brighter than vegetation in both true and false color.
- Shape: Elongated, sometimes fan-shaped at the toe (debris fan).
- Sediment-laden rivers: The Toriu River appears brown in the post-event image due to suspended sediment from upstream landslides.
4. Correlate with Rainfall Data
Use GPM IMERG (Integrated Multi-satellitE Retrievals for GPM) to estimate precipitation totals during the cyclone. NASA’s Earth Observatory provides near-real-time maps. For Maila, GPM data showed extreme rainfall rates—likely exceeding 500 mm over several days—concentrated over the Baining Mountains. To visualize:
- Visit the GPM data access portal and download the IMERG “late run” product for the event period (April 6–12, 2026).
- Overlay the rainfall layer on the Landsat image (using GIS or Kepler.gl). The heaviest rainfall should coincide with the landslide locations.
5. Terrain and Landslide Susceptibility
Landslides are more likely on steep slopes with unstable soils. A digital elevation model (DEM) from SRTM or ALOS can help. Steps:
- Download a DEM for the Baining Mountains (30 m resolution).
- Derive slope (in degrees) using GIS: —
slope = arctan(√(dz/dx)² + (dz/dy)²)— areas with slope > 25° are particularly hazardous. - Overlay landslide polygons from step 3 to verify they fall on steep slopes.
- Note that tropical cyclones can trigger landslides on slopes that are normally stable because of extreme soil saturation—hence even moderate slopes (15–20°) can fail during such events.
6. Integrate Findings
Create a final map or narrative that combines:
- Cyclone track and intensity (slow motion, intense rainbands).
- Pre- and post-event Landsat imagery showing change.
- GPM rainfall totals showing spatial distribution.
- DEM slope analysis confirming topographic susceptibility.
This integrated approach explains why the Gazelle district landslides occurred: unprecedented rainfall on steep terrain, compounded by the cyclone’s prolonged presence.
Common Mistakes
- Confusing clouds with landslides: In post-event imagery, residual clouds can appear bright white or light gray. Use multiple dates or a cloud-free image for verification.
- Ignoring sediment plumes: Rivers may look brown due to erosion, but that doesn’t always indicate a landslide upstream—check for direct scar-to-stream connections.
- Assuming cyclone intensity alone causes landslides: A fast-moving storm can produce less rain per area than a slow mover like Maila. Always factor in storm speed.
- Using only true color imagery: False-color (e.g., SWIR band) can better differentiate soil from vegetation. Relying solely on natural color may miss subtle scars under thin cloud or shadow.
Summary
Analyzing cyclone-induced landslides requires integrating satellite imagery, rainfall data, and terrain information. The 2026 Papua New Guinea event illustrates how an unusually slow and intense tropical cyclone—in a region normally spared from such storms—can trigger deadly landslides. By following the steps above, you can replicate this analysis for other events. The key takeaways: always compare pre- and post-event images, use GPM for rainfall context, and never overlook terrain steepness. This knowledge can help improve early warning systems and disaster response in vulnerable regions.