CCWD: Using DSM2 Historical July20, Two Applications of the DSM2 Historical Model: Modeling Contaminant Spills and Using Salinity Fingerprints to Improve DSM2 Real-Time Forecasts Tom Rose and Marianne Guerin
CCWD: Using DSM2 Historical July20, Background CCWD uses Delta Water WQ Implications –Salinity –DOC Supply Issues –Direct delivery –Blending –Emergency storage Forecasting WQ at intakes to help Operations vital –G-Model – zero-dimensional –DSM2 – accuracy variable at intakes – time-consuming
CCWD: Using DSM2 Historical July20, Talk Logistics – Using the Historical Model Modeling Contaminant Spills –Motivation for the work –Methodology –Results –Future direction Improving Forecasts –Background –Objective –Method –Findings –Future work – suggestions? Background –CCWD Water Supply and Quality Issues –Using DSM2 Historical Model
CCWD: Using DSM2 Historical July20, Emergency Modeling: Contaminant Spills Motivation –Operations – needs to know what to do NOW –CCWD informed immediately, or, after the event Operational ?s: When/Where/How Much/How Long –Safety of water supply Toxicity Concentration –Ability to deliver water Arrival time at intakes Which intakes (Old River, Rock Slough, Mallard) –Cost Using Los Vaqueros reservoir Damage to pumps Duration of shut-down
CCWD: Using DSM2 Historical July20, Use Historical Model and PTM IDEA – For a given set of hydrological and operational conditions, find two+ times in Historical Model that bound this hydrology –SAC Flow and DCC –SJR flow and HORB –Exports – SWP and CVP –CCWD – Old River, Rock Slough Use PTM to run particle tracking models for first arrival time of contaminant
CCWD: Using DSM2 Historical July20, Method – Quick and Dirty Method: –Use prepared plots to find candidate times –Use HEC-DSSVue tables to narrow search –Run PTM at two (or more) times by injecting particles at source First arrival times at Old River Rock Slough – flux.txt Visualize for qualitative information for location of contaminant
CCWD: Using DSM2 Historical July20, Hypothetical Example 03/06/06: spill at 1:00 AM, downstream of Vernalis, ~ DSM2 node 3 SAC+SJ <83K cfs; Exports: ~6.8K cfs; No HORB or DCC
CCWD: Using DSM2 Historical July20, PTM Results: spill arrives in less than two days
CCWD: Using DSM2 Historical July20, Compare Real Result and Closest Scenario Mar 2006 Feb 1999
CCWD: Using DSM2 Historical July20, Compare Two Scenarios Feb 2000Feb 1999Feb 2000Feb 1999
CCWD: Using DSM2 Historical July20, Next Step ? Quick and Dirty – result in ~ hour –Example gave reasonable, fairly conservative estimates –BUT, may not be able to find good brackets Real-time DSM2 – –fairly time-consuming –need lots of data (but not EC if just a spill) –accuracy variable – see next talk More work with Historical DSM2?
CCWD: Using DSM2 Historical July20, Use DSM2 salinity fingerprinting in conjunction with field data to characterize the sources of modeled salinity error at CCWD intakes in the Delta, and at selected other locations The Objectives are to: –Identify systematic bias (seasonal, operational, …) –Quantity the error to allow us to put error bars on DSM2 forecasts –Possibly develop relationships to correct bias Were NOT trying to calibrate the model Using the Historical Model to Improve Forecasts
CCWD: Using DSM2 Historical July20, CCWDs Old River Intake
CCWD: Using DSM2 Historical July20, Method Compare residual (model – data) with: –Salinity fingerprints –NDOI (Net Delta Outflow Index) –DICU (Delta Island Consumptive Use) –Operations DCC (Delta Cross Channel) HORB (Head of Old River Barrier) Exports (SWP, CVP) Look for and characterize bias at: –Jersey (RSAN018), Holland (ROLD014), Bacon (ROLD024), CCWD Old R. Intake (ROLD034): –Started looking at Jersey Point – easiest –Started investigating North, moved South
CCWD: Using DSM2 Historical July20, EC-MTZ: Major Contributor to Jersey Point Modeled Error
CCWD: Using DSM2 Historical July20, Residual < 0: Occurs as Salinity is Falling in Late Fall
CCWD: Using DSM2 Historical July20, Residual > 0: Occurs in Fall as Salinity is Rising ….
CCWD: Using DSM2 Historical July20, JP Residual: Interesting Relationship With DICU/NDOI
CCWD: Using DSM2 Historical July20, Holland Tract: EC-MTZ Relationship Evident, But Not as Clean as JP
CCWD: Using DSM2 Historical July20, Bacon: Residual Changes w/EC-MTZ Are Messy (at Best), Data Questionable
CCWD: Using DSM2 Historical July20, Old River Intake: Model Usually Underestimates EC, Especially In Summer and Fall as EC-MTZ Peaks
CCWD: Using DSM2 Historical July20, Old River & EC-AG: Residual > 0 in Winter or Spring With Ag Events
CCWD: Using DSM2 Historical July20, Summary of Findings There is a suggestive relationship between modeled error and EC-MTZ at all 4 stations investigated: –Clearest at Jersey Point –Story more complicated as move south along Old River Relationships w/other salinity sources and operations: –No apparent relationship with DCC or HORB operations –No apparent relationship with EC from SJR, Eastside –Not sure about EC from Ag or Sac R. Modeled EC at Old River: –Underestimates EC as salinity increases in late summer and fall; related to EC-MTZ –Overestimates EC in Winter, Spring; related to EC-AG events Modeled EC at Jersey Point related to DICU/NDOI –Error greatest when DICU is a substantial portion of NDOI in the fall
CCWD: Using DSM2 Historical July20, Future Work More work with residuals: –Look for relationship with export operations, including CCWD diversions –Look at some station on Middle River –Include Volumetric Fingerprinting Look closer at ROLD034: –Incorporate more data –Quantify seasonal error –Look for other contributions to error Jersey Point: –Quantify error for EC-MTZ