Hydrological Texture and Its Variations
Unified Fine-Resolution Large-Scale Hydrology Modeling
The Unified Fine-Resolution Large-Scale Hydrology Modeling agenda focuses on developing computationally efficient approaches that retain fine spatial representation of hydrologic processes while supporting simulations across large and heterogeneous domains. HYSTAR is being developed within this framework as a continuous two-dimensional hydrologic model that extends the traditional time–area concept to spatially explicit watershed simulation.
The modeling framework is intended to integrate rainfall input, infiltration and runoff generation, spatial flow routing, subwatershed interactions, and watershed response within a common computational structure. Parallel processing, spatial decomposition, and scalable numerical implementation are being explored to enable applications ranging from small watersheds to large regional systems. This development is intended to support both process-based hydrologic investigation and operational prediction while reducing the traditional tradeoff between model resolution, computational efficiency, and spatial extent.
HYSTAR model was prepared for 2038 US watersheds USGS has monitored. A user can explore the watershed and run the model with prepared input data here: https://hystarusgs.thehydrology.org/
HYSTAR
HYdrological Simulation using Time-ARea method
HYSTAR represents watershed hydrology on a grid connected by directed flow paths. Rainfall and runoff received from upstream cells are partitioned into infiltration and effective runoff; effective runoff drives hydraulic calculations and time-area redistribution, while soil-water accounting carries antecedent conditions between events. This revision describes the integrated modular source identified as version 0.4.0 and reviewed on 12 September 2026. It adds the unified watershed configuration, optional-observation behavior, selectable native parallelism, and Spark calibration execution while preserving the governing model equations.
Equations below use consistent explanatory notation. Where the implementation differs from an ideal conservation equation or textbook expression, the difference is stated rather than silently corrected. The integration checks summarized here cover configuration validation, initialization, and bounded one-event runs; they do not constitute new full-period calibration, watershed validation, or multi-node Spark performance evidence. Source identifiers are defined in the final register.
The dissertation and hydrology manuscript establish the time-area concept, reinfiltration, hourly soil moisture, and spatial runoff–soil-water feedback. The current source determines what executes in this release. Earlier package manuals contain groundwater and reservoir routines that are not proof of an active, configurable implementation in the current driver. The 2019 manuscript is incomplete and contains provisional software metadata and case-study gaps. The March 2025 manuscript was subsequently supplied and reviewed; it retains historical case-study results and incomplete evaluation sections, and is the primary source for the revised article.
The active source supports CN runoff, soil-water accounting, evaporation and transpiration, a lumped baseflow conversion, shared overland/channel routing, sediment calculations, native kernels, calibration drivers, and event-boundary restart. No Green-Ampt module was found in the active R or calibration code. GA is therefore presented as a proposed scientific extension. The inspected Spark drivers distribute model candidates and complex-evolution work; they do not partition the cells of a single simulation over Spark executors.