Exploring Water Management Strategies for Mitigating Local Drought Impacts in the Netherlands using Data-Driven methods previously used for Simulations to Projections

Extreme events like droughts and floods can have significant impact on the environment and society. As a result effective water management strategies are necessary to limit and mitigate these impacts. In the past decade, the Netherlands has experienced several extreme drought events, raising increasing interest in adapting water management practices, traditionally focused on floods, to address droughts more directly and effectively. Machine learning techniques have previously been tested for the same region in terms of seasonal forecasting1 and projections2 under different warming scenarios3, showing the additional benefit of these techniques in downscaling large-scale input data to local-scale relevant information for water managers. This recent work aims to go a step further to explore water management options for drought mitigation by incorporating machine learning in a framework of hydrological simulations, water management scenarios and impact functions. By incorporating the insights gained from previous work, a closer focus is given on human aspects and its impact on local drought management. We developed a Multi-Target Long Short-Term Memory (LSTM) model which facilitates the exploration of different water management options. An essential finding is that taking proactive actions earlier can further limit drought impacts and help to mitigate long recovery periods that would have been observed otherwise. With the Multi-LSTM water management model we can potentially reduce drought impact by 3-5% for the droughts in 2003, 2015 and 2018. As a results, this work yields valuable insights for operational water management and potential improvements in water management strategies with machine learning techniques to effectively address future drought events. 1) Hauswirth, S. M., Bierkens, M. F. P., Beijk, V., and Wanders, N.: The suitability of a seasonal ensemble hybrid framework including data-driven approaches for hydrological forecasting, Hydrol. Earth Syst. Sci., 27, 501517, https://doi.org/10.5194/hess-27-501-2023, 2023. 2) Hauswirth SM, van der Wiel K, Bierkens MFP, Beijk V and Wanders N (2023) Simulating hydrological extremes for different warming levelscombining large scale climate ensembles with local observation based machine learning models. Front. Water 5:1108108. doi: 10.3389/frwa.2023.1108108 3) Van der Wiel, K., Wanders, N., Selten, F. M., Bierkens, M. F. P. (2019). Added value of large ensemble simulations for assessing extreme river discharge in a 2 C warmer world. Geophysical Research Letters, 46, 2093 2102.