The impact of atmospheric rivers on the water level and hydrology in a lake in south-Central Chile

Atmospheric rivers (ARs) are major drivers of hydroclimatic extremes in mid-latitude regions, with important implications for water resource management and climate services. This study evaluates the hydroclimatic and hydrological response of Lake Villarrica in south-central Chile to AR landfalls during 2000–2024. ARs were identified using an ERA5-based catalog and classified as Zonal (ZARs) or Tilted Atmospheric Rivers (TARs) according to integrated vapor transport direction. We identified 236 ARs, with TARs occurring more frequently, particularly during austral winter. In situ hydrometeorological observations were complemented with Sentinel-3B and Surface Water and Ocean Topography satellite altimetry, while gaps in streamflow and lake-level records were reconstructed using the Génie Rural á 4 paramètres Journalier hydrological model and a Random Forest approach, respectively. AR landfalls produced a consistent hydroclimatic response characterized by enhanced precipitation, increased minimum air temperature, reduced maximum air temperature and potential evapotranspiration, and increased streamflow and lake level. Median responses reached 19.9 mm for precipitation, 56.4 m 3 s −1 for streamflow, and 0.20 m for lake level. Positive lake-level responses occurred following 87.0% of ZARs and 89.8% of TARs. Although TARs showed slightly larger and more persistent anomalies, differences in response magnitude between AR orientations were not statistically significant. The first post-landfall local maximum occurred at a median lag of 4 days (IQR: 2–8 days) for lake level and 6 days (IQR: 3–10 days) for streamflow, demonstrating a delayed hydrological response following AR landfall. Satellite altimetry showed high agreement with ground observations, supporting its integration into operational monitoring systems. The findings highlight the potential of combining atmospheric diagnostics, hydrological modeling, and data-driven techniques to enhance climate services, particularly in forecasting hydrological responses to extreme events. This integrated framework contributes to improved risk assessment, early warning systems, and adaptive water resource management under a changing climate. Practical Implications The findings of this study provide the following actionable insights for improving water resource management and climate adaptation strategies in south central Chile, and similar mid-latitude regions affected by atmospheric rivers (ARs). The integration of satellite altimetry, hydrological modeling, and AR tracking offers a replicable framework for enhancing operational climate services. Early warning systems for extreme events AR landfalls were consistently associated with substantial and measurable hydroclimatic responses in the Lake Villarrica basin, including enhanced precipitation, increased streamflow, and rising lake levels. Positive lake-level responses occurred following approximately 90% of the analyzed ZAR and TAR events, with a median lake-level increase of 0.20 m. Importantly, the hydrological response persisted beyond AR landfall. The first post-landfall local maximum occurred after a median of 4 days for lake level and 6 days for Toltén River outflow, indicating that hydrological responses can continue for several days after landfall. This delayed response provides a potentially useful window for intensified hydrological monitoring, public warnings, and coordination with emergency services. These responses were observed for both AR orientations. Although TARs tended to produce slightly larger and more persistent anomalies in some variables, differences in response magnitude between ZARs and TARs were not statistically significant (Mann–Whitney U test, p > 0.05). Therefore, the present results do not support the use of AR orientation alone as a criterion for differentiating hydrological monitoring protocols. Instead, both ZARs and TARs should be considered capable of producing substantial hydrological responses in the Lake Villarrica basin. Because Lake Villarrica's outlet (Toltén River) is a natural, unregulated outflow, these implications are particularly relevant for downstream flood preparedness, continued post-landfall monitoring, public communication, and coordination with emergency services rather than active reservoir operations. The median lake-level increase of 0.20 m provides a quantitative measure of the typical response observed in this study; however, additional validation against observed flood impacts would be required before using lake-level changes to establish operational warning thresholds. Satellite monitoring integration The validation of Sentinel-3B and SWOT satellite altimetry against in situ observations (Spearman's Correlation of 0.93 and 0.96, respectively) demonstrates that space-based monitoring can effectively complement ground-based stations. This is particularly valuable for Lake Villarrica and other Chilean lakes, where in situ gaps are common. The SWOT mission offers broader spatial coverage and higher temporal resolution, capturing seasonal fluctuations in lake-level dynamics. For practitioners, this means that even in data-scarce regions or during periods of ground-station failure, reliable lake-level estimates can be obtained from freely available satellite products. The study's Random Forest reconstruction method demonstrates that historical lake level records can be extended back to 2000 using a combination of satellite data and meteorological variables, enabling long-term trend analysis and improved baseline assessments. While satellite altimetry products (such as SWOT and Sentinel-3B) typically involve a short data-processing and dissemination latency rather than true real-time streaming, their update cycle may complement post-landfall situational monitoring, particularly given that the hydrological responses identified here persisted for several days after AR landfall. Case study: Operational application during 2023–2024 season The 2023–2024 period provides a concrete example of the relationship between AR occurrence and lake-level variability. During this period, the study identified multiple AR events (especially TARs) coinciding with rapid increases in Lake Villarrica water level, from approximately 213.5 m to values above 215.5 m. Based on the response patterns identified in this study, such events illustrate how AR information could be incorporated into a staged monitoring framework: - Before landfall: AR forecasts and atmospheric moisture-transport information could be used to identify periods requiring increased hydrological attention - Landfall to initial response: Activate monitoring protocols, using SWOT data to track water level changes in near real-time. - Peak response window: Implement preemptive management strategies based on simulated hydrological responses. Although the GR4J model is applied here strictly for historical simulation and gap-filling rather than real-time live forecasting, its framework helps anticipate volume trends. Because Lake Villarrica drains naturally via the Toltén River without major artificial dam regulation, lake surges (such as anomalies reaching 0.2 m) directly translate into flood risks for low-lying areas, making these insights vital for mitigating downstream impacts. For Lake Villarrica, where the Toltén River represents a natural and unregulated outlet, these actions would primarily concern monitoring, public communication, and coordination with emergency services rather than active discharge management. Climate services for long-term planning Beyond event-scale responses, this study supports long-term climate adaptation. The observed interannual variability in AR frequency suggests that the character of extreme precipitation events may be shifting. Water resource planners should incorporate this evolving AR climatology into infrastructure design. The suppression of potential evapotranspiration during AR events (reductions of up to 0.6 mm day 0) also affects water balance calculations. Standard operational models that assume average PET conditions may underestimate net water gains during AR-dominated periods, leading to conservative reservoir management. Adjusting PET inputs during AR events would improve accuracy. Transferability to other regions While this study focuses on Lake Villarrica, the methodological framework is transferable to other mid-latitude lakes influenced by ARs, including those in the Pacific Northwest of the United States, western Europe, New Zealand, and Japan. The combination of (1) AR tracking using ERA5 or similar reanalysis products, (2) satellite altimetry from Sentinel-3 A and SWOT, (3) GR4J hydrological modeling, and (4) Random Forest reconstruction requires only publicly available data and open-source tools. This framework therefore provides a practical approach for investigating AR-related hydrological responses in regions where long-term in situ observations are limited, although site-specific calibration and validation would be required before operational application.