Dr Kilian Hermes

Profile

I am a postdoctoral research fellow in the Atmospheric and Cloud Dynamics research group at the Institute for Climate and Atmospheric Science (ICAS). My research is focused on data-driven approaches to weather prediction, ranging from forecast lead times of a few hours to multiple weeks. In particular, I deploy machine learning to improve weather forecasts for Africa.

My current research uses AI-based methods to produce more accurate predictions of rainfall and drought in West Africa. Such prediction support farmers to make critical decisions on planting or harvesting crops, and are crucial for food security in this region. This work is part of the Cumulus project, a collaboration between UK research institutions and Universities and meteorological agencies in Ghana and Senegal. Cumulus is funded by the Gates Foundation and the Foreign, Commonwealth and Development Office.

My PhD research was focused on mineral dust, an essential component of Earth’s climate system. I developed the first observation-based short range forecasts ("nowcasts") of dust storms, based on satellite observations and deep learning methods. The nowcast model DustCast is free and open source and allows deterministic and probabilistic short-term predictions of storms that are often missed by currently operational weather prediction models, and can help to improve forecasts for high-impact dust storms. DustCast is available as a live product through the UKCEH nowcasting portal.

An adapted version of DustCast has successfully been deployed in the WISER-EWSA project for predicting convective storms. DustCast has further been deployed for the EW4energy project in Ghana for general nowcasting purposes. Both projects are part of the United Nations Early Warnings for All initiative to ensure that everyone on Earth is protected from hazardous weather, water, or climate events through life-saving early warning systems by the end of 2027.

I am alumnus of the SENSE CDT - Centre for satellite data in environmental science, funded by NERC and the UK space agency.

Research interests

• Applied machine learning
• High-impact weather
• Satellite remote sensing
• Mineral dust
• Nowcasting
• African Meteorology

Qualifications

• PhD Meteorology, University of Leeds, United Kingdom
• MSc Meteorology, Karlsruhe Institute of Technology (KIT), Germany

Student education

I have previously demonstrated on undergraduate modules that teach data analysis and running numerical weather prediction models in UNIX-based HPC environments.

Research groups and institutes

• Institute for Climate and Atmospheric Science
• Dynamics and Clouds