Appendix 1—Data processing and computation to characterize hydrology and compare water quality of four watersheds in Puerto Rico
This appendix summarizes the procedures used to characterize the hydrology and to compare the water quality among the watersheds studied as part of the U.S. Geological Survey's (USGS) Water, Energy, and Biogeochemical Budgets (WEBB) research program in Puerto Rico and parallel work in Panama. Data from the first fifteen years (1991-2005) of the Puerto Rico component have been processed for this report, an undertaking that entailed working with several million discharge measurements, thousands of measurements from automated rain gages, and several thousand samples analyzed for some aspect of water quality, ranging from suspended sediment and electrical conductivity to comprehensive chemical analyses. Discharge records from stream gages and the water-quality data sets were processed separately to check for errors and to fill in discharge-data gaps with estimated values. The hydrologic and water-quality data sets were then merged into two types of master files. One contains all the primary hydrology data and interpretive information about hydrologic processes, such as calculating trends, tagging recessions and exceptional storms, with one hydrologic-data file for each gage site. The other file type holds primary water-quality data, chemical parameters derived from the primary data, and an abbreviated hydrological characterization of each sample, with samples from all sites and sources in a single water-quality data file. This water-quality data file, with about 6,600 records for 15 years of data, can be manipulated in conventional spreadsheets. This appendix is divided into seven sections. The first discusses the computational approach used to handle this large data set. The second covers hydrological data processing. The third describes the assembly of the water-quality data, including calculated parameters such as saturation indices, cyclic-salt correction, and dissolved bedrock. The fourth describes the solute- and solid-load computations built around the program LOADEST (Runkel and others, 2004). The fifth details how runoff and yield percentile ranges are calculated. The sixth presents summaries of the data set. The last compares this data set with other large data sets assembled by the USGS, in terms of the range of process rates measured.