Nepal Flood Response Mapping
Satellite-based potential inundation analysis | August 2026
Introduction
This analysis examines potential flood inundation in the Bhote Koshi region of Nepal using satellite imagery and Google Earth Engine. The objective was to compare pre-event and post-event surface conditions and identify areas showing signs of newly developed water coverage.
Analysis Question
What changed after the event?
The analysis uses satellite observations to compare surface conditions before and after the event and identify locations where water presence appears to have increased. This is an analytical screening workflow, not an official disaster map.
Before & After
Pre-event reference vs latest post-event acquisition
Pre-event Sentinel-2 composite
27 July – 27 August 2026
Pre-event Reference
Sentinel-2 surface reflectance composite built from imagery acquired between 27 July and 27 August 2026, serving as the baseline for change detection.
Latest post-event acquisition
29 August 2026 · Tile 45RUL
Post-event Acquisition
Latest available Sentinel-2 acquisition after the event, dynamically identified from the Google Earth Engine archive. Tile 45RUL.
Visual placeholders represent the analytical composite periods. No fabricated imagery is shown. The interactive Earth Engine application displays the actual satellite observations.
Methodology
Analysis Workflow
Sentinel-2
Surface reflectance imagery acquisition
Cloud Masking
SCL-based removal of clouds, shadows, cirrus, and snow
MNDWI
Modified Normalized Difference Water Index calculation
Change Detection
Pre-event vs post-event MNDWI comparison
DEM / Slope Filtering
SRTM-derived slope constraint (< 15°)
Historical Water Context
JRC Global Surface Water occurrence masking
Potential Inundation
Terrain-constrained analytical output layer
Study Area
The Bhote Koshi region of Nepal, a steep mountain river valley prone to seasonal flood events and terrain-driven hazards.
Pre-event Reference
A Sentinel-2 surface reflectance composite was built for the period 27 July – 27 August 2026 to establish baseline surface conditions.
Dynamic Post-event Search
The workflow searches the available Sentinel-2 archive after the event and dynamically identifies the latest available acquisition, rather than relying on a single hard-coded date.
Cloud Masking
The Sentinel-2 Scene Classification Layer (SCL) is used to remove cloud shadow, medium and high probability clouds, cirrus, and snow or ice contamination.
Water Detection
The Modified Normalized Difference Water Index (MNDWI) is calculated to enhance the contrast between water and surrounding land surfaces.
Temporal Change Detection
Pre-event and post-event MNDWI values are compared to identify areas showing increased water presence relative to the baseline.
Terrain Filtering
The SRTM DEM is used to derive slope, and potential inundation is constrained to lower-slope terrain (slope < 15°) to reduce false positives on steep slopes.
Historical Water Context
JRC Global Surface Water is used to provide historical water-occurrence context and reduce confusion with persistent water bodies.
Final Analytical Layer
A terrain-constrained potential inundation layer is produced, representing locations where satellite-derived water indicators increased and satisfied all applied constraints.
Datasets
Data & Methods
Copernicus Sentinel-2 Surface Reflectance
Optical satellite imagery providing multi-spectral surface reflectance for water index computation and change detection.
SRTM DEM
Shuttle Radar Topography Mission digital elevation model used to derive slope and constrain analysis to low-slope terrain.
JRC Global Surface Water
Historical water-occurrence dataset used to distinguish newly developed water from persistent water bodies.
Google Earth Engine
Cloud computing platform enabling satellite data access, processing, and dynamic archive search at scale.
Key Parameters
Analysis configuration
> 0.1
> 0.2
Slope < 15°
27 Jul – 27 Aug 2026
29 Aug 2026
45RUL
The workflow updates as new imagery becomes available
The workflow does not permanently depend on a single post-event image. It searches the available Sentinel-2 archive after the event and identifies the latest available acquisition, allowing the analysis to update as new imagery becomes available through Google Earth Engine. This dynamic post-event analysis is one of the key strengths of the approach.
Results
Current Analytical Result
~0.03 km²
Terrain-constrained potential inundation
The analysis identified approximately 0.03 km² of terrain-constrained potential inundation within the latest available Sentinel-2 imagery coverage.
Result Interpretation
The detected areas represent locations where satellite-derived water indicators increased relative to the pre-event reference and satisfied the applied terrain and historical-water constraints. This is an analytical indication requiring further investigation.
Important
Limitations & Interpretation
- The output represents a satellite-derived analytical indication of potential inundation.
- It is NOT an officially mapped flood boundary.
- It should NOT be interpreted as a confirmed flooded area.
- The latest Sentinel-2 acquisition does not provide complete fresh coverage of the entire study area.
- Therefore, the approximately 0.03 km² estimate represents only the terrain-constrained potential inundation detected within the latest available imagery coverage.
- It should NOT be interpreted as the total flood extent across the Bhote Koshi region.
- Optical satellite observations can also be affected by cloud cover, image availability, terrain effects, and other sources of uncertainty.
- Independent validation would be required before using the result for operational decision-making.
Recommended Validation Sources
Interactive Application
Explore the Interactive Analysis
The Google Earth Engine application provides an interactive interface for exploring the satellite observations, derived water indicators, temporal change, terrain constraints, and potential inundation results.
Google Earth Engine Application
Explore Interactive AnalysisTechnology
Tools & Methods
"This analysis demonstrates how Earth Observation data can be transformed into a rapid, reproducible workflow for identifying areas that may warrant further investigation during flood events."