Ventura River Watershed Model

1
Ventura River Watershed Model

• Dr. Jon Butcher
• 2/24/2009

2
Content of Presentation

• Introduction
• Supporting Data and Model Development
• Model Calibration and Validation
• Sources of Uncertainty
• Conclusions
• Next Steps
• Questions ?

3
Purpose of this Effort

• Develop, test, and document a hydrologic
  simulation model of the Ventura River watershed
• Provide basis for improved flood forecasting
• Framework for water supply evaluations
• Examine response to changes in land use,
  management
• Provide basis for future water quality simulation

4
What is a Simulation Model?

• All models are wrong some are useful George
  Box
• A simulation model is a mathematical
  approximation of the complex natural world
• Rigorous calibration and validation procedures
  are used to test and document how well the model
  performs
• Why do we need a model?
• Observation is preferable but many things cant
  (yet) be observed, such as
• Response to changing conditions (land use,
  climate, dams, land management)
• Prediction of risk from rare events (floods,
  droughts, fires)
• Testing likely response to management scenarios

5
The HSPF Model

• HSPF Hydrological Simulation Program FORTRAN
• Most well-established and widely applied
  comprehensive dynamic watershed simulation model
• Supported by USEPA and USGS
• Approved for use by FEMA
• Previously applied to Santa Clara and Calleguas
  Creek watersheds in Ventura Co.
• Basic model tasks Combine information on land
  characteristics, land use, and soils/geology with
  continuous weather time series to produce
  estimates of runoff, streamflow, and water
  quality throughout the modeled watershed

6
Weather

• Weather drives everything in complex ways
• Precipitation primary input
• Potential Evapotranspiration primary output
  evaporation and plant transpiration
• Temperature controls evaporation determine
  whether precipitation is rain or snow

7
Representing the Land Surface

• Land Use
• Impervious Cover
• Soils
• Slopes

8
Unique Aspects of the Ventura Watershed Model

• Represent Land Use Change over Time
• Include Fire Impacts
• Hydrologic Response Unit (HRU) Approach

9
Land Use Land Cover
10
Land Use Change Over Time

• Land use change over time is small
• Developed land increases from 5.4 to 6.7 of
  watershed
• Change in some individual subwatersheds is more
  substantial

11
Irrigation

• Additional source of soil moisture
• Largely recycled from deep ground water
• No major sources of imported water
• Occurs on both agricultural and developed
  pervious land
• Use CIMIS approach to calculate potential
  irrigation application rates
• Most crops and lawns are irrigated most pasture
  is not

Potentially Irrigated Agricultural Lands
12
Soils and Slopes
13
Significant Fires in Ventura River Watershed
(1967-2006)

• Fires change the properties of the soils and
  vegetation
• Burned areas have reduced infiltration and
  increased erosion, but also reduced
  evapotranspiration
• Effects may persist for 2-5 years
• Fire impacts are more significant that other
  forms of land use change in the watershed

14
Fire Impacts on Hydrology

• Burned areas have higher flood peaks and higher
  baseflows particularly during dry years

15
Watershed Segmentation

• Define the spatial resolution of the model
  areas are lumped within subwatersheds
• 88 Model subwatersheds
• Average size 1600 acres
• Isolate FEMA tributaries
• Breakpoints correspond to gages, monitoring
  stations, point source inputs
• Arranged to account for changing in flow routing
  over time

16
Meteorology

• Rain Gages assigned based on proximity
• Modified to conform to rainfall patterns
• Elevation has a big effect!
• Balance positive and negative bias

17
Hydrologic Response Units

• Model unit areas within each subwatershed are
  represented as Hydrologic Response Units (HRUs)
• Each HRU represents a unique combination of land
  use/ land cover/ soil group/ precipitation gage,
  on a unit area basis
• Model parameters can be adjusted by HRU
• All HRUs are simulated for the entire model
  application period linked to reach model in
  different combinations over time to account for
  land use change

18
Representing the Stream Network

• Stream Segmentation
• Stream Hydraulics
• Reservoirs
• Debris and Detention Basins
• Diversions and Withdrawals
• Interactions with Ground Water

19
Stream Segmentation

• Stream segments correspond to the upland
  subwatershed
• Each subwatershed has a defined stream reach
• Isolate each FEMA tributary
• Additional reaches are added to represent
  reservoirs, detention basins, and diversions
• Final segmentation reflects stakeholder input and
  review

20
Stream Hydraulics

• HSPF represents hydraulics through Functional
  Tables of volume-stage-discharge relationships
• Determine these from HEC-RAS flood elevation
  models where available (Mainstem, San Antonio
  Creek)
• Estimate for other reaches based on trapezoidal
  shape approximation and regional regressions on
  dimensions using LIDAR.

21
Reservoirs - Casitas

• Characterize reservoir operations from reported
  data

22
Reservoirs - Matilija

• Sedimentation has reduced pool storage over time
  from 7000 to about 500 AF
• Model suggests that the effective storage (water
  that can be released in excess of inflows) is
  greater than 500 AF due to temporary storage in
  the sand on the reservoir bottom
• Represent releases using gaged flow below the dam
  with weir equation for spillage

23
Debris and Detention Basins
24
Diversions and Withdrawals

• Withdrawals from Casitas (about 20,000 AF/yr)
• Robles Diversion Releases from Matilija
  Reservoir to Casitas Reservoir (about 9,000
  AF/yr)
• Withdrawals via Foster Park Diversion from
  Mainstem
• Flood Control Diversions
• Live Oak Diversion
• Rancho Matilija Diversion
• East Ojai Drain
• McDonald Canyon Drain
• Happy Valley Drain
• Miramonte Drain
• Skyline Drain

25
Interactions with Ground Water

• Portions of system naturally lose to or gain from
  ground water
• Losses are enhanced by ground water pumping
  (e.g., Ojai Basin)
• No detailed groundwater models exist
• Ground water interactions specified iteratively,
  consistent with
• Geologic information
• Pumping information
• Overall water balance

26
Unique Aspects

• Link to HRU model to represent change in land use
• Changes in flow routing and reservoir management

27
Model Calibration and Validation Approach

• Calibration
• Adjust model parameters to achieve optimal fit
• Carried out for 1996-2006
• Validation
• Test the performance of the calibrated model
• Carried out for 1986-1995
• Compare results at 8 stream gages to criteria
  specified in the modeling Quality Assurance
  Project Plan
• Extended application (1967-2006) to examine
  performance in predicting peak flow events

28
Summary of Performance of the Calibrated Models

• Calibration and validation successful, with a few
  exceptions
• Some problems associated with Coyote Creek and
  Santa Ana Creek, where gage records appear
  suspect
• R2 values for daily flows at other gages range
  from 86 to 94

29
Representation of the Water Balance

• 23 of rainfall reaches the ocean

30
Fit to Individual Stream Gages San Antonio Creek
31
Fit to Individual Stream Gages Ventura R. _at_
Foster Park
32
Fit to Individual Stream Gages Santa Ana Creek

• Most, but not all gages fit well
• Observed and simulated flows deviate for Santa
  Ana Creek
• Observed flows dont correlate well with
  rainfall discrepancy may be due to infrequent
  maintenance and calibration of the gage

33
Prediction of Flood Peaks

• Predict individual events
• Strive for match between predictions of 100-year
  flood using Bulletin 17B methodology for observed
  and simulated flood peaks

100 yr Event
34
Prediction of Flood Peaks

• Divergence at some stations appears to be due to
  spatial variability in precipitation

35
Sources of Uncertainty

• Lack of a Ground Water Model
• Primarily affects low to medium flow prediction
• Translating point rainfall to subwatershed areal
  totals
• May mis-estimate effective rainfall
• Does not fully account for timing
• Actual irrigation rates
• Detailed hydraulics for channels outside of
  Ventura River mainstem and San Antonio Creek

36
Conclusions

• Model calibration and validation Successful
• Model performs well in reproducing all aspects of
  the water balance and replicating gaged flows
• Some discrepancies associated with gage record
  certain sites
• Simulation of high flow events is good to
  excellent at most locations
• Model is ready for for use for in scenario
  evaluation for flow prediction and analysis

37
Recommendations

• Develop a basin ground water model and link to
  the surface water model
• Develop HEC-RAS flood elevation models for
  additional segments of interest and use to inform
  watershed model
• Test and improve calibration of flow gages on
  Coyote Creek and Santa Ana Creek
• Consider use of integrative rainfall measurement
  techniques, such as Doppler radar interpretation

38
Next Phase Natural Conditions Analysis

• Remove reservoirs and revert streams to natural
  path and channel form
• Remove groundwater pumping
• Remove point sources
• Convert the human landscape to natural cover
• Remove agriculture, developed land,
  transportation
• Replace with LANDFIRE Potential Natural
  Vegetation Group cover
• Simulate using 1967-2007 weather for direct
  comparison to current conditions run

39
Questions?