Natural Monthly Flow Estimates for the Conterminous United States, 1950-2015

This metadata record describes monthly estimates of natural stream flows for greater than 2.5 million stream reaches, defined by the National Hydrography Dataset (NHD) Version 2.0, in the conterminous United States for the period 1950-2015. A statistical machine learning technique - random forest modeling - was applied to estimate natural flows using 200 potential predictor variables. The dataset is organized by level 3 ecoregions, with each ecoregion having its own directory. Separate .csv files for each stream reach in the NHD network belonging to a given ecoregion are contained within the respective ecoregion directories. The stream reach .csv files are named as the NHD COMID corresponding to that stream reach. Methods used to calibrate the random forest models, and references to predictor data sources are detailed in the Journal of the American Water Resources Associate paper "A database of natural monthly flow estimates from 1950-2015 for the conterminous United States" by Miller and others (2018). See the discussion paper link in the "Related External Resources" section for access.