Calibration-Validation SchemeTwo types of model runs are used to calibrate and validate DSM2. For both Hydro and Qual, a realtide run, based on historic, observed data, may be used. In this mode, 15 minute or hourly historic stages and ECs are used for Hydro and Qual at the downstream boundary (Martinez), and hourly or daily historic flows and water quality values are used for rim locations, pumping, and internal sources and sinks such as agricultural diversions, drainages, and seepage from channels to islands. Gates are operated according to historic timing data. Typically realtide runs are several months or a year in length. A planning run is used primarily to calibrate and validate Qual. In this mode, a 19-year repeating mean tide is used as the Martinez stage boundary in Hydro. Monthly averaged flows and exports are used instead of daily flows. Gates are fixed in one position for each month. A tidefile is generated for each month, representing the average tidal flows for that month. A series of tidefiles are thus generated, and Qual run over the series. Planning runs are usually 15-25 years in length.
Calibration DescriptionThe most recent calibration, finished in July 1997, is a relatively small adjustment to the first calibration of DSM2, finished in June 1997 and documented in our 1997 Annual Progress Report. This recalibration was needed because of a generalization and more accurate treatment of irregular cross-sections within DSM2-Hydro. The changes made with respect to the previous calibration were to decrease Manning's N coefficients in the Suisun Bay and lower Sacramento River, and to increase the dispersion coefficients in the lower San Joaquin River. Both coefficients in those areas are now somewhat more realistic, and it illustrates indirectly once again the importance of having a realistic channel bathymetry. Additional information on the calibration and validation of the model can be found ont he section's world wide web home page: http://wwwdelmod.water.ca.gov.
DWR/Delta Modeling SectionDelta Simulation Model II - Calibration PlotsThis section provides detailed plots of the DSM2 Hydro and Qual calibration periods.
Delta Simulation Model II - Hydro CalibrationHydro: Stage amplitude, instantaneous flow, and residual flows are provided for comparison to observed data.
Stage Amplitudes We find it helpful to plot the daily maximum and minimum amplitudes of stages to see how well the downstream boundary tidal stage at Martinez propagates upstream into the estuary. Flows, and thus salinity, are quite sensitive to how the stage is propagated-an attenuated propagation will reduce tidal flow excursions and reduce ocean salt intrusion. Stages are easily attenuated by increasing channel friction (Manning's N coefficient). Flows Four different time periods were used to compare computed with observed flows: May 1988 (mostly tide cycle measurements), December 1992-January 1993, May 1994-July 1994, and October 1994. Different periods were required to cover a range of barrier operations and available data. May 1988-Instantaneous Measurements
North Delta
Jersey Point-Three Mile Slough
Central Delta Delta Simulation Model II - Qual CalibrationQual: In general the plots for this run are an improvement over the previous calibration, more so in the Western Delta where the exaggerated increased salinity in the Suisun Bay has been tempered to only a moderate increase. As a result, salinities in the lower Sacramento River are lower than observed, and a little lower even than the previous run, though we believe still acceptable.
Real-Tide EC The period January to August 1992 was chosen to calibrate DSM2 using real (observed) tides (stages) at Martinez, along with observed hourly salinity at Martinez. Daily rim hydrologies were used, as well as historic gate timings. However, we may have incorrect data or model operation with respect to the barriers in the Southern Delta (Old and Middle River); see the Union Island plot as an example.
Planning EC A Planning EC run differs from a real tide run in two significant respects: the downstream stage is a 19-year averaged stage, period 25 hours, in which the stage at t=26 hours is equal to the stage at t=1 hour; and rim flows and gate operations are monthly averages or approximations of historic data. Thus, a Planning Qual run is not intended to strictly simulate a historic condtion. Nevertheless it is still useful to compare such a run to historic data to see general trends over many years of different hydrologies. 1968-1991
1985-1991
Goto: Chapter 9 Goto: 1998 Annual Report Goto: Annual Reports |
