Enhancing Catchment Based Hydrological Model Performance through Dynamic Sub-Discretization Using Land Surface Attributes

Abstract Submission: To improve NOAA’s operational water prediction capabilities, version 4.0 of the National Water Model (NWM) will be reformulated as a mosaic of performant models using the Next Generation Water Resources Modeling Framework (NextGen). The design of the NextGen Framework allows execution of scientifically selected model formulations connected to a common hydrofabric. The hydrofabric defines the catchments, flowlines, water bodies and their connectivity at any desired resolution. In line with this, conceptual or physical-conceptual catchment based hydrological models are often preferred over high-resolution models due to their reduced computational demands, making them more practical for large-scale or long-term simulations. However, the simplifications in lumped models can lead to reduced accuracy when applied uniformly in catchments with significant land-surface heterogeneity.

This study explores the potential to enhance lumped model performance by introducing dynamic discretization based on key land-surface heterogeneities such as land use, soil type and soil layering, slope aspect, and elevation. By incorporating varying levels of discretization to create Hydrological Response Units (HRUs) that reflect spatio-temporal variations in watershed characteristics, we aim to improve predictive capabilities without significantly increasing computational costs. We implement this discretization and compare models performance due to these heterogeneities with those applied uniformly, using the Conceptual Functional Equivalent (CFE) and the Catchment Arid/Semi-arid Model (CASAM) hydrological models. The study focuses on different hydroclimatic regions, including semi-arid and humid subtropical environments in select CAMELS (Catchment Attributes and Meteorology for Large-Sample Studies) basins with a focus on identifying conditions under which the consideration of sub-catchment scale heterogeneities yields significant performance gains. This research offers a method for selecting optimal discretization strategies and demonstrates how dynamic discretization can enhance the performance of conceptual and physical-conceptual models used for operational water management and climate change adaptation efforts.

Keywords: Conceptual Functional Equivalent (CFE), National Water Model, Dynamic Hydrological Response Unit, Rainfall-Runoff Modeling

Learning Objectives/Expected Outcome (Optional) :