Large Sample Basin Attribute Generation and Interpretation

In recent years several large-sample hydrometeorological datasets have been developed and used as inputs in both process-based and machine learning hydrological models, often for runoff prediction in ungauged basins.  Large sample hydrology datasets take information from a rapidly evolving array of geospatial data sources to create indices describing basin attributes associated with runoff-generating processes.  

In this study we discuss nuances of computational representation of basins, attribute interpretation with respect to physical processes, attributes vs. applications, the rate of change of spatial information sources, and the rapid growth and use of open source software tools.  Preliminary findings from generating a large sample dataset of ungauged basin attributes (~1M basins) are presented to support convergence towards standardized computational methods for basin attribute selection and calculation.