Countryside in Senegal

Prolonged drought in Kenya

1 October 2009 00:00 UTC

Countryside in Senegal
Countryside in Senegal

In 2009, highly depressed rainfall was once again observed over a large portion of the Kenya.

Last Updated

06 June 2022

Published on

01 October 2009

By Peter Ambenje and Ignatius Gitonga (KMD) and Jochen Kerkmann (EUMETSAT)

The prolonged 2009 drought in Kenya is a culmination of cumulative effects of about 11 seasons (except 2006) of depressed rainfall over most parts of the country since 2003.

In Kenya, climate change signals in the recent past have manifested themselves in increases in temperature over the country (in Nairobi, for example, the night-time and early morning (minimum) temperatures have increased from about 11°C in the 1960s to about 13°C currently) as well as in changes in rainfall patterns. Most parts of the country would normally receive significant and reliable rainfall during the 'Long Rains' season that starts in March and ends in May.

As such this is the main season for agricultural activities and also contributes significantly to other rainfall-dependent activities such as hydro-electric power generation and livestock farming among others. Over a long period of time, the rainfall received during this period over many areas is becoming less and less. In other words, there is a general deceasing trend in the 'Long Rains' season over most areas. This has adversely affected many rainfall-dependent activities in the country.

For example, during the 'Long Rains' (March to May) 2009 season, Kericho received 435mm of rainfall compared to the expected 681mm; Marsabit received a paltry 35mm compared to the expected 387mm; Nyeri got 282mm instead of 433mm and Nairobi had 295mm instead of 492mm. Dry conditions persisted over most part of Kenya until mid October 2009, save for the western parts that experienced slightly enhanced rainfall in September 2009.

There are several ways of monitoring the rainfall performance reflected by the vegetation evolution on the surface. Remote sensing is one of the various ways and depicts a very good spatial and temporal coverage. The Metop-A Normalized Difference Vegetation Index (NDVI) shown below (Figure 1) is a simple numerical indicator that can be used to monitor the density and vigor of green vegetation. It is basically a calculation of the differences between Advanced Very High Resolution Radiometer (AVHRR) channels 1 and 2.

The product shown below is a weekly composite created from daily maps by selecting, on pixel basis, the data of the day with the highest NDVI. The daily map is created selecting the most nadir pixel (from all overpasses, the pixel with the highest satellite viewing angle). Useful applications of NDVI products include classifying land cover, estimating crop acreage, and detecting plant stress. NDVI values greater than 0.1, generally, denote increasing degrees in the greenness and intensity of vegetation. Values between 0 and 0.1 are commonly characteristic of rocks and bare soil, and values less than 0 often indicate clouds or snow.

The NDVI image below shows the situation of the drought as of the beginning of October 2009. As can be seen by the red-orange colours, large areas of Kenya and Tanzania are affected by the drought with the most severe impact centered on northwest Kenya. The only area excluded from the drought is the southwest part of Kenya towards Lake Victoria.

Usually, for drought monitoring the NDVI anomaly (the difference between the average NDVI for a particular month and the average NDVI for the same month over many years) is used. If not available, one can also use animated NDVI products to follow the development of rain/vegetation during the rain season. This can be seen in the following sequence of SPOT NDVI products (Jan–Oct 2009, 10-day composites, source: VITO / SPOT-VGT4 Africa), which shows little growth of plants in northwest, central and northeast Kenya during the 2009 'Long Rains' season.

As forecast by the Kenya Meteorological Department (KMD, see press release from 26 August 2009), albeit with a little delay, October 2009 brought the eagerly awaited rains to much of East Africa, and plants started to recover. This is confirmed by the KMD report from 27 Oct 2009 and the Metop NDVI map from 5 Nov 2009 (source: NOAA).

According to Meteosat imagery, the first major convective systems hit Somalia and Eastern Kenya between 13 and 15 October 2009 (see Day Microphysics RGB product, 14 Oct 2009 12 UTC) followed by a second major event on 26–27 October (see Day Microphysics RGB product, 26 Oct 2009 12 UTC). However, around 5 November 2009 the seasonal rainfall generating system called the Inter-Tropical Convergence Zone (ITCZ) moved further south to Tanzania, Zambia and northern Mozambique. This has created dry conditions over most parts of Kenya and is likely to affect the overall seasonal rainfall performance in terms of the distribution and the total amounts.

Prolonged drought in Kenya
Figure 1: Metop Normalized Difference Vegetation Index, 1 October 2009

NDVI map for Africa, 1 Oct 2009 (source: NOAA Satellite Information Service)
NDVI map for Africa, 5 Nov 2009 (source: NOAA Satellite Information Service)

Another simple way of estimating monthly cloud and rainfall amounts is to generate monthly average satellite images, either from infrared channels or from RGB composites (see July, August, September and October averages shown below). This can be done from 15-minute imagery (best solution) or, as a kind of 'poor man's solution', from daily e.g. 12 UTC images. For the Day Microphysics RGB monthly averages shown below, blue colour denotes cloud-free /no rain areas, while red colour stands for areas with high coverage of cold ice clouds and precipitation. It can be seen from these images that, in the months July to September 2009, clouds/rains are limited more or less to southwest Kenya, while in October 2009 clouds and rains developed also in Somalia, Ethiopia and the coastal/central areas of Kenya.

Meteosat-9 Day Microphysics RGB
Figure 2: Meteosat-9 Day Microphysics RGB, July 2009 monthly average (from 12 UTC images)
Meteosat-9 Day Microphysics RGB
Figure 3: Meteosat-9 Day Microphysics RGB, August 2009 monthly average (from 12 UTC images)
Meteosat-9 Day Microphysics RGB
Figure 4: Meteosat-9 Day Microphysics RGB, September 2009 monthly average (from 12 UTC images)
Meteosat-9 Day Microphysics RGB
Figure 5: Meteosat-9 Day Microphysics RGB, 10-26 Oct average (from 12 UTC images)