AI Increases Global Access to Reliable Flood Forecasts

Floods are one of the most common and impactful natural disasters, with a disproportionate impact in developing countries that often lack dense streamflow monitoring networks. Accurate and timely warnings are critical for mitigating flood risks, but accurate hydrological simulation models typically must be calibrated to long data records in each watershed where they are applied. We developed an Artificial Intelligence (AI) model to predict extreme hydrological events at timescales up to 7 days in advance. This model significantly outperforms current state of the art global hydrology models (the Copernicus Emergency Management Service Global Flood Awareness System) across all continents, lead times, and return periods. AI is especially effective at forecasting in ungauged basins, which is important because only a few percent of the world's watersheds have stream gauges, with a disproportionate number of ungauged basins in developing countries that are especially vulnerable to the human impacts of flooding. We produce forecasts of extreme events in South America and Africa that achieve reliability approaching the current state of the art in Europe and North America, and we achieve reliability at between 4 and 6-day lead times that are similar to current state of the art nowcasts (0-day lead time). Additionally, we achieve accuracies over 10-year return period events that are similar to current accuracies over 2-year return period events, meaning that AI can provide warnings earlier and over larger and more impactful events. The model that we develop in this paper has been incorporated into an operational early warning system that produces publicly available (free and open) forecasts in real time in over 80 countries. This work using AI and open data highlights a need for increasing the availability of hydrological data to continue to improve global access to reliable flood warnings.

[1]  Bryan A. Tolson,et al.  In Defense of Metrics: Metrics Sufficiently Encode Typical Human Preferences Regarding Hydrological Model Performance , 2023, Water Resources Research.

[2]  Sella Nevo,et al.  Caravan - A global community dataset for large-sample hydrology , 2023, Scientific Data.

[3]  J. Rentschler,et al.  Flood exposure and poverty in 188 countries , 2021, Nature Communications.

[4]  H. Gupta,et al.  Deep learning rainfall-runoff predictions of extreme events , 2021, Hydrology and Earth System Sciences.

[5]  Sepp Hochreiter,et al.  Large-scale river network modeling using Graph Neural Networks , 2021 .

[6]  Martin Gauch,et al.  Uncertainty Estimation with Deep Learning for Rainfall-Runoff Modelling , 2020, Hydrology and Earth System Sciences.

[7]  V. Babovic,et al.  Hydrologically Informed Machine Learning for Rainfall-Runoff Modelling: Towards Distributed Modelling , 2020, Hydrology and Earth System Sciences.

[8]  P. Salamon,et al.  Daily ensemble river discharge reforecasts and real-time forecasts from the operational Global Flood Awareness System , 2020, Hydrology and Earth System Sciences.

[9]  H. Tabari.  Climate change impact on flood and extreme precipitation increases with water availability , 2020, Scientific Reports.

[10]  M. Thieme,et al.  Global hydro-environmental sub-basin and river reach characteristics at high spatial resolution , 2019, Scientific Data.

[11]  S. Hochreiter,et al.  Toward Improved Predictions in Ungauged Basins: Exploiting the Power of Machine Learning , 2019, Water Resources Research.

[12]  Christopher Edwards,et al.  Hydrologic Modeling as a Service (HMaaS): A New Approach to Address Hydroinformatic Challenges in Developing Countries , 2019, Front. Environ. Sci..

[13]  S. Hochreiter,et al.  Towards learning universal, regional, and local hydrological behaviors via machine learning applied to large-sample datasets , 2019, Hydrology and Earth System Sciences.

[14]  Joachim Denzler,et al.  Deep learning and process understanding for data-driven Earth system science , 2019, Nature.

[15]  S. L. Sellars,et al.  “Grand Challenges” in Big Data and the Earth Sciences , 2018, Bulletin of the American Meteorological Society.

[16]  E. Wood,et al.  A Drought Monitoring and Forecasting System for Sub-Sahara African Water Resources and Food Security , 2014 .

[17]  Bryan A. Tolson,et al.  An efficient framework for hydrologic model calibration on long data periods , 2013 .

[18]  Bernhard Lehner,et al.  Global river hydrography and network routing: baseline data and new approaches to study the world's large river systems , 2013 .

[19]  J. McDonnell,et al.  A decade of Predictions in Ungauged Basins (PUB)—a review , 2013 .

[20]  L. Alfieri,et al.  GloFAS – global ensemble streamflow forecasting and flood early warning , 2012 .

[21]  Stéphane Hallegatte,et al.  A Cost Effective Solution to Reduce Disaster Losses in Developing Countries: Hydro-Meteorological Services, Early Warning, and Evacuation , 2012 .

[22]  E. Todini.  Hydrological catchment modelling: past, present and future , 2007 .

[23]  P. E. O'connell,et al.  IAHS Decade on Predictions in Ungauged Basins (PUB), 2003–2012: Shaping an exciting future for the hydrological sciences , 2003 .

[24]  D. Moorhead,et al.  Increasing risk of great floods in a changing climate , 2002, Nature.

[25]  S. Hochreiter,et al.  Long Short-Term Memory , 1997, Neural Computation.

[26]  R. Vignon.  HEALTH ORGANIZATION , 1925, Techniques hospitalieres, medico-sociales et sanitaires.

[27]  G. Nearing,et al.  NeuralHydrology - A Python library for Deep Learning research in hydrology , 2022, J. Open Source Softw..

[28]  Berit Arheimer,et al.  Global catchment modelling using World-Wide HYPE (WWH), open data and stepwise parameter estimation , 2019 .

[29]  Cred Unisdr,et al.  The Human Cost Of Natural Disasters: A global perspective , 2015 .

[30]  C. Land.  Effect of calibration data series length on performance and optimal parameters of hydrological model , 2010 .

[31]  Paul J. Pilon,et al.  Guidelines for Reducing Flood Losses , 2004 .