Combine images during a flood

During Nigeria’s 2012 floods, different satellites contributed different evidence. A 2013 paper by Nigerian practitioners describes NigeriaSat-2 archival imagery as a pre-flood reference for Kogi and Bayelsa. Radar observations supplied important flood information when clouds limited new optical views. The response therefore depended on a combination of records, rather than one spacecraft providing every required image.

A baseline helps identify roads, settlements and usual water patterns before an event. It must be dated: a road may have changed since the archive acquisition, and a different season can alter river extent or vegetation. A comparison should distinguish a real change from a difference in alignment, sensor response, season or processing. An old image is useful evidence, but it is not a timeless map of normal conditions.

The two NASA images accompanying this lesson were acquired by Terra’s MODIS instrument on 20 October 2008 and 13 October 2012. They show the Niger and Benue region using visible and infrared measurements displayed in false colour. Water is blue, vegetation green and clouds pale blue-green. They are comparison images from another satellite, not NigeriaSat acquisitions.

The wider blue river areas in the 2012 view are visible evidence of the changed water extent at that observation time. Colour alone is not a universal depth gauge: different water conditions can affect its appearance. Clouds also obscure parts of the scene. These two dates cannot establish the precise time of the flood peak, the maximum extent everywhere or whether a particular building was damaged.

Complementary evidence in a flood response

The practitioner account describes radar products helping map areas that were difficult to observe optically during the response. Radar and optical images measure different interactions with the surface. Their interpretation still requires suitable geometry, processing and validation; radar is not an automatically perfect outline of every flooded street. A product should preserve which source supports each part of the map.

In a fictional comparison, Dataset A is an older clear optical image, Dataset B is a newer radar observation and Dataset C is a current optical image mostly covered by cloud. A sound workflow uses A for baseline context, evaluates B for event-time water evidence and marks C’s obscured areas as unobserved. Filling cloud-covered areas with “dry land” would invent a measurement. Keeping an explicit unknown category is more informative.