Laserfiche WebLink
of monthly flow characterizations for each station (filenname, charact.xls). The step-by-step <br />approach is given in detail as follows: <br />1. Downloading data •fles -Individual data files for each gaging station are extracted from the <br />CRDSS using tstool. The desired file format should contain no summary information. The <br />suggested file naming scheme is the station's 8-digit ID number, followed by the extension <br />*.tsl (e.g. 09034500.ts1). <br />2. Filling missing data -Missing data for the period 1909-1996 in the *.tsl files is filled using <br />the MSM. A Windows interface to the MSM program was developed by Ayres Associates to <br />handle different formatting needs and make the program more user-friendly. The interface <br />prompts the user for up to 200 *.tsl files to be filled, and allows for the selection of the <br />different data filling options provided in the MSM. The interface creates an input file for the <br />MSM, called infile.dat, and runs the program. The MSM outputs two files, a detailed output <br />file called outfile.dat, and a summary information file called sum.dat. The output file <br />contains filled data for all of the stations (filled values are followed by an "e" for estimated), <br />and the summary file contains details about the data filling process, including the stations <br />used for the regression, goodness of fit, and computed statistics for the original and filled <br />data sets. ~~ ~~~ ~-'~ `' <br />Several options are provided in the MSM. For more information about these options, refer to the <br />subtask 11.10 Memorandum. The options used in this portion of Task 11 were: (1) simple linear <br />regression, (2) cyclic ornon-cyclic depending on Standard Error of Prediction (SEP); (3) search <br />all stations, (4) all stations dependent, (5) extra output table, (6) 1909 as the beginning year of <br />the extended record, and (7) 5 as the minimum number of concurrent values required for the <br />extension equation. Future modifications to the MSM and the interface are planned for Taslc <br />11.10, but will not affect any of these options. <br />When the SEP option is chosen in the MSM, missing data is filled through regression against the <br />gage that provides the least SEP, whether cyclic or non-cyclic, within the same basin or outside. <br />Therefore, although it would seem more likely for the independent and dependent gages to reside <br />in the same basin, it is not a constraint. Crossing basin boundaries with the regression equations <br />is an issue in some of the earlier years of record. For part of 1909 and all of 1910, gaps exist in <br />all of the data sets, except those in the Colorado Basin. Tlus means that regression is forced to <br />occur bet`veen gages residing in different basins, some that are not contiguous. Although these <br />regressions may not be as good as others, they still pass the 95% confidence interval test required <br />for filling by the MSM. All of the seventeen gages were filled by the MSM using the options <br />noted, and were output individually as *.fil files. <br />Characterizing the data -Monthly streamflow data is characterized one station at a time <br />using the SCT. The SCT workbook contains several spreadsheets and two macros. It <br />performs data analysis on the original ~`.tsl files, and characterization on the *.fil files. Data <br />analysis consists of calculating normal distribution parameters (the mean and standard <br />deviation), and computing 25% and 75%quantiles. Characterizations are completed using <br />the calculated quantiles as cutoff values for wet, average, and dry years. A readme file in the <br />workbook contains instructions for spreadsheet/macro use. Input requirements are: user <br />name; 8-digit station ID; paths, directories, and extensions for the tstool and filled data files; <br />Appendix E E-71 <br />