 Differentiable process-based hydrologic models, Delta models, have been shown to be able to accurately predict streamflow in ungaged areas using deep network-based parameterization pipelines. These models can also capture the effects of climate change and other environmental factors on streamflow. Additionally, they can be trained using multiple sources of data such as meteorological data, soil moisture, and evaporation. This article was authored by Defone, HBAC, K Lawson, and others.