Some stations give a weak match with the ERA5 reanalysis: annual (October 1st - September 31st) total rainfall.

Stations from Mozambique and Zimbabwe give a good match with ERA5.

The Burundi rain gauge data broadly agree with ERA5 total annual rainfall.

There is an excellent match between annual total rainfall derived frpom Tanzanian rain gauges and ERA5 reanalysis.

Output using an older version of the esd-package. When a wider group of rain gauge measurement are used, then there is a lesser match with ERA5.

A simple formula based on the wet-day frequency and wet-day mean precipitation (black curve) gave a good description of the number of days per year with more than 20 mm - observed numbers marked by red curve.

The attendants of the workshop computed future climate projections for local rainfall based on a small selection of CMIP6 (SSP3-70) global climate model simulations. Similar projections were also derived for the wet-day frequency and wet-day mean precipitation.

The band of downscaled results from the small ensemble matches the year-to-year variations quite well. The historical trend is consistent with the results, 

The band of downscaled results from the small ensemble matches the year-to-year variations quite well. The historical trend is slightly upwards, whereas the model results may seem to be flatter.

A crude estimate of the probability of heavy rainfall, and hence how many to expect per year, is based on two key parameters: the wet-day frequency and the wet-day mean precipitation.