Flooding in West Africa

By Simon R Proud, Rasmus Fensholt, Stephanie Horion (University of Copenhagen)

Unusually heavy rainfall in the summer and autumn of 2012 has caused widespread flooding in West Africa. These floods have caused problems across the region, with Nigeria hit particularly hard, Reliefweb stated that more than 400 have died and more than one million forced to move since early July. As well as the direct threat posed by rising waters the flood also presents an array of indirect threats. Inundated land can prove to be a breeding ground for locusts, a pest that has a large impact upon the growth of crops.

The wet conditions also increase the danger posed by diseases — particularly those transmitted through water such as cholera. However, the heavy rains can also be a benefit as farmers have reported that the increased rainfall has encouraged crop growth, lessening the chances of food shortages later in the year. Therefore, the accurate monitoring of the spatial extent and duration of yearly flooding of West African rivers is important in the assessment of the continued feasibility of traditional flood recession agriculture, but can also improve monitoring of hazards to people, irrigated agriculture, buildings and infrastructure.

The Spinning Enhanced Visible and InfraRed Imager (SEVIRI) aboard Meteosat Second Generation (MSG) is particularly useful for flood monitoring in Africa. The geostationary orbit of MSG provides a constant view of the African continent and the high temporal resolution maximises the likelihood of gaining a cloud-free view of the land surface — something that has proven to be a problem for polar-orbiting sensors such as MODIS[1]. The two images in Figure 1 shows the recent flooding in West Africa from the perspective of SEVIRI. One image is from mid-October 2012 and the other from mid-October 2008.

The Senegal river in the upper-left of the image and the Niger and Benue rivers towards the bottom-right all show much larger areas of water in 2012 than they did in 2008. The images shown in this figure are derived from the Red, Near-InfraRed and Short-Wave InfraRed bands and have been corrected so as to more closely resemble the natural colours of the Earth. As well as allowing a direct visual comparison of flooded area these images can be processed to produce a number of other variables — such as water indices and BRDF parameter values — that are useful in quantifying the area of land that has been affected by flooding.

Such processing is done at the University of Copenhagen on a daily basis to produce daily cloud-free land surface reflectances, flooded area maps[2], crop growth information[3] and rainfall estimates. The latter is of value not only in monitoring of flood events, but also when combined with hydrological models, in producing short-term forecasts of possible flooding.

The SEVIRI does, however, have the drawback of low spatial resolution. In the majority of spectral bands SEVIRI is limited to a pixel spacing of greater than 3km, which compares unfavourably to polar instruments that typically have a pixel spacing of less than 1km. This means that it is difficult to measure the flood extent with high accuracy; particularly as for small rivers and tributaries the flood may cover an area that is much smaller than the SEVIRI pixel size. But the High Resolution Visible (HRV) channel aboard SEVIRI can help overcome this.

This broadband channel operates across the visible spectrum with a pixel spacing of around 1km at nadir. Because it is broadband it cannot be used to produce the other variables derived from the other SEVIRI channels. It is still of great use in quantifying the small scale extent of floods, as shown in Figure 2. A number of flooded areas are visible in the HRV images that are not visible in the colour composite.

The much higher spatial resolution of this channel allows more precise estimation of flooded area, as can be seen here when comparing the right-hand side of the two images.The Benue river covers much larger areas in 2012.

The images were received and processed by the Earth Observation Group at the University of Copenhagen.

References

[1] Fensholt, R. et al (2011) Analysing the advantages of high temporal resolution geostationary MSG SEVIRI data compared to Polar Operational, International Journal of Applied Earth Observation and Geoinformation, 13, 721--729
[2] Proud, S.R., Fensholt, R., Rasmussen, L.V. and Sandholt, I. (2011) Rapid response flood detection using the MSG geostationary satellite, International Journal of Applied Earth Observation and Geoinformation, 13, 536--544
[3] Proud, S.R. and Rasmussen, L.V. (2011) The influence of seasonal rainfall upon Sahel vegetation, Remote Sensing Letters, 2, 241--249