From SAR Data to Actionable Maps: Building an Open-Source Flood Detection Pipeline with Python

Floods can be devastating. They can happen quickly and leave communities struggling to recover. This topic is particularly important now as climate change increases the frequency of extreme weather events, making effective flood detection essential.

What is SAR Data?

Synthetic Aperture Radar (SAR) is a type of radar used to create images of landscapes. Unlike optical satellites, which rely on clear skies, SAR can see through clouds and rain. This makes it especially useful for flood detection. However, SAR data can be noisy and complex, which can complicate the analysis.

Building the Flood Detection Pipeline

In this article, we’ll walk through the steps to create a pipeline that turns raw SAR data into actionable flood maps. Here are the three main challenges we’ll address:

Challenge 1: Handling Zero-Value Borders

When using the Sentinel Application Platform (SNAP) to export SAR data, you might find zero-value borders in the images. These borders can skew your data analysis.

Solution:

1. Treat zero values as NaN (Not a Number).
2. Convert the data to decibels (dB).
3. Crop the data to only include valid information.

Here’s a sample code snippet to crop the data:

rows = np.where(~np.all(np.isnan(dB_full), axis=1))[0]
cols = np.where(~np.all(np.isnan(dB_full), axis=0))[0]
dB = dB_full[rows.min():rows.max() + 1, cols.min():cols.max() + 1]

Challenge 2: Isolating Actual Floodwater

A simple threshold of -17 dB can detect all water, but it also picks up permanent water bodies like rivers and lakes. This can lead to false positives in flood detection.

Solution:

1. Use OpenStreetMap (OSM) to fetch permanent water bodies.
2. Rasterize this data to match the SAR grid.
3. Subtract the permanent water from the detected mask to isolate the floodwater.

Here’s how you can fetch permanent water data:

osm_water = ox.features_from_bbox(bbox=bbox, tags={"natural": "water"})
flood_only = mask_clean(~binary_dilation(permanent_water, iterations=2))

Challenge 3: Addressing QGIS Rendering Issues

When exporting transparent GeoTIFFs, QGIS can render transparent pixels as solid black. This can make it hard to visualize the data correctly.

Solution:

1. Instead of using RGBA (which includes red, green, blue, and alpha for transparency), create a single-band raster.
2. Use a dedicated NoData value (255) which QGIS treats as transparent.

Here’s an example of how to set this up:

overlay = np.where(final_flood_raster == 1, 1, 255).astype(np.uint8)

The Result

By following these steps, the pipeline outputs:

• Vector Data: GeoJSON or Shapefiles with accurate area calculations.
• Interactive Maps: A Folium HTML map that allows toggling between satellite and street views.
• QGIS-Ready Rasters: Clean overlays that work immediately upon import.

Conclusion

Building a flood detection pipeline using SAR data is a valuable skill. It helps transform complex data into useful information that can aid in disaster response and planning.

Merits

• Provides accurate flood detection.
• Utilizes open-source tools, making it accessible.
• Enhances disaster management efforts.

Demerits

• Requires familiarity with Python and GIS tools.
• Processing can be resource-intensive.
• Initial setup may be complex for beginners.

Caution

This article is for educational purposes. Be sure to replace any placeholder values in the code with your actual data. Always verify claims against the original source before relying on them.

Frequently asked questions

• What is SAR data? — SAR data is radar data used to create images of landscapes, effective in cloudy conditions.
• Why is optical satellite data not useful during floods? — Optical satellites rely on clear skies, which are often not present during floods.
• What is the purpose of the flood detection pipeline? — The pipeline transforms raw SAR data into actionable flood maps.
• What programming language is used for the pipeline? — Python is used to build the flood detection pipeline.
• What is the role of OpenStreetMap in this pipeline? — OSM provides data on permanent water bodies to help isolate floodwater.
• What is QGIS? — QGIS is a free and open-source geographic information system used for mapping and analyzing spatial data.

Tags

#python #opensouce #flooddetection #SAR #GIS #QGIS #OpenStreetMap #climatechange #floodmapping #dataanalysis