Flood

Flooding and surface water, district by district: land that radar has seen flooded again and again, where a model says large rivers could reach, how lakes and reservoirs have changed, and two past flood events. None is a live flood map.

About this module

Flood-prone land Derived by Tinga Lens Experimental shows land that radar saw under water in at least two of the years on record. Tinga Lens works this out from every Sentinel-1 image of Ghana for June to October. It is an experimental product: rice fields, salt pans and lagoon edges can be counted, and flooding in towns and under trees is missed. Land along Lake Volta is reported separately. Read the methods.

River flood hazard Published model shows the share of each district inside the zone that large rivers could reach in a 1-in-100-year flood, from the European Commission's global flood model, with the people and built-up area inside that zone. It is a model, so it covers every district, but it leaves out small streams, city drainage and the sea. Read the methods.

Flood 2023 and Flood 2020 Observed show land that Sentinel-1 radar satellites saw under water during two named flood events, the lower Volta in 2023 and northern Ghana in 2020, from the Copernicus Global Flood Monitoring service. Water that is normally there is not counted. Flooding under trees and between buildings is mostly invisible to radar, so the figures are a lower bound. Read the methods.

Surface water change Observed shows where permanent water has appeared or disappeared from 1984 to 2021, from the Global Surface Water record of Landsat images. The record ends in 2021. New water includes reservoirs, a lake refilling after its 1980s low, dugouts and water-filled mining pits, and the map cannot tell them apart. Read the methods.

The views measure different things and should not be added together. None of them is "flood risk", which would also need to account for how well people can cope.

For flood warnings, follow the Ghana Meteorological Agency, the Ghana Hydrological Authority and NADMO.