Humans cannot wait for the world to change to make decisions about the future. Climate change, although acute in its impacts, is a slow burn that will transform our planet over the coming generations. Today, long-term decisions are being made across sectors that will impact water supply for cities, energy generation and the feasibility of a renewable grid, and flood resilience for decades to come.
At Upstream Tech, we’re excited to officially launch HydroForecast Long-term, our solution to the complex challenge of arming those making decade-scale decisions with the best possible information about the future of water, temperature and precipitation.
HydroForecast Long-term offers streamlined access to a complete picture of how water availability may change and shift over the years to 2100. Skipping the tedious steps of processing large amounts of data, our service hands organizations decision-ready data to plan for critical infrastructure investments, water supply availability, energy generation commitments, and much more.
HydroForecast Long-term leverages the same award winning theory-guided machine learning approach found in our operational forecasts with the addition of temperature and precipitation projections from global climate models. The result is a probabilistic view of water supply over the next decades that gives organizations confidence to make climate-resilient decisions.
HydroForecast Long-term makes it easy to break the cycle of relying on past conditions to indicate what the future will look like. This has been a standard practice for many, partly due to the prohibitive resources needed to process large datasets, and also because the past was in many cases a reasonable predictor of the future. Times have changed, and research as well as anecdotal evidence suggests that climate change is causing increases in flood magnitudes and frequency<sup><a href="#1">1</a></sup>, while simultaneously increasing drought frequency and duration<sup><a href="#2">2</a></sup>.
Most existing process-based hydrologic models are not well suited to adapt to these new complexities and basin changes since they were built on assumptions between inputs (snow, rainfall) and output (streamflow) that no longer hold true as the climate warms, the jet stream shifts, and the glaciers melt. Recalibrations are possible, but they are costly, time-intensive, and manual.
HydroForecast leverages the advantages of machine learning science to teach the model underlying hydrologic relationships. Trained on a diverse set of basins, the model is constrained to make physically realistic predictions without being constrained to predict within the historical record.
Don’t see your question here? Tell us about it! Bring your questions and get in touch with our team to learn more about HydroForecast Long-term. To get an in-depth look at the model and the problems it solves, register for our upcoming webinar!
<sup id="1">1</sup> Marsooli, R., Lin, N., Emanuel, K. et al. Climate change exacerbates hurricane flood hazards along US Atlantic and Gulf Coasts in spatially varying patterns. Nat Commun 10, 3785 (2019). https://doi.org/10.1038/s41467-019-11755-z
<sup id="2">2</sup> Chiang, F., Mazdiyasni, O. & AghaKouchak, A. Evidence of anthropogenic impacts on global drought frequency, duration, and intensity. Nat Commun 12, 2754 (2021). https://doi.org/10.1038/s41467-021-22314-w