For agriculture, it is extremely important to know how much it rained on a particular field. However, rainfall is variable in space and time and it is impossible to have rain gauges everywhere. Therefore, remote sensing instruments such as radar are used to provide wide spatial coverage. Rainfall estimates drawn from remotely sensed observations will never exactly match the measurements that are carried out using rain gauges, due to the inherent characteristics of both sensors. Currently, radar observations are "corrected" using nearby gauges and a single estimate of rainfall is provided to users who need to know how much it rained. This competition will explore how to address this problem in a probabilistic manner. Knowing the full probabilistic spread of rainfall amounts can be very useful to drive hydrological and agronomic models -- much more than a single estimate of rainfall.