The Sahel Rainfall Index was developed by Zwarts et al. (2009) and updated by Zwarts et al. (2023a); please refer to these sources for detailed insight into the data and methodology used.

The Sahel Rainfall Index as shown on this site is updated annually, which has been made possible by largely automating the underlying data collection and analysis process. Here, we provide a brief overview (based on Zwarts et al. 2023a) of the methodology used to calculate the Sahel Rainfall Index shown on this site. The data used are from a variety of sources; data for recent years come mainly from NCAR’s monthly precipitation data datasets, supplemented by data from Senegal (ANACIM).

Rain falls in the Sahel during only a few months of the year, mostly as the result of occasional tropical storms. Patterns of precipitation are erratic and vary between sites, which means estimating average annual rainfall for the Sahel requires data from many weather stations each year. The earliest annual rainfall measurements on record for the region are from stations along the Atlantic coast (e.g. Saint Louis, starting in 1848, and Dakar, starting in 1853). In later years, measurements were also made at stations further inland (e.g. Timbuktu, starting in 1897). Rainfall across the entire Sahel was first measured in 1920. The number of weather stations in the Sahel gradually increased during the first part of the 20^th century, peaking at 860 in 1961, and has since declined to 103 stations in 2020 (Figure A).

To quantify year-to-year variation in rainfall, multiple researchers over the years have published a rainfall index for the Sahel (see Zwarts et al. 2023a for a summary). These give annual rainfall as percent deviation from the long-term average, and in many cases were standardised by dividing percent deviation by standard deviation. However, because the number of weather stations has declined and rainfall measurements at some stations were collected only on an irregular basis, it is harder to calculate the Sahel Rainfall Index for recent years. To address this issue and maximize the number of weather stations used, Zwarts et al. (2023a) used years with complete datasets to calculate, for each station, the percentage of annual rainfall that occurred in each month, then applied these average figures to estimate monthly rainfall in years where data was missing. For any given year, each station was included in the dataset only if it measured more than 70% of the annual rainfall that year. This approach introduced additional error, but it incorporated data from twice as many weather stations for the years since 2000 and more than three times as many stations for the years since 2018.

The next steps were to convert annual rainfall at each of the selected stations into percent deviation from that station’s average over the period 1950–1975, and calculate the average percent deviation for all stations combined. This dataset was then used to identify correlations between each individual station and the average across all stations. Some stations—most either in the desert to the north or the hyper-humid zone to the south—were uncorrelated or only weakly correlated with the average percent deviation, and we omitted them from the index (see Zwarts et al. 2023a for details). In total, we used data from 148 rainfall stations to calculate the Sahel Rainfall Index presented on this site. The number of stations used per year and the standard error are shown in the figure below. Corresponding data can be downloaded on the Home page.