Correcting Problems with Existing Codes

7.1.1.1 Original vs. New Precipitation Raster File

At the end of his research, Emerick discovered that the watershed boundary for Little Pine Creek generated by the WATERSHED program was inaccurate because it did not include the entire watershed area as described in the published data. He recommended that future investigators use the watershed boundary produced by the program Geo-HMS instead. During the course of the present research, we were unable to reproduce this problem. The Little Pine Creek watershed boundary generated by the WATERSHED program provides a perfectly accurate description of the true watershed area.

Since this problem with the watershed boundary was discovered at the end of the previous research, all the precipitation raster files from during the last study were developed from a precipitation grid file that was created using an inaccurate watershed outline. Figures 7.1 and 7.2 show that the greatest difference between the areas described by the correct and incorrect watershed boundaries occurs in the northern most reaches of the watershed, where the old file underestimates the actual watershed area. This difference in this omitted area is small when compared to that of the entire watershed, so one would expect that any differences between the model’s results obtained using either file would be small as well. This might be true under most conditions, but not for data generated using the August 17, 2002 precipitation event. Figure 6.12 shows that the greatest precipitation depths during the entire event occur during the first hour over the northernmost reaches of the watershed. Since the old precipitation grid file does not include some of this area, the results generated using the old and new watershed outlines should differ noticeably.

Figure 7.1: Previous watershed outline Figure 7.2: Corrected watershed outline

Figure 7.3 provides a comparison of the runoff hydrographs at the watershed outlet generated using precipitation grid raster files created from the previous and new watershed boundaries. Table 7.1 compares the peak flow (Qp) and time to peak (Tp) values for each case, and shows that the resulting values generated using the new file are 14% and 5.5% greater, respectively, than the originals. In light of this discovery, all results from our present research were generated using the correct delineation of the Little Pine Creek watershed area, as defined by the WATERSHED program.

Table 7.1: Hydrograph Qp and Tp values using 2 different Little Pine Creek watershed outlines

Case Qp (cfs) Tp (min)

Original precipitation data 154.36 255 New precipitation data 179.88 270 % Difference between data 14.19% 5.56%

0 20 40 60 80 100 120 140 160 180 200 0 100 200 300 400 500 600 700 800 Time (min) Q (cfs)

orig precip data new precip data

Figure 7.3: Hydrographs resulting from 2 different outlines of Little Pine Creek watershed

7.1.1.2 Improvements to the Runoff Code

While exploring the effect of subdividing the watershed on the hydrograph at the outlet, the code RUNOFF began behaving inconsistently. Cases run using the same number of sub-watershed divisions yielded substantially different hydrograph Qp results depending on how the sub- watershed divisions were arranged. Further investigation into this matter revealed that the runoff code would generate different runoff volumes for the same area of the watershed, receiving the same depth of precipitation input depending on how it was subdivided. As it turns out, the source of this discrepancy is the way in which the original runoff code computes the 15-minute precipitation time increments required by the Muskingum routing module by averaging the 5- minute incremented time-series used to perform calculations in the code. In the original scheme, the code uses only the first and last of the 5-minute incremental Q values within any 15-minute period to compute the average, neglecting the weights of the second and third values. Averaging in this way, without considering the weights of all four 5-minute incremental values that fall within a particular 15-minute time interval will produce different total runoff values for different sub-watershed arrangements, because each considers different 5-minute incremental Q values

when creating the averages. The updated version of the runoff code, RUNOFF_MTS, fixes this problem by using all four 5-minute incremental Q values (t = 0, 5, 10 and 15) to compute any one 15-minute average. Table 7.2 compares the total runoff generated using the original and new runoff codes for an area in the northwest portion of the watershed shown in Figure 7.4, assuming two and four sub-watershed divisions. The total runoff volume predicted by the new code for this area is essentially the same for both cases, while the volume predicted by the original code differs substantially between the two cases.

The new RUNOFF_MTS code operates with a greater degree of independence from the user as well. The original design requires that the user run the code for each individual sub- watershed to generate runoff. The new design offers the user the option of computing runoff for all sub-watershed areas at once, to save time, or individually.

Table 7.2: Comparison of runoff values generated by code Sbws No. sbws-15 sbws-16 sbws-15 sbws-16 sbws-17 sbws-18 sbws-15 sbws-16 sbws-15 sbws-16 sbws-17 sbws-18 Runoff Vol (ft3) Total Runoff Vol (ft3) Area 1 Area 2 274810 153644 96131 103831 139031 60670 139294 58207 428454 399663 415598 415168

Original RUNOFF code New RUNOFF_MTS code

Area 1 Area 2

244513 171086 104851 112815