OHD/HL/DMIP - Model Instructions

A.   General Comments

Calibration will be carried out by comparing observed and simulated streamflow only at the designated basin outlet (indicated in brown in the description below) during the calibration period. Even though observed streamflow data may exist at some interior nested locations, modelers are asked to ignore these data in the calibration process. One emphasis of DMIP is to assess how well distributed models predict streamflow at interior locations, especially at ungaged sites.

Modelers will be asked to generate and submit to HL two basic types of simulations at specified points. The first type of simulation will be generated using initial or uncalibrated values of the hydrologic model parameters (and any hydraulic routing parameters). This test is to determine how well "physically based" models perform with parameters derived from physical data. Participants will submit their uncalibrated simulations from both the calibration and verification periods. The second type of simulation is generated after hydrologic and hydraulic model parameters are calibrated at the basin outlet. This simulation is meant to show how much calibration is required and what improvement in simulation accuracy is gained. Participants will submit their simulations (using calibrated parameters) for both the calibration and verification periods.

During the same model runs to generate the basin outlet hydrographs (with both outlet calibrated and uncalibrated parameters), participants will also simultaneously generate simulations at two types of interior points. The first type of interior point is where observed streamflow data are available. As stated above, there should be no calibration using these interior observed data. These "blind" simulations will be used to assess how well interior processes can be simulated when calibration was performed using only basin outlet data. Not all of the basins have observed interior streamflow data. The second type of point is an ungaged location along the main channel or a major tributary of the basin. These simulations will be analyzed by HL personnel to assess the variability of simulations from the various distributed models. Click here is see the DMIP basin maps showing the interior computational points. See section B below for a more precise description of the interior point locations.

Modelers will thus have to discretize their models in order to generate the interior hydrographs at the specified locations.

The Illinois River flowing through Arkansas and Oklahoma presents a good opportunity for participants to test their models on 3 different scales of basins, each having observed streamflow data at interior locations. The smallest basin, Baron Fork at Eldon, OK. has a drainage area of 307 sq. miles with an interior gage on Peacheater Creek at Christie, OK. Next to the Baron Fork basin is the Illinois River above the gage at Watts, OK. This basin has a drainage area of 635 sq. miles and includes the interior gage on the Illinois River at Savoy, Arkansas. The Illinois River at Watts, OK. is nested within the largest basin, the Illinois River above Tahlequah, OK, having a drainage area of 959 square miles. The Tahlequah basin also includes an interior gage on Flint Creek at Kansas, OK. Thus, the Illinois River above Tahlequah, OK. contains three interior gage locations.

The Elk River basin above the gage in Tiff City, Missouri, and the Blue River basin above the gage in Blue, Oklahoma, have no interior gage locations. These two basins represent additional cases for testing and comparison of distributed hydrologic models.

If participants chose to use the energy forcing data, they are directed to first use the data listed under 'Observed Energy Forcing Fields' for model simulations.

For those participants who can afford to do so, a) repeat and resubmit their simulations a second time using the NCEP forcing data, and b) characterize differences between the two forcing data sets and how these differences manifest themselves in the model simulations.

B.   Specific Instructions - Illinois River basins, Blue River basin, Elk River basin

  

 B.1.  Model Run Periods            

          1.  Calibration period:  May 1993 to May 31,  1999       

          2.  Verification period:   June 1, 1999 to July 31, 2000   

                         

 B.2. Time Step: Participants are requested to generate simulations in Z time 

       (GMT).  Simulations should have an hourly time step or have an ordinate 

        spacing that includes hourly values to facilitate comparison to the USGS 

        observed hourly discharge data. The USGS hourly discharge data is 

        supplied by HL in Z time to correspond to the time zone of the NEXRAD 

        precipitation data.

                    

  B.3. Illinois River Basin:  Baron Fork with basin outlet at USGS gage 07197000 

       at Eldon, Oklahoma. Drainage area 307 square miles.  USGS gage location: 

       Lat. 35d 55' 16" Lon. 94d 50' 18", on downstream left abutment of bridge 

       on State Highway 51, 0.4 miles southeast of Eldon.         

       a. Generate 2 simulations at the basin outlet that span both the calibration 

       and validation periods:        

          1.  with uncalibrated/initial parameters          

          2.  with calibrated parameters 

       b. While generating the 2 basin outlet simulations, compute interior

                 simulations at:

          1. Peacheater Creek at USGS gage 07196973 at Christie, Ok.  drainage         

             area 25 square miles. Gage:  Lat. 35d 57' 17" Lon. 94d 41' 46", 

             0.4 miles upstream of junction with Baron Fork.   No calibration

             is allowed using observed streamflow data at this point. 

             It is to be a "blind" simulation.

                

          2.  Ungaged location on channel at Lat. 35d 54' 38",   Lon. 94d 32'   

              16", drainage area 58.4 square miles [Note: before 2/23/2002, 

              the area estimate given on this site was 80.66 square miles.  

              The 80.66 estimate was derived using a 400-m resolution DEM (See DMIP DEM data page.), 

              but there is a big discrepancy between this area

              and the area derived from the 30-m DEM (58.4 square miles).  We believe the area

              derived from the 30-m DEM is more accurate.]   

                                     

 B.4.  Illinois River Basin:  Illinois River with basin outlet at USGS gage

       07195500 at Watts, Oklahoma. Drainage area 635 square miles. 

       USGS gage location: Lat 36d 07' 48" Lon 94d 34' 19", on downstream side of pier

       of bridge on Highway 59, 1.5 miles north of Watts.           

       a.  Generate 2 simulations at the basin outlet that span both the calibration and 

           validation periods:        

           1.  with uncalibrated/initial parameters            

           2.  with calibrated parameters        

       b.  While generating the 2 basin outlet simulations, compute interior

           simulations at:             

           1.  Illinois River  at USGS gage 07194800 at Savoy, Arkansas.  Drainage area 

               167 square miles. Gage: Lat. 36d 06'  11" Lon. 94d 20' 39"   

               No calibration is allowed using observed streamflow data at this point.  

               It is to be a "blind" simulation.             

           2.  Ungaged location on channel at Lat. 36d 2' 53", Lon. 94d 19' 16", 

               drainage area 76.49 square miles. 

                                        

                 

  B.5.  Illinois River Basin: Illinois River with basin outlet at USGS gage 

        07196500 at Tahlequah, Oklahoma.  Drainage area: 959 square miles. Gage 

        location: Lat. 35d 55' 22", Lon. 94d 55' 24", 0.2 miles downstream from 

        U.S. Highway 62, 2.2 miles northeast of Tahlequah.

        a.  Generate 2 simulations at the basin outlet that span both the calibration and 

            validation periods:        

            1.  with uncalibrated/initial parameters            

            2.  with calibrated parameters  

        b.  While generating the 2 basin outlet simulations, compute interior 

            simulations at: 

            1.  Illinois River at USGS gage in Watts, OK

            2.  Illinois River at USGS gage in Savoy, Arkansas.

            3.  Flint Creek at USGS gage 07196000 in Kansas, OK.  Drainage area 

                110 square miles.  Gage location:  Lat. 36d 11' 11", 

                Lon. 94d 42' 24" upstream from bridge on U.S. Highway 412.

        (Note: no specific calibration using observed streamflow at  these points, 

        even though calibration was performed for the Illinois River at Watts, OK 

        for the runs in item  B.4 above.)

   

  B.6.  Elk River with basin outlet at USGS gage 07189000 in Tiff City, Missouri. 

        Drainage area 872 square miles.  Gage location: Lat. 36d 37' 53" Lon. 94d 

        35' 12", on bridge on State Highway 43, 3.0 miles southeast of Tiff City.

        a.  Generate 2 simulations at the basin outlet that span both the calibration and 

            validation periods:        

            1.  with uncalibrated/initial parameters            

            2.  with calibrated parameters

        b.  While generating the 2 basin outlet simulations, compute interior 

            simulations at ungaged location on channel at Lat. 36d 35' 38", 

            Lon. 94d 9' 17", drainage area 122.93 square miles. 

        (Note: there is no interior observed streamflow for this basin). 

                  

  B.7.  Blue River with designated basin outlet at USGS gage 07332500 in Blue, 

        Oklahoma.  Drainage area 476 square miles. Gage location: Lat. 33d 59' 49" 

        Lon. 96o 14' 27', on bridge on U.S. Highway 70, 1.0 mile west of Blue, 

        Oklahoma.    

        a.  Generate 2 simulations at the basin outlet that span both the calibration and 

            validation periods:        

           1.  with uncalibrated/initial parameters            

           2.  with calibrated parameters

        b.  While generating the 2 basin outlet simulations, compute interior 

            hydrographs at 

            1  Ungaged location on main channel of the Blue River at Lat. 34d 

               30' 24", Lon. 96d 40' 30",  drainage area 59.15 square miles.

            2  Ungaged location on main channel of the Blue River at Lat. 34d 

               26' 39", Lon. 96d 37' 30", drainage are 116.87 square miles. 

        (note: there is no interior observed streamflow for these two points)