We use the framework of conformal prediction to investigate the impact of temporal and spatial information on uncertainty estimates within hydrological predictions.
Abstract
Uncertainty estimates are fundamental to assess the reliability of predictive models in hydrology. We use the framework of conformal prediction to investigate the impact of temporal and spatial information on uncertainty estimates within hydrological predictions. Integrating recent information significantly enhances overall uncertainty predictions, even with substantial gaps between updates. While local information yields good results on average, it proves to be insufficient for peak-flow predictions. Incorporating global information improves the accuracy of peak-flow bounds, corroborating findings from related studies. Overall, the study underscores the importance of continuous data updates and the integration of global information for robust and efficient uncertainty estimation.
Links
The code repository is available at https://doi.org/10.5281/zenodo.10674231. The trained base model (LSTM) and the utilized model states, as well as the global HopCPT models, are available at https://doi.org/10.5281/zenodo.10653863. The trained CMAL models for the non-PUB experiments are available at https://doi.org/10.5281/zenodo.10654345, and the CMAL models for the PUB experiments are available at https://doi.org/10.5281/zenodo.10654399.
Citation
@Article{auer2024hopcpt,
author = {Auer, A. and Gauch, M. and Kratzert, F. and Nearing, G. and Hochreiter, S. and Klotz, D.},
title = {A data-centric perspective on the information needed for hydrological uncertainty predictions},
journal = {Hydrology and Earth System Sciences},
volume = {28},
year = {2024},
number = {17},
pages = {4099--4126},
doi = {10.5194/hess-28-4099-2024}
}