We tested whether environmental maps can be reconstructed better by selecting similar past situations locally and combining past maps in three ways, including simple averaging and two MPS-based stochastic methods. Each method has advantages: averaging was fastest and most accurate overall, while stochastic methods better preserved some spatial variability and high values. Overall, the simpler averaging approach remains the best baseline, but the preferred method depends on the goal.

Mountain streams can rise within minutes of heavy rain, but much of the streamwater may have been stored underground for far longer. We combine hydrological measurements with stable water isotopes to simulate streamflow in a Swiss Alpine catchment. The simulations show that >80 % of summer streamflow came from groundwater, while direct rainfall contributions were small and brief. Fast stream rises therefore do not necessarily mean that rainwater moves quickly through the landscape.

Streamflow observations contain information about snow, but their potential to constrain seasonal snow mass reconstructions remains underexplored. Using inverse hydrological modeling, we show that streamflow is particularly effective at constraining catchment-aggregated melt rates, but that non-uniqueness in the snow–streamflow relationship and uncertainties in the inverse modeling chain can easily limit inversion performance.

Our study presents a method to estimate high-resolution snow water equivalent (HR-SWE) using low-resolution climate data (LR-CD). By using a data-driven approach, we analyze historical weather patterns from LR-CD to generate HR-SWE maps. Machine learning and statistical relationships between LR-CD and HR-SWE enable estimation for dates without HR-SWE data. This method enhances water resource management and climate impact assessments, especially in data-scarce regions.

We aim to combine multiple global climate models (GCMs) to enhance the robustness of future projections. We introduce a novel approach, called "α pooling", aggregating the cumulative distribution functions (CDFs) of the models into a CDF more aligned with historical data. The new CDFs allow us to perform bias adjustment of all the raw climate simulations at once. Experiments with European temperature and precipitation demonstrate the superiority of this approach over conventional techniques.