Chuck Spalding has reported a bug in the representation of streams in the model. It affects the model for those stress periods after reservoirs go on line. At each of the reservoir locations, cells within the reservoir reaches that are upstream of the reservoir pool are omitted. In most cases these amount to 1 or 2 cells, however, in the case of Harry Strunk Reservoir 12 cells are omitted representing a length of 7 or 8 miles. Please take a look at the MODFLOW STREAM Package 12.str. A PDF is attached showing 2 figures. The first figure is the stream package under steady-state conditions and the second figure is the stream package in December 2000, representing post-reservoir conditions.
The aforementioned PDF file shows the streams before and after the reservoirs as reproduced here:
The same information can be displayed in an alternative format as shown here with a light blue color used to represent stream cells that are active before and after the reservoirs come on line, red cells represent stream cells that are active only before the reservoirs come on line, and yellow cells represent stream cells that are active only after the reservoirs come on line.
The same information is displayed in the image below zoomed in on Harry Strunk reservoir on the right and Hugh Butler on the left.
Note that in both cases there are cells that appear only before or only after the reservoirs come on line. The reason why this occurs is best illustrated by the schematic diagram of the middle portion of the stream network shown below.
The stream network was adopted from the original USGS work by Landon et. al. The system topology, and in particular the Segment and Reach numbering was adopted verbatim. On Red Willow creek, for example, the sequence Segment 60, Segment 132, Segment 159, Segment 160 was adopted, with Segment 132 representing Hugh Butler Reservoir. The topology of Medicine Creek is more complex as can be seen above. It is, however, of interest to note that Segment 170 represents Medicine Creek above Harry Strunk, Segment 183 represents Harry Strunk Reservoir, and Segment 184 represents Medicine Creek below Harry Strunk Reservoir. The division between Segments 170 and 183 occurs at the Medicine Creek Above Harry Strunk gage.
In order to simplify the post-processing, streams and reservoirs are represented using the MODFLOW STREAM package. This permits the impacts to be captured by a small number of accounting points in the HYDMOD package which can then be post-processed for accounting purposes. Unfortunately, there is a limitation in the HYDMOD package that requires the number of segments and reaches per segment to remain the same throughout the simulation. This creates a particular problem in the case of the reservoirs. Reservoirs are simulated as they were built. Therefore, prior to a reservoir coming on line, the stream is simulated crossing the reservoir. When the reservoir starts to store water, the reservoir stage is set based on historical stage measurements, while the reservoir area is represented using a number of cells. A different number of cells is typically required to simulate the pre-reservoir stream and the reservoir. In order to accommodate the HYDMOD package limitation, padding stream reaches were used. In the padding reaches, the stream-aquifer conductance term was set to zero. This allows water in the stream to be routed through these stream reaches, but have no interaction with the groundwater system.
Reservoirs were simulated by setting an inflow into the segment containing the reservoir of 1,000,000 cfs. This flow rate was chosen to readily distinguish reservoir segments from stream segments, and to allow for any amount of reservoir leakage that could reasonably occur. The reservoir leakage is then calculated by subtracting 1,000,000 from the stream flow at the end of the reservoir segment. Note, however, that since stream flows can only be specified at the beginning of a segment, this requires reservoirs to occupy an entire segment.
In the case of Harry Strunk reservoir, Segment 183 is used to represent the reservoir. However, as noted in the Nebraska email, Segment 183 actually starts at the Medicine Creek Above Harry Strunk gage which is several miles above the reservoir proper. Since the reservoir cells were selected to only cover the reservoir proper, there is a resulting gap in the Medicine Creek stream cells between the gage and the reservoir.
The accounting calculations for Medicine Creek adds the impacts calculated above Harry Strunk reservoir to the impacts at the confluence with the main stem. The above Harry Strunk reservoir accounting point was Segment 170 Reach 5 which was the last reach above the reservoir represented. However, since Segment 170 new is extended by 12 reaches, a new accounting point is needed at Segment 170 Reach 17. This accounting point was named AcctHarryStrunk.
A new simulation was generated to evaluate the impacts of the above changes. The simulation was named V12s using the previously established naming scheme.
The program fixstr was written to modify the input file to mkstr as well as produce new HYDMOD and accounting package input files. The fixstr program takes the 11.str.0 input file used to define the pre-reservoir stream network and produces 12.str.0. The difference between 11.str.0 and 12.str.0 is that the first 12 reaches of Segment 183 are appended to Segment 170 as described above. The mkstr program is then modified to read 12.str.0 instead of 11.str.0, and to produce 12s.str instead of 12.str. Furthermore, the fixstr program creates 12s.hyd from 12.hyd by adding the AcctHarryStrunk location and modifying the HarryStrunk location from Segment 183 Reach 19 to Segment 183 Reach 8 to account for the fact that Harry Strunk Dam is now at a renumbered reach. Finally, the fixstr program modifies the accounting parameter file acct.par to produce acct.12s which replaces SI1700056841000 with SI170017AcctHarryStrunk.
The MODFLOW V12s simulations was then be repeated with the V12p7 input files except that 12.str is replaced by 12s.str and 12.hyd is replaced by 12s.hyd. Output files and input files that were changed are provided.
There are some changes in the heads in the immediate vicinity of the twelve stream reaches that are appended to Segment 170. The differences between the heads predicted by V12p7 and V12s are shown for the steady state, 12/31/1940, 12/31/1970 and 12/31/2000. The head differences are the result of the heads being pulled up between V12p7 and V12s when the stream is loosing, and the heads being pulled down when the stream is gaining. Prior to September 1949 when the reservoir comes on line, there is, of course, no difference between V12p7 and V12s. At the end of the simulation, the maximum increase in head from V12p7 to V12s is 14.5 feet, and the maximum decline in head is 25.1 feet. In 8 cells the decline exceeded 10 feet, in 66 cells the decline exceeded 1 foot and in 197 cells the decline exceeded 0.1 feet. In 1 cell, the increase exceeded 10 feet, In 6 cells, the increase exceeded 1 foot, and 10 cells the increase exceeded 0.1 feet. Therefore, a head change of greater than 10 feet was observed in 9 cells, a change of 1 foot was observed in 72 cells and a change of greater than 0.1 feet was observed in 207 cells. The maximum distance a 0.1 foot head change was observed from Medicine creek was approximately 10 miles.
As is to be expected, the predicted impacts for Colorado and Kansas remain unchanged between V12p7 and V12s. The impacts for Nebraska remain unchanged except on Medicine Creek and Harry Strunk reservoir as summarized below:
|1981-2000 Average Annual Impact (acre-feet)|
|Nebraska Pumping V12p7||11999||360||165073|
|Nebraska Pumping V12s||13151||195||166060|
|Nebraska Pumping Change||+1152||-165||+987|
|Nebraska Mound V12p7||7998||0||16328|
|Nebraska Mound V12s||7976||0||16310|
|Nebraska Mound Change||-22||0||-18|
|Nebraska Impact Net Change||+1174||-165||+1005|