A Hybrid Physics–AI Model to Improve Hydrological Forecasts

Abstract

The National Oceanic and Atmospheric Administration have developed a very high-resolution streamflow forecast using National Water Model (NWM) for 2.7 million stream locations in the United States. However, considerable challenges exist for quantifying uncertainty at ungauged locations and forecast reliability. A data science approach is presented to address the challenge. The long-range daily streamflow forecasts are analyzed from Dec. 2018 to Aug. 2021 for Alabama and Georgia. The forecast is evaluated at 389 observed USGS stream gauging locations using standard deterministic metrics. Next, the forecast errors are grouped using watersheds’ biophysical characteristics, including drainage area, land use, soil type, and topographic index. The NWM forecasts are more skillful for larger and forested watersheds than smaller and urban watersheds. The NWM forecast considerably overestimates the streamflow in the urban watersheds. The classification and regression tree analysis confirm the dependency of the forecast errors on the biophysical characteristics. A densely connected neural network model consisting of 6 layers (Deep Learning, DL) is developed using biophysical characteristics, NWM forecast as inputs, and the forecast errors as outputs. The DL model successfully learns location invariant transferrable knowledge from the domain trained in the gauged locations and applies the learned model to estimate forecast errors at the ungauged locations. A temporal and spatial split of the gauged data shows that the probability of capturing the observations in the forecast range improved significantly in the hybrid NWM-DL model (82±3 %) than in the NWM-only forecast (21±1 %). A tradeoff between overly constrained NWM forecast and increased forecast uncertainty range in the DL model is noted.

Enhancing Streamflow Prediction Physically Consistently Using Process-Based Modeling and Domain Knowledge: A Review

“…To develop physically consistent data-driven streamflow models, various hydrological models, ranging from large-scale models like the United States National Water Model (NWM) [61] and the Weather Research and Forecasting hydrological model (WRF-Hydro) [62], to simpler models like the Tank model [63], MISDc (Modello Idrologico Semi-Dis-…”

“…To develop physically consistent data-driven streamflow models, various hydrological models, ranging from large-scale models like the United States National Water Model (NWM) [61] and the Weather Research and Forecasting hydrological model (WRF-Hydro) [62], to simpler models like the Tank model [63], MISDc (Modello Idrologico Semi-Distribuito in continuo) [64], and Soil Moisture Accounting and Routing (SMAR) [65], have been coupled with different data-driven models. SWAT, HBV, and GR4J are widely used models (Table 1).…”

“…PBM-simulated streamflow has been used as an input variable in the DDM framework, primarily to address data scarcity and improve accuracy. This involves combining hydrological model output using data from various sources or merging different hydrological model simulation results using the same input data or both [21,32,62,[97][98][99]. Additionally, hydrological model-simulated subbasin-scale results have been utilized to enhance predictions at the outlet of the watershed or as a routing technique [100][101][102].…”