This dataset contains selected basin characteristics from the GAGES-II dataset (https://water.usgs.gov/lookup/getspatial?gagesII_Sept2011) that were considered for use in regression equations for predicting flow-duration curves as well as additional variables that were constructed or derived from these basin characteristics. Descriptions of basin characteristics used in regression equations are provided (BCdescriptions.xlsx). Additional processing of the selected basin characteristics was done on a region-by-region basis, including deletion of variables with a large (greater than 50%) fraction of the same value (usually zero). A table indicating the basin characteristics considered for use in regression equations for each study region is provided (BCs_byRegion.xlsx). After transformation to reduce skewness and to convert to a similar range of values, these basin characteristics were used to fit regression equations for predicting FDCs. Basin characteristic selection and transformation were performed separately for each study region using all the USGS streamgages that were identified as being of “reference” quality in the GAGES-II dataset. Of the reference quality streamgages, only those having at least 10 complete water years (WYs) during the study period from WY1981 through WY2013 were used in regression model building; these are identified in the dataset. The basin characteristics are presented in their original units (unstransfBCs.new.xlsx) and as transformed (transfBCs.new.xlsx); a table specifying the transformations is provided (tranfBC_info.new.xlsx). Methods are fully described by Over and others (in review).

This dataset contains the selected regression models interpreted in the concurrent publication (Over and others, 2018). To each group of gaged basins in each region with one basin left out, a regression model selection procedure was applied to select the best set of basin characteristics for fitting regression models to each quantile in a set of up to three contiguous groups of FDC quantiles (called here a “flow regime”) for each of four or five binary methodology choices (thus 16 or 32 methodological combinations). This dataset provides the definitions of the selected flow regimes, the regression model parameters (intercepts and coefficients of the selected basin characteristics and their standard errors), variance inflation factor (VIF) values of the selected basin characteristics, goodness-of-fit statistics (adjusted-R-squared and AIC) of the fitted models, and predicted FDCs for each of the methodological combinations. This dataset contains a zipped file (Model_info.zip) of 1 or 2 CSVs per study region containing model information grouped by the number of basin characteristics considered in the regression models.

This dataset contains the selected regression models interpreted in the concurrent publication (Over and others, 2018). To each group of gaged basins in each region with one basin left out, a regression model selection procedure was applied to select the best set of basin characteristics for fitting regression models to each quantile in a set of up to three contiguous groups of FDC quantiles (called here a “flow regime”) for each of four or five binary methodology choices (thus 16 or 32 methodological combinations). This dataset provides the definitions of the selected flow regimes, the regression model parameters (intercepts and coefficients of the selected basin characteristics and their standard errors), variance inflation factor (VIF) values of the selected basin characteristics, goodness-of-fit statistics (adjusted-R-squared and AIC) of the fitted models, and predicted FDCs for each of the methodological combinations. This dataset contains a zipped file (Model_info.zip) of 1 or 2 CSVs per study region containing model information grouped by the number of basin characteristics considered in the regression models.

This data release contains the input used and the output files interpreted in the publication "Refinement of a Regression-Based Method for Prediction of Flow-Duration Curves of Daily Streamflow in the Conterminous United States". This data release contains daily streamflow data for 1,378 streamgages in 19 study regions in the conterminous U.S. (CONUS) from October 1, 1980 through September 30, 2013 from mostly undisturbed watersheds. This data release also contains the empirical flow-duration curves (FDCs) derived from this daily streamflow data, presented as 27 quantiles ranging from 0.02 to 99.98 percent nonexceedance probabilities. Selected basin characteristics from the GAGES-II dataset (https://water.usgs.gov/lookup/getspatial?gagesII_Sept2011) were transformed to reduce skewness and to convert to a similar range of values and then used, directly or in the construction of additional variables, to fit regression equations for predicting FDCs. Basin characteristic selection and transformation was performed separately for each study region. The basin characteristics considered for use in candidate regression models are presented in their original units and as their transformed values; a table specifying the variable transformations is also provided. To each group of gaged basins in each region with one basin left out, a regression model selection procedure, with four or five binary methodological choices thus 16 or 32 methodological combinations, was applied to select the best regression model for each contiguous group of FDC quantiles (called here a “flow regime”). This data release provides the definitions of the selected flow regimes, the regression model parameters, goodness-of-fit statistics of the fitted models, and predicted FDCs for the gaged basin that was left out for the selected FDC regression models for each of the methodological combinations.

This data release contains the input used and the output files interpreted in the publication "Refinement of a Regression-Based Method for Prediction of Flow-Duration Curves of Daily Streamflow in the Conterminous United States". This data release contains daily streamflow data for 1,378 streamgages in 19 study regions in the conterminous U.S. (CONUS) from October 1, 1980 through September 30, 2013 from mostly undisturbed watersheds. This data release also contains the empirical flow-duration curves (FDCs) derived from this daily streamflow data, presented as 27 quantiles ranging from 0.02 to 99.98 percent nonexceedance probabilities. Selected basin characteristics from the GAGES-II dataset (https://water.usgs.gov/lookup/getspatial?gagesII_Sept2011) were transformed to reduce skewness and to convert to a similar range of values and then used, directly or in the construction of additional variables, to fit regression equations for predicting FDCs. Basin characteristic selection and transformation was performed separately for each study region. The basin characteristics considered for use in candidate regression models are presented in their original units and as their transformed values; a table specifying the variable transformations is also provided. To each group of gaged basins in each region with one basin left out, a regression model selection procedure, with four or five binary methodological choices thus 16 or 32 methodological combinations, was applied to select the best regression model for each contiguous group of FDC quantiles (called here a “flow regime”). This data release provides the definitions of the selected flow regimes, the regression model parameters, goodness-of-fit statistics of the fitted models, and predicted FDCs for the gaged basin that was left out for the selected FDC regression models for each of the methodological combinations.

Digital datasets were used to develop basin characteristics values that are used in multiple regression equations and tested for the use in predicting flow-duration curves (FDCs) in ungaged areas of Indiana. Several basin characteristics are easily derived from StreamStats basin delineations, such as basin area (https://streamstats.usgs.gov/ss/). Other basin characteristics require ancillary datasets as input. The data provided through this data release are those data that have been collected, tested, and ultimately selected as a basis for FDC development. These include PRISM 3-Month Average Precipitation, Thickness and Coarseness of Quaternary Sediments, and Soil Available Water Capacity. There are 6 continuous parameter grids (CPGs; for a definition, see https://pubs.usgs.gov/ds/412/section3.html) included in the Indiana portion of this data release: PRISM December to February Average Precipitation (prdecfeb00_cpg), PRISM September to November Average Precipitation (prsepnov00_cpg), index of permeability of Quaternary surficial sediments (qss_permb_cpg), index of permeability of Quaternary surficial sediments multiplied by their thickness (permbxthick_cpg), STATSGO Permeability (statsperm_cpg), and Available Water Capacity (watcapinil_cpg).

Digital datasets were used to develop basin characteristics values that are used in multiple regression equations and tested for the use in predicting flow-duration curves (FDCs) in ungaged areas of Indiana. Several basin characteristics are easily derived from StreamStats basin delineations, such as basin area (https://streamstats.usgs.gov/ss/). Other basin characteristics require ancillary datasets as input. The data provided through this data release are those data that have been collected, tested, and ultimately selected as a basis for FDC development. These include PRISM 3-Month Average Precipitation, Thickness and Coarseness of Quaternary Sediments, and Soil Available Water Capacity. There are 6 continuous parameter grids (CPGs; for a definition, see https://pubs.usgs.gov/ds/412/section3.html) included in the Indiana portion of this data release: PRISM December to February Average Precipitation (prdecfeb00_cpg), PRISM September to November Average Precipitation (prsepnov00_cpg), index of permeability of Quaternary surficial sediments (qss_permb_cpg), index of permeability of Quaternary surficial sediments multiplied by their thickness (permbxthick_cpg), STATSGO Permeability (statsperm_cpg), and Available Water Capacity (watcapinil_cpg).

This dataset contains predicted flow-duration curves (FDCs) for 1,378 streamgages in 19 study regions in the conterminous U.S. from mostly undisturbed watersheds using regression models constructed with four or five binary methodological choices (thus 16 or 32 methodological combinations). The predicted FDCs are presented as 27 quantiles ranging from 0.02 to 99.98 percent nonexceedance probabilities. These data are also contained within the CSV files in child item 4, "Selected regression models for predicting FDC quantiles for selected streamgages in the conterminous United States", but are separated out here to provide more convenient access in a single Excel file (predictedFDCs.xlsx). These data support a concurrent publication (Over and others, 2018).

This dataset contains predicted flow-duration curves (FDCs) for 1,378 streamgages in 19 study regions in the conterminous U.S. from mostly undisturbed watersheds using regression models constructed with four or five binary methodological choices (thus 16 or 32 methodological combinations). The predicted FDCs are presented as 27 quantiles ranging from 0.02 to 99.98 percent nonexceedance probabilities. These data are also contained within the CSV files in child item 4, "Selected regression models for predicting FDC quantiles for selected streamgages in the conterminous United States", but are separated out here to provide more convenient access in a single Excel file (predictedFDCs.xlsx). These data support a concurrent publication (Over and others, 2018).

This dataset contains the empirical flow-duration curves (FDCs) derived from complete water years of daily streamflow data for 1,378 streamgages in 19 study regions in the conterminous U.S. from October 1, 1980 through September 30, 2013 from mostly undisturbed watersheds contained in child item 1, "Daily streamflow data for selected streamgages in the conterminous United States", of this data release. The empirical FDCs are presented as 27 quantiles ranging from 0.02 to 99.98 percent nonexceedance probabilities. Because streamflow data less than 0.005 cfs are reported as zero, they are considered to be censored values. To handle these censored data values, two versions of the FDC quantiles from streamgage records were computed: (1) empFDCs.unfilled.xlsx - where the quantiles were estimated from the original data and (2) empFDCs.filled.xlsx – where the censored quantile values were filled with estimated positive values. With the method used for filling the censored quantiles, which relies on a lognormal fit to the data, occasionally the data values estimated for the largest censored values were larger than the smallest noncensored data values. This sometimes resulted in increases to the quantiles greater than the censoring level. As a result, some of the noncensored flow quantile values in the filled dataset are greater than the corresponding noncensored flow quantile values in the unfilled dataset. Methods are fully described by Over and others (2018).