Zachary P. McEachran

We present a knowledge‐guided machine learning framework for operational hydrologic forecasting at the catchment scale. Our approach, a Factorized Hierarchical Neural Network (FHNN), has two main components: inverse and forward models. The inverse model uses observed precipitation, temperature, and streamflow data to generate a representation of the current underlying catchment state. The forward model predicts streamflow using the learned catchment state. The FHNN architecture is designed to model multi‐scale processes and capture their interactions, a critical ability for flood modeling. FHNN also improves forecasts based on real‐time data through an inference‐based data integration approach using inverse modeling. FHNN's data integration approach improves forecasts in response to observed data more efficiently than data assimilation methods that require computationally intensive optimization. We compare the FHNN to a leading deep learning alternative (autoregressive LSTM) on the large‐sample CAMELS‐US data set, and operational flood forecast data from the US National Weather Service (NWS). Official NWS flood forecasts are generated by expert human forecasters using a physics‐based model, in a human‐in‐the‐loop process. Thus, we assess the flood forecast ability of FHNN by directly comparing its performance against these NWS expert‐derived forecasts. The human forecaster creates a more accurate forecast within the first 12–18 hr of a forecast's issuance, but FHNN has significantly better predictions thereafter. This research lays the groundwork for leveraging the predictive performance of AI‐based models with the expertise in forecasting agencies to produce better river forecasts.

We present a knowledge-guided machine learning framework for operational hydrologic forecasting at the catchment scale. Our approach, a Factorized Hierarchical Neural Network (FHNN), has two main components: inverse and forward models. The inverse model uses observed precipitation, temperature, and streamflow data to generate a representation of the current underlying catchment state. The forward model predicts streamflow using the learned catchment state. The FHNN architecture is designed to model multi-scale processes and capture their interactions, a critical ability for flood modeling. FHNN also improves forecasts based on real-time data through an inference-based data integration approach using inverse modeling. FHNN’s data integration approach improves forecasts in response to observed data more efficiently than data assimilation methods that require computationally intensive optimization. To show the performance of FHNN, we compare the FHNN to a leading deep learning alternative (autoregressive LSTM) on the large-sample CAMELS-US dataset, and operational flood forecast data from the US National Weather Service (NWS). Official NWS flood forecasts are generated by expert human forecasters using a physics-based model, in a human-in-the-loop process. Thus, we assess the flood forecast ability of FHNN by directly comparing its performance against these NWS expert-derived forecasts. The human forecaster creates a more accurate forecast within the first 18 hours of a forecast’s issuance, but FHNN has significantly better predictions thereafter. This research lays the groundwork for leveraging the predictive performance of AI-based models with the expertise in forecasting agencies to produce better river forecasts.

Empirical assessment of how forest disturbance affects streamflow has been traditionally limited to small, experimental catchments. However, larger catchments where landscape management occurs have emergent drivers of streamflow at scale, and thus may exhibit novel responses to land cover disturbance. We used statistical models of water yield and annual maximum peak streamflow for multiple large (> 50 km²) forested catchments in the low‐relief glaciated region of central North America to investigate how forest disturbance and climatic variability affect water yield and peak flows in similar landscapes. We used linear models, linear mixed effects models, and probabilistic flood‐frequency analysis with Bayesian parameter estimation in two case studies. These included: (1) a wildfire that burned ~30% of the 650 km² wilderness Upper Kawishiwi catchment, and (2) 11 catchments within the St. Louis River Basin ranging from 56 to 8880 km² with a patchwork disturbance regime wherein ~0.25% to 1% of the catchment is harvested or converted to non‐forest land use each year. We also assessed the most likely hydrological recovery year after forest disturbance, and the relative importance of stationary and nonstationary drivers of streamflow. We found forest disturbance correlated with declines in water yield for low‐level disturbance regimes in some catchments, but that water yield increased in response to the large‐scale wildfire. Positive and negative associations of forest disturbance with peak flows were observed. Hydrologic recovery time ranged from 5 to 13 years for water yield and peak flows following disturbance. Despite these effects of forest disturbance on streamflow, effects of climatic variability and stationary catchment size factors were more prominent streamflow drivers. Basins larger than ~50 km² in low‐relief glaciated regions can be impacted by forest cover change even on <30% of basin area, but climatic variability and catchment spatial scale has a larger effect than forest disturbance.

The catchment approach has been traditionally limited to small, experimental catchments where water fluxes can be determined with high accuracy. However, larger catchments where landscape management occurs have emergent drivers of streamflow at scale, and thus may exhibit novel responses to land cover disturbance. We used statistical models of water yield and annual maximum peak streamflow for multiple forested catchments in the low-relief glaciated region of central North America to investigate how forest disturbance may affect water yield and peak flows in similar landscapes. We utilized linear models, linear mixed effects models, and probabilistic flood-frequency analysis, with Bayesian parameter estimation in two case studies in Minnesota, USA: 1) a wildfire comprising ~30% of a 650km ² wilderness Upper Kawishiwi catchment, and 2) 11 catchments within the St. Louis River Basin ranging from 56 to 8,880 km ² with a patchwork disturbance regime wherein ~0.25% to 1% of the catchment is harvested or converted to non-forest land use each year. We also assessed for the most likely hydrological recovery year after forest disturbance, and the relative importance of stationary and nonstationary drivers of streamflow. We found forest disturbance correlated with declines in water yield for low-level disturbance regimes, but that water yield increased in response to the large-scale wildfire. Positive and negative associations of forest disturbance with peak flows were observed, generally with low confidence. Hydrologic recovery time ranged from 5 to 12 years for water yield and peak flows following disturbance. Despite these effects of forest disturbance on streamflow, effects of climate variability and stationary catchment size factors were more prominent drivers of streamflow. Basins larger than ~50 km ² in low-relief glaciated regions were resilient to forest cover change when it comprised <30% of basin area, but climate change may have a larger effect than could be mitigated by land management.

Managed forests generally produce high water quality, but degradation is possible via sedimentation if proper management is not implemented during forest harvesting. To mitigate harvesting effects on total watershedsediment yield, it is necessary to understand all processes that contribute to these effects. Forest harvesting best management practices (BMPs) focus almost exclusively on overland sediment sources, whereas in‐and‐near stream sources go unaddressed although they can contribute substantially to sediment yield. Thus, we propose a new framework to classify forest harvesting effects on stream sediment yield according to their direct and indirect processes.Direct effectsare those caused by erosion and sediment delivery to surface water from overland sources (e.g., forest roads).Indirect effectsare those caused by a shift in hydrologic processes due to tree removal that accounts for increases in subsurface and surface flows to the stream such that alterations in water quality are not predicated upon overland sediment delivery to the stream, but rather in‐stream processes. Although the direct/indirect distinction is often implicit in forest hydrology studies, we have formalized it as a conceptual model to help identify primary drivers of sediment yield after forest harvesting in different landscapes. Based on a literature review, we identify drivers of these effects in five regions of the United States, discuss current forest management BMPs, and identify research needs.