Potter Valley Project HEC ResSim Model...

Potter Valley Project HEC‐ResSim

Hydrologic Model Validation Review

Prepared by the Water Supply Modeling Subgroup for the Water Supply Working

Group and Ad‐Hoc Committee

October 3, 2018

Objectives of PVP Model Validation Effort

• Ensure that new HEC‐ResSim model is accurately portraying primary performance metrics under existing conditions (Lake Pillsbury Storage, Eel River Releases, Diversions to Potter Valley).

• Enable group confidence that upcoming hydrology scenarios (and comparisons) are accurately portrayed by the model to a level where outputs can be used for water supply evaluation and ecological evaluations on both the Eel River and Russian River basins.

Model Validation Review: Compared

Three Different Outputs (performance metrics)

1.Lake Pillsbury Storage

2.Cape Horn Dam E‐11 Flows

3.Diversion Tunnel E‐16 Flows

E‐2 flows, EBRR flows, and other metrics are considered lower priority for illustration here, but should be accurate if above metrics are validated

Assumptions for Validation Process

• Daily time step for Water Year (WY) 1911‐2017 Simulation time series; WY 2007‐2017 Calibration/Validation time series

• Unimpaired flows obtained from recent mass balance computations (Cardno 2018)

• Observed storage and flows are accurately measured/computed

• Drought/low water supply variances and PG&E construction variances (i.e., abnormal operations) *not* programmed in model runs. Deviations due to variances noted in output.

• New predictive linear relationship used for the model to predict date of gate closure. This linear relationship is only an estimation of actual operations based on observed data, so deviations in storage due to gate closure noted in output. Note: actual gate closure operations are based on best available data at the time (e.g., snowpack estimates), whereas modeled gate closure is based on known spring/summer cumulative inflow.

• Uses 2017 Block Water Design hydrograph, initiated when E‐2 flows drop below 250 cfs starting after April 1st. Block water releases made from Scott Dam slide gates.

• Notation in upcoming performance metrics note the name of the database row used in the plot

• Blue line=Observed, Red line=Simulation Model prediction

PG&E Variances During Validation Period

• Construction variance periods:

1) Nov 2014 ‐ Feb 20152) Nov 2015 ‐ Feb 20163) Nov 2017 ‐ Jan 2018

• Drought/low water supply variance periods:

1) Dec 2013 ‐ Feb 20142) May 2015 ‐ Dec 20153) July 2016 ‐ Oct 2016

Examples of assumptions specific to the metrics• Variances

• Focusing on lower flows for E‐11

• End of fall storage is a priority for storage metric

Model drains the reservoir (5,000 ac‐ft minimum), i.e., what would have happened without a drought variance

Drought variance (reduced releases) was implemented May 2015

Lake Pillsbury Storage: Example of divergence of simulation from historical operations during a drought variance

Flows below Cape Horn Dam (E‐11) shown on log scale to focus on lower flows, i.e. summer base flows

Observed Conditions and Existing Model Runs

• Observed PVP Ops:

• Modeled Existing (Validation):

Overview of WY 2007‐2017

METRIC #1: Lake Pillsbury Storage

Comparison of:OBSERVED PVP Ops: Number 4

MODELED EXISTING (Validation): Number 5

METRIC #2: Flows Below Cape Horn Dam (E‐11)



METRIC #3: Flows through PVP Diversion

Tunnel (E‐16)Comparison of:

OBSERVED PVP Ops: Number 5MODELED EXISTING (Validation): Number 3

Note E‐11 charts are scaled to log axis

Overview of WY 2007‐2017, Part 2

Due to actual later gate closure

Due to drought variance reduced releases

Difference in storage due to later gate closure

Differences at lower flows mainly due to modeled versus actual buffer flows. In recent years PG&E has typically run larger buffer flows due to increased awareness of license compliance risks.

Powerhouse outages & construction variances not modeled, resulting in modeled diversions that did not actually occur. Additionally, drought/low water supply variances not modeled, resulting in model diversions going to zero when the reservoir was drained (in reality, reduced diversions were occurring).

Discretionary flow decisions are extremely situational and so it is challenging to capture operators’ professional judgment as model logic. Model tends to be over‐optimistic about diversions, and so a cap (170 cfs ) was set on discretionary flows. The cap was calibrated to capture monthly behavior.

Due to difference in buffer flows released





Comparison of:OBSERVED PVP Ops: Number 2MODELED EXISTING (Validation):

Number 2


Tunnel (E‐16)


Number 3

WY 2007 – WY 2008

Estimated Unimpaired flows are between Scott Dan And Cape Horn Dam are higher than observed.

Operational buffer flows are different between the observed and modeled.

Operators released discretionary flow through the tunnel.

Model uses logic to estimate discretionary flow through the tunnel. Actual discretionary flow is subject to operator’s discretion.



Number 5



Number 2


Number 3


Tunnel (E‐16)

WY 2008 – WY 2009

Flood gates remained open longer than assumed by the model, reducing reservoir storage later in the year.

Estimated Unimpaired flows are between Scott Dan And Cape Horn Dam are higher than observed.

Tunnel diversions shut down due to high flows. Model shuts down diversions at 7,000 cfsflow into Cape Horn Dam.






Number 2

METRIC #2: Flows Below Cape Horn

Dam (E‐11)

METRIC #3: Flows through PVP

Diversion Tunnel (E‐16)


Number 3

WY 2009 – WY 2010

Different buffer flows for MF11 are assumed in model.







Number 2


Dam (E‐11)



OBSERVED PVP Ops: Number 5MODELED EXISTING (Validation):

Number 3

WY 2010 ‐WY 2011

Flood gates remained open longer than assumed by the model.

Model assumes discretionary flow being released through the tunnel, while not occurring in observed.

Estimated peak unimpaired flows are higher than observed.

Ramp up & down on either side of increased discretionary flow


Possibly block water releases







Dam (E‐11)





WY 2011 – WY 2012

Flood gates assumed to close more slowly in model.








Number 2


Dam (E‐11)





WY 2012 – WY 2013








Number 2


Dam (E‐11)



OBSERVED PVP Ops: Number 5MODELED EXISTING (Validation):

Number 3

WY 2013 – WY 2014

Model drains the reservoir (5,000 ac‐ft minimum)

Difference in drawdown due to Dry Spring Exclusion not being calculated in the model, but is in observed conditions.

Dry Spring Exclusion not calculated in the model.

Not enough modeled Lake Pillsbury storage for tunnel diversions.


Emergency drought variance approved in Dec. 2013.



Number 5


Number 2




Dam (E‐11)


Number 3

WY 2014 – WY 2015


Spill under‐forecasted because model drained reservoir without variance.

Block water release assumed in the model.

Estimated peak unimpaired flows are higher than observed.

Observed construction variance Nov/15/2014 –3/5/2015. Model assumes discretionary flow.

Observed construction variance Nov/15/2014 –3/5/2015. Model assumes discretionary flow.

Difference in storage due to Dec. 2013‐Feb. 2014 emergency drought variance (not modeled).



Number 5


Number 2


Dam (E‐11)


Number 3



WY 2015 – WY 2016

Model drains the reservoir (5,000 ac‐ft minimum)

The difference in the troughs is attributable to implementation of drought variance (lower reservoir releases). The model does not incorporate lowered flow releases from drought variance.

Implementation of drought variance (lower reservoir releases). The model does not incorporate lowered flow releases from drought variance.




Number 5


Number 2


Dam (E‐11)




Number 3

WY 2016 – WY 2017

Flood gates remained open longer than assumed by the model, reducing reservoir storage early in the year, but drought variance made up for storage by end of summer.

Implementation of drought variance (lower reservoir releases). The model does not incorporate lowered flow releases from drought variance.

Modeled discretionary flows capped at 170 cfs.



Number 5


Number 2


Dam (E‐11)


Number 3



Wrap up and Discussion

• Any additional Validation work needed?

Date post:	21-Jan-2021
Category:	Documents
Upload:	others
View:	2 times
Download:	0 times

Potter Valley Project HEC ResSim Model...

Documents