Hydrological Droughts in the Netherlands: from Simulations to Projections using Data-Driven Methods

Abstract

Droughts in Europe have become more frequent and recent examples such as the droughts in 2003, 2015, 2018, and 2022, highlight their impact. The past few extreme drought events showed that even countries that are usually less affected by droughts, due to their large natural water availability, experience increasing challenges related to them. One prime example includes the Netherlands, which is usually known for its abundance of water and advanced water management system. While this system was originally designed to prevent floods and reduce flood impacts, the past extreme drought events showed that this design choice leads to additional drought impacts. This made it clear that a shift in water management strategies is needed to improve drought preparedness for future drought events. In this thesis, which was a joint collaboration with the National Water Authority of the Netherlands (Rijkswaterstaat), the current water management challenges related to drought events in the Netherlands are addressed by exploring the potential of machine learning techniques for simulating, forecasting and projecting of hydrological droughts, and the effect of water management on potential drought impact mitigation. By combining practical insights from stakeholders with recent scientific developments, showing the prospect and strength of such collaborations, this thesis highlights the potential of machine learning techniques and hybrid modelling approaches to improve hydrological modelling for drought monitoring, forecasting and projections and calls on water managers to advance the state-of-the-art of operational water management with the use of machine learning techniques to reduce future drought impacts in the Netherlands.