Open Data Cube and Deep Learning to Exploit Earth Observation Data Precipitation

In this video, Mónica Estébanez Camarena, Ph.D. candidate at TU Delft, talks about the Schools and Satellites (SaS) project (https://tahmo.org/schoolandsatellites/), in which she has been working since 2019. The project was born as a collaboration between TU Delft, TAHMO, PULSAQUA, Smartphones4Water, and the Ghana Meteorological Agency (GMet). SaS is one of the pilot projects of the Citizen Science Earth Observation Lab (https://www.cseol.eu/). The goal is to produce a Deep Learning model able to estimate rainfall from various Earth observation data, using citizen science rainfall measurements together with TAHMO and Gmet rain gauge data as ground-truth. The first study region is the north of Ghana but it could hopefully be extended to other regions in West Africa.

The first version of the model – RainRunner v.0 – classifies 3-hour intervals in rain/no-rain using only Meteosat thermal infrared data and achieves performances competitive with state-of-the-art satellite rainfall products.

SaS is coming to an end in December 2021, when Mónica will travel together with local partners across the north of Ghana, closing the project with workshops where the citizen scientists will learn about the outcomes of the project, the use their data were given and what will happen next with these outputs. We are really thankful to all the volunteers that have been measuring rainfall with us for the past 2 years!

After the end of SaS, Mónica will keep developing the model as part of her Ph.D.