Long-term information on irrigation water withdrawal is crucial to understand how agriculture affects water resources, but observations are often too short. We developed a statistical model to reconstruct annual irrigation water withdrawal in France from 2000 to 2022 at the catchment scale using a national database. The reconstructed time series reveal widespread increases in irrigation water withdrawal and provide a basis for water management and climate change impact studies.

To improve hydrological uncertainty estimation, recent studies have explored machine learning (ML)-based post-processing approaches. Among these, quantile random forests (QRF) are increasingly used for their balance between interpretability and performance. We develop a hydrologically informed QRF trained in a multi-site setting. Our results show that the regional QRF approach is beneficial, particularly in catchments where local information is insufficient.

Water resources will be heavily impacted by climate change in the future, with low flows and water demand satisfaction expected to decline. This study uses an integrated water resources management model to examine future water demand scenarios, revealing that climate change will be the primary driver of changes. While adapting water uses could mitigate negative impacts, this will not be enough to adapt to climate change. The irrigation sector is expected to be the most impacted.

We study the variations in annual streamflow and explicit their dependence to climate variations, in order to understand their causes and to provide tools for a rapid assessment of the impact of climate change on water resources. By making explicit the dependency of streamflow elasticity to aridity, we are able to propose a regionalized elasticity formula with physically-realistic elasticity coefficients.

We tested whether a simple model combining basic water balance principles with artificial intelligence can predict yearly river flow across many regions and years. Using data from over 3,000 river basins in eight countries, we found that the model works well when predicting future years in the same basin, but is less accurate in new regions because estimating long-term average flow remains difficult. This highlights both the promise and limits of this approach for large-scale water management.