Generalized least squares (GLS) techniques were used with at-site statistics, from PeakFQ, and drainage area for streamgages in Puerto Rico to calculate regional regression equations for ungaged locations. The GLS approach assigns weights based on uncertainty of the observations (ie. record length and variance) and accounts for correlated streamflows and time-sampling errors (Farmer and others, 2019). Two separate regions were defined for the use of these equations to minimize residuals, and performance metrics from WREG were analyzed to determine drainage area as the only explanatory variable provided the best results. Values for the correlation smoothing function in region 1 were alpha equal to 0.006 and theta of 0.96 and for region 2 alpha equal to 0.004 and theta of 0.94. This data release includes (1) a PeakFQ folder that contains input files, both .EXP and .PRT (output from PeakFQ), for use in WREG; (2) a SiteInfo folder for use in WREG that contains information about each site used, separated by region; and (3) a Results folder showing the WREG output of each annual exceedance probability flow, separated by regression region where applicable. References Cited: Farmer, W.H., Kiang, J.E., Feaster, T.D., and Eng, K., 2019, Regionalization of surface-water statistics using multiple linear regression: U.S. Geological Survey Techniques and Methods, book 4, chap. A12, 40 p., Accessed August 4, 2020 at https://doi.org/10.3133/tm4A12.

A generalized least squares (GLS) technique was used with at-site n-day low-flow time series and frequencies, from SWToolbox (Kiang and others, 2018), and 14 basin characteristics (basin perimeter, drainage area, maximum basin elevation, mean total annual reference evapotranspiration, groundwater head, length of longest flow path, minimum basin elevation, runoff-curve number, relief, ruggedness, slope ratio, percentage of hydrologic soil type B, available water capacity, and total length of streams in a basin) for 28 selected streamflow gaging stations in Puerto Rico to calculate regional n-day low-flow frequency regression equations for ungaged stream locations. The GLS approach assigns weights based on uncertainty of the observations (ie. record length and variance) and accounts for correlated streamflows and time-sampling errors (Eng and others, 2009). Performance metrics from the Weighted-multiple-linear REGression Program (WREG) were analyzed to determine drainage area, mean total annual reference evapotranspiration, and minimum basin elevation as the explanatory variables that provided the best results. Values for the correlation smoothing function in were alpha equal to 0.003 and theta equal to 0.96. This data release includes a SWToolbox_output_files folder that contains supporting data used to format the WREG input files, xday_lowflow folders (where x refers to 1, 7, 14, or 30) that contain (1) the WREG input files FlowChar, LP3G, LP3K, LP3s, SiteInfo, and the 28 station nday timeseries files (USGSxxxxxxxx); and (2) a WREG_gls_results folder that contains output for the 2yr, 5yr, and 10yr nonexceedance probability flows. References Cited: Eng, K., Chen, Y., and Kiang, J.E., 2009, User’s guide to the weighted-multiple-linear-regression program (WREG version 1.0): U.S. Geological Survey Techniques and Methods, book 4, chap. A8, 21 p., https://pubs.usgs.gov/tm/tm4a8. Kiang, J.E., Flynn, K.M., Zhai, T., Hummel, P., and Granato, G., 2018, SWToolbox—A surface-water tool-box for statistical analysis of streamflow time series: U.S. Geological Survey Techniques and Methods, book 4, chap. A–11, 33 p., https://doi.org/10.3133/tm4A11.

A generalized least squares (GLS) technique was used with at-site n-day low-flow time series and frequencies, from SWToolbox (Kiang and others, 2018), and 14 basin characteristics (basin perimeter, drainage area, maximum basin elevation, mean total annual reference evapotranspiration, groundwater head, length of longest flow path, minimum basin elevation, runoff-curve number, relief, ruggedness, slope ratio, percentage of hydrologic soil type B, available water capacity, and total length of streams in a basin) for 28 selected streamflow gaging stations in Puerto Rico to calculate regional n-day low-flow frequency regression equations for ungaged stream locations. The GLS approach assigns weights based on uncertainty of the observations (ie. record length and variance) and accounts for correlated streamflows and time-sampling errors (Eng and others, 2009). Performance metrics from the Weighted-multiple-linear REGression Program (WREG) were analyzed to determine drainage area, mean total annual reference evapotranspiration, and minimum basin elevation as the explanatory variables that provided the best results. Values for the correlation smoothing function in were alpha equal to 0.003 and theta equal to 0.96. This data release includes a SWToolbox_output_files folder that contains supporting data used to format the WREG input files, xday_lowflow folders (where x refers to 1, 7, 14, or 30) that contain (1) the WREG input files FlowChar, LP3G, LP3K, LP3s, SiteInfo, and the 28 station nday timeseries files (USGSxxxxxxxx); and (2) a WREG_gls_results folder that contains output for the 2yr, 5yr, and 10yr nonexceedance probability flows. References Cited: Eng, K., Chen, Y., and Kiang, J.E., 2009, User’s guide to the weighted-multiple-linear-regression program (WREG version 1.0): U.S. Geological Survey Techniques and Methods, book 4, chap. A8, 21 p., https://pubs.usgs.gov/tm/tm4a8. Kiang, J.E., Flynn, K.M., Zhai, T., Hummel, P., and Granato, G., 2018, SWToolbox—A surface-water tool-box for statistical analysis of streamflow time series: U.S. Geological Survey Techniques and Methods, book 4, chap. A–11, 33 p., https://doi.org/10.3133/tm4A11.

An ordinary least squares (OLS) technique was used with at-site flow-duration exceedance probabilities, from SWToolbox (Kiang and others, 2018), and 14 basin characteristics (basin perimeter, drainage area, maximum basin elevation, mean total annual reference evapotranspiration, groundwater head, length of longest flow path, minimum basin elevation, runoff-curve number, relief, ruggedness, slope ratio, percentage of hydrologic soil type B, available water capacity, and total length of streams in a basin) for 28 selected streamflow gaging stations in Puerto Rico to calculate regional flow-duration regression equations for ungaged stream locations. The final flow-duration regression equations were developed in R (R Core Team, 2019) using the “lm” function. Performance metrics were analyzed to determine drainage area, mean total annual reference evapotranspiration, and minimum basin elevation as the explanatory variables that provided the best results. This data release includes a SWToolbox_output_files folder that contains supporting data used to format the R input file, a FlowDuration_regression_models folder that contains .txt files showing the final regression models for the 99th, 98th, 95th, 90th, 80th, 70th, 60th, and 50th percent exceedance probabilities, and a FlowDuration_regression_model_plots folder that contains .pdf files showing supporting graphics output for the final regression models. References Cited: Kiang, J.E., Flynn, K.M., Zhai, T., Hummel, P., and Granato, G., 2018, SWToolbox—A surface-water tool-box for statistical analysis of streamflow time series: U.S. Geological Survey Techniques and Methods, book 4, chap. A–11, 33 p., https://doi.org/10.3133/tm4A11. R Core Team, 2019, R—A language and environment for statistical computing: R Foundation for Statistical Computing, Vienna, Austria, accessed February 11, 2021, at https://www.R-project.org/.

An ordinary least squares (OLS) technique was used with at-site flow-duration exceedance probabilities, from SWToolbox (Kiang and others, 2018), and 14 basin characteristics (basin perimeter, drainage area, maximum basin elevation, mean total annual reference evapotranspiration, groundwater head, length of longest flow path, minimum basin elevation, runoff-curve number, relief, ruggedness, slope ratio, percentage of hydrologic soil type B, available water capacity, and total length of streams in a basin) for 28 selected streamflow gaging stations in Puerto Rico to calculate regional flow-duration regression equations for ungaged stream locations. The final flow-duration regression equations were developed in R (R Core Team, 2019) using the “lm” function. Performance metrics were analyzed to determine drainage area, mean total annual reference evapotranspiration, and minimum basin elevation as the explanatory variables that provided the best results. This data release includes a SWToolbox_output_files folder that contains supporting data used to format the R input file, a FlowDuration_regression_models folder that contains .txt files showing the final regression models for the 99th, 98th, 95th, 90th, 80th, 70th, 60th, and 50th percent exceedance probabilities, and a FlowDuration_regression_model_plots folder that contains .pdf files showing supporting graphics output for the final regression models. References Cited: Kiang, J.E., Flynn, K.M., Zhai, T., Hummel, P., and Granato, G., 2018, SWToolbox—A surface-water tool-box for statistical analysis of streamflow time series: U.S. Geological Survey Techniques and Methods, book 4, chap. A–11, 33 p., https://doi.org/10.3133/tm4A11. R Core Team, 2019, R—A language and environment for statistical computing: R Foundation for Statistical Computing, Vienna, Austria, accessed February 11, 2021, at https://www.R-project.org/.

Flow-duration statistics at the 99th, 98th, 95th, 90th, 80th, 70th, 60th, and 50th percent exceedance probabilities and annual n-day low-flow statistics for the 1-, 7-, 14-, and 30-day mean low flows with 2-year (0.5 nonexceedance probability), 5-year (0.2 nonexceedance probability), and 10-year (0.1 nonexceedance probability) recurrence intervals were computed for 28 selected streamflow gaging stations in Puerto Rico. The 28 selected streamflow gaging stations were required to have 10 or more years of daily mean streamflow data through water year 2018. The flow-duration statistics and n-day low-flow frequencies were computed using the U.S. Geological Survey program, SWToolbox. Regional regression equations were developed to estimate flow-duration statistics and n-day low-flow frequencies at ungaged stream locations using selected basin characteristics as explanatory variables. These variables were determined from digital spatial datasets and geographic information systems using the most recent data available, as referenced in the U.S. Geological Survey web application, StreamStats, and published in Kolb and Ryan (2021). An ordinary least-squares procedure in R Studio was used to develop the final regional flow-duration regression equations using drainage area, mean total annual reference evapotranspiration, and minimum basin elevation as the explanatory variables. A generalized least squares procedure in the U.S. Geological Survey program, WREG, was used to account for cross-correlation of sites and develop the final regional n-day low-flow frequency regression equations using drainage area, mean total annual reference evapotranspiration, and minimum basin elevation as the explanatory variables. This data release includes two child pages: Puerto Rico Flow-Duration Regression Files and Puerto Rico N-day Low-Flow Regression Files, a BasinCharacteristics.csv file that contains 47 basin and climatic characteristics considered in the analyses, a BasinCharacteristics_corrrelation_charts folder that contains .pdf files showing correlation matrices, a R_regsubsets_output folder that contains .txt and .pdf files showing results of the "regsubsets" analyses, a Trend_statistics_nday_lowflow_timeseries.csv file that contains the SWToolbox Mann-Kendall tau statistics, and a NWIS_rdb_files folder that contains the .rdb files used in the analyses. References Cited: Kolb, K.R., and Ryan, P.J., 2021, Basin Characteristic Rasters for Puerto Rico StreamStats, 2021: U.S. Geological Survey data release, https://doi.org/10.5066/P9HK9SSQ.

Flow-duration statistics at the 99th, 98th, 95th, 90th, 80th, 70th, 60th, and 50th percent exceedance probabilities and annual n-day low-flow statistics for the 1-, 7-, 14-, and 30-day mean low flows with 2-year (0.5 nonexceedance probability), 5-year (0.2 nonexceedance probability), and 10-year (0.1 nonexceedance probability) recurrence intervals were computed for 28 selected streamflow gaging stations in Puerto Rico. The 28 selected streamflow gaging stations were required to have 10 or more years of daily mean streamflow data through water year 2018. The flow-duration statistics and n-day low-flow frequencies were computed using the U.S. Geological Survey program, SWToolbox. Regional regression equations were developed to estimate flow-duration statistics and n-day low-flow frequencies at ungaged stream locations using selected basin characteristics as explanatory variables. These variables were determined from digital spatial datasets and geographic information systems using the most recent data available, as referenced in the U.S. Geological Survey web application, StreamStats, and published in Kolb and Ryan (2021). An ordinary least-squares procedure in R Studio was used to develop the final regional flow-duration regression equations using drainage area, mean total annual reference evapotranspiration, and minimum basin elevation as the explanatory variables. A generalized least squares procedure in the U.S. Geological Survey program, WREG, was used to account for cross-correlation of sites and develop the final regional n-day low-flow frequency regression equations using drainage area, mean total annual reference evapotranspiration, and minimum basin elevation as the explanatory variables. This data release includes two child pages: Puerto Rico Flow-Duration Regression Files and Puerto Rico N-day Low-Flow Regression Files, a BasinCharacteristics.csv file that contains 47 basin and climatic characteristics considered in the analyses, a BasinCharacteristics_corrrelation_charts folder that contains .pdf files showing correlation matrices, a R_regsubsets_output folder that contains .txt and .pdf files showing results of the "regsubsets" analyses, a Trend_statistics_nday_lowflow_timeseries.csv file that contains the SWToolbox Mann-Kendall tau statistics, and a NWIS_rdb_files folder that contains the .rdb files used in the analyses. References Cited: Kolb, K.R., and Ryan, P.J., 2021, Basin Characteristic Rasters for Puerto Rico StreamStats, 2021: U.S. Geological Survey data release, https://doi.org/10.5066/P9HK9SSQ.

Regional regression equations were calculated in Puerto Rico with generalized least squares techniques to estimate flood frequency statistics at ungaged locations using drainage area as the only explanatory variable. The island was divided into 2 regions to minimize residuals. The region division that resulted in lower and more balanced residuals runs primarily north-south near the center of the island, mostly along an 8-digit hydrologic unit code (HUC8) boundary. The division line runs through a HUC8 polygon on the southern end of the island, but care was taken to include entire watersheds and consideration was given where hydrologic and physiographic properties differed. This data release includes geographic information system files that define the polygons for both regions.

Regional regression equations were calculated in Puerto Rico with generalized least squares techniques to estimate flood frequency statistics at ungaged locations using drainage area as the only explanatory variable. The island was divided into 2 regions to minimize residuals. The region division that resulted in lower and more balanced residuals runs primarily north-south near the center of the island, mostly along an 8-digit hydrologic unit code (HUC8) boundary. The division line runs through a HUC8 polygon on the southern end of the island, but care was taken to include entire watersheds and consideration was given where hydrologic and physiographic properties differed. This data release includes geographic information system files that define the polygons for both regions.

Estimates of the magnitude of peak-flows were updated for the 50-, 20-, 10-, 4-, 2-, 1-, 0.5-, and 0.2-percent chance exceedance levels for 91 rural, unregulated streamgaging stations on the main island of Puerto Rico. These stations required 10 or more years of annual peak-flow record, using data to 2017, for inclusion in the study. The magnitude and frequency of floods at selected streamgages in Puerto Rico were estimated using the U.S. Geological Survey PeakFQ program and updated methods outlined in Bulletin 17C (England and others, 2018). Regional regression equations were calculated to estimate flood frequency statistics at ungaged locations using selected basin characteristics as explanatory variables. These variables were determined from digital spatial datasets and geographic information systems using the most recent data available, as referenced in the U.S. Geological Survey web application, StreamStats, and published in Kolb and Ryan (2021). A generalized least squares procedure in the U.S. Geological Survey program, WREG, was used to account for cross-correlation of sites and develop the final regional regression equations using drainage area as the only explanatory variable. Two separate regions were defined for regression equation use in this study to minimize residuals. NOTE: All of the data in the previous version can be found in version 1.1. References Cited: England J.F., Jr., Cohn, T.A., Faber, B.A., Stedinger,J.R., Thomas,W.O.,Jr., Veilleux,A.G., Kiang,J.E., and Mason,R.R., Jr., 2018, Guidelines for determining flood flow frequency —Bulletin 17C: U.S. Geological Survey Techniques and Methods, book 4, chap. B5, 148p., https://doi.org/10.3133/tm4B5. Kolb, K.R., and Ryan, P.J., 2021, Basin Characteristic Rasters for Puerto Rico StreamStats, 2021: U.S. Geological Survey data release, https://doi.org/10.5066/P9HK9SSQ.