Flow Conditioned Parameter Grids for Mechanistic, Statistical, and Machine Learning Hydrologic Models: A Seamless Basin Characteristic Dataset for the Contiguous United States

Abstract:

Mechanistic, statistical, and machine learning models of hydrologic systems require parameters representative of land-surface conditions. Often, parameters are computed for modeling units or the area upstream of a forecast point using a delineated boundary and zonal statistics, which require time-consuming delineation of modeling units. Instead, to efficiently generate model parameters or basin characteristics, we introduce the Flow Conditioned Parameter Grid (FCPG) Tools, an open-source package to pre-compute model parameters using a flow direction grid. A FCPG is generated by first deriving surface water flow direction and upstream contributing area grids from a digital elevation model for a selected area. Second, the parameter of interest, either gridded data or rasterized points, lines, or polygons, is accumulated following the flow direction grid. Third, the accumulated parameter grid is normalized by the upstream contributing area grid to produce a grid where each cell holds the mean upstream value of the parameter. The calculation of the final grid can also account for missing data. The FCPG can then be queried to, for example, parameterize a mechanistic hydrologic model, provide basin characteristics for statistical models of high and low flows, train a machine learning model, or drive Web applications that serve hydrologic predictions for user-specified locations. This software package has been used to generate a seamless dataset of common basin characteristics for the contiguous United States including mean upstream elevation, slope, minimum and maximum monthly temperature, annual precipitation, land cover class, and latitude. These data will be publicly available via a Web service from the USGS. The FCPG Tools can be used to extend this dataset or to create new, purpose-built FCPGs.