An Evaluation of Water Quality Modeling Used to Inform Operational Decisions for the NYC Water Supply

Mark S. Zion, Donald C. Pierson and Elliot M. SchneidermanNew York City Department of Environmental Protection

Watershed Science and Technical ConferenceSeptember 13-14, 2012West Point, New York

Adao H. MatonseCUNY Institute for Sustainable Cities

2

Introduction

• DEP has developed an extensive suite of water quality and water system models which are used to analyze turbidity transport in the Catskill System.

• Elevated turbidity can be an issue for the Catskill System of the NYC water supply during and after periods of high streamflow.

• During recent storm events, DEP’s water quality and water system models were used to help inform operational decisions and reduce turbidity entering the water distribution system.

• Informed use of operations during event periods can help to mitigate the impacts of turbidity on the water supply

• How well does the modeling system work in predicting the range in future turbidity which is subsequently used to inform operations?

3

System Description

Kensico

Ashokan Reservoir OptionsD

IVID

ING

WEI

R

EAST BASIN

WESTBASIN

RELEASECHANNELCATSKILL

AQUEDUCT

GATEHOUSE

DIVIDINGWEIRGATE

4

Model Description

Kensico

CEQUAL-W2 Two Dimensional Reservoir ModelsAshokan

Esopus Creek

DividingWeir

CatskillAqueduct

ReleaseChannel

Spillway

Catskill Influent

Catskill Effluent

Delaware Effluent

Delaware Influent

• CEQUAL-W2 Models adapted to simulate turbidity transport by Upstate Freshwater Institute (UFI).

• Models are connected using OASIS-W2 combined modeling software and applied in a positional analysis mode.

• Linked models can also be run using LinkRes software developed by UFI and adapted for positional analysis by DEP.

5

Catskill Influent

Catskill Effluent

Delaware Effluent

Delaware Influent

Model Description

Kensico

Turbidity Profile

CEQUAL-W2 Two Dimensional Reservoir ModelsExample Longitudinal Profiles

Temperature

6

Model DescriptionOASIS system model combined with CEQUAL-W2 Reservoir Models

• OASIS Model used to simulate aqueduct flows and reservoir releases and spills

• Combined OASIS-W2 is backbone of Operations Support Tool (under development)

7

Reservoir Water Quality Model(CEQUAL-W2)

Meteorologic Input(Temp., RH, Solar Rad) Flows

OASIS Water System Model

Turbidity

Reservoir Storage

Effluent Turbidity

Initial Conditions(Reservoir Water

Temperature/Turbidity Profiles; Reservoir

Storage Levels;Snowpack)

Time Series Input (Forecast)

Initial Conditions (Snapshot)

Model

Results

Model Description – Application Method

8

Reservoir Water Quality Model(CEQUAL-W2)

Meteorologic Input(Temp., RH, Solar Rad) Flows

OASIS Water System Model

Turbidity

Reservoir Storage

Effluent Turbidity

Initial Conditions(Reservoir Water

Temperature/Turbidity Profiles; Reservoir

Storage Levels;Snowpack)

Time Series Input (Forecast)

Initial Conditions (Snapshot)

Model

Results

Model Description – Application Method

• Current reservoir water surface elevations

• Limnological surveys• Data from automated

profiles of water column temperature and turbidity

• Snow survey results

9

Reservoir Water Quality Model(CEQUAL-W2)

Meteorologic Input(Temp., RH, Solar Rad) Flows

OASIS Water System Model

Turbidity

Reservoir Storage

Effluent Turbidity

Initial Conditions(Reservoir Water

Temperature/Turbidity Profiles; Reservoir

Storage Levels;Snowpack)

Time Series Input (Forecast)

Initial Conditions (Snapshot)

Model

Results

Model Description – Application Method

• Forecast based on historical record or weather service analysis• Multiple traces representing many potential future input time series

10

Reservoir Water Quality Model(CEQUAL-W2)

Meteorologic Input(Temp., RH, Solar Rad) Flows

OASIS Water System Model

Turbidity

Reservoir Storage

Effluent Turbidity

Initial Conditions(Reservoir Water

Temperature/Turbidity Profiles; Reservoir

Storage Levels;Snowpack)

Time Series Input (Forecast)

Initial Conditions (Snapshot)

Model

Results

Model Description – Application Method

Multiple time series results representing each of the forecast traces

11

Model run for 57 inflow traces based on historical flow and meteorologic record (1948-2004)

Statistics of traces can be translated to cumulative probability function

Model Description – Application MethodExample of Positional Analysis

12

Case Study – Late December 2011

• These simulations are run to provide an estimate of the time that it might take for Ashokan Reservoir turbidity to reduce to a level at which alum use would no longer be necessary.

• In late August and early September, Tropical Storms Irene and Lee greatly impacted turbidity in the Ashokan and Schoharie Reservoirs and alum use was initiated on Catskill inflow to Kensico Reservoir.

• At the time of these model runs in late December 2011, the impact on the Catskill system was still significant.

• On December 20 Ashokan turbidity ranged from 22-170 NTU. Schoharie turbidity on December 14 ranged from 45-60 NTU.

• Alum continued to be used to reduce impact of turbidity entering Kensico Reservoir. In addition, the Catskill Aqueduct flow was reduced through the use of stop shutters.

Schoharie Reservoir, Jan 5, 2012

• Simulations consisted of integrated OASIS/W2 model run.

13

Ashokan Reservoir – Water Surface ElevOST Simulations of Dec. 20, 2011

Ashokan Reservoir – East Basin

Ashokan Reservoir – West Basin

Date

Individual Traces

Median

10th and 90th

Percentiles

14

Ashokan Reservoir – Water Surface ElevOST Simulations of Dec. 20, 2011 vs. Actual Elevation

Ashokan Reservoir – East Basin

Ashokan Reservoir – West Basin

Date

Individual Traces

Median

10th and 90th

Percentiles

Recorded Elevation 2012

15

Schoharie Reservoir ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Shandaken Tunnel

Date

Individual Traces

Median

10th and 90th Percentiles

16

Schoharie Reservoir ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Shandaken Tunnel

Date

Individual Traces

Median

10th and 90th Percentiles

Keypoint Data - Shadaken Tunnel Portal

17

Ashokan Reservoir – West Basin ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Ashokan West Basin Entering Dividing Weir Gate

Date

Individual Traces

Median

10th and 90th Percentiles

18

Ashokan Reservoir – West Basin ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Ashokan West Basin Entering Dividing Weir Gate

Date

Individual Traces

Median

10th and 90th Percentiles

Keypoint Data - West Basin Gate House (middle level)

19

Ashokan Reservoir – East Basin ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Turbidity Ashokan East Basin near Gate House

Date

Individual Traces

Median

10th and 90th Percentiles

20

Ashokan Reservoir – East Basin ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Turbidity Ashokan East Basin near Gate House

Date

Individual Traces

Median

10th and 90th Percentiles

Keypoint Data - East Basin Gate House

21

Ashokan Reservoir – Water Surface ElevOST Simulations of Dec. 20, 2011 vs. Actual Elevation

Ashokan Reservoir – East Basin

Ashokan Reservoir – West Basin

Date

Individual Traces

Median

10th and 90th

Percentiles

Recorded Elevation 2012

22

Ashokan Reservoir – Water Surface ElevOST Simulations of Dec. 20, 2011 vs. Actual Elevation

Ashokan Reservoir – East Basin

Ashokan Reservoir – West Basin

Date

Individual Traces

Median

10th and 90th

Percentiles

Recorded Elevation 2012

23

Schoharie Reservoir ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Shandaken Tunnel

Date

Individual Traces

Median

10th and 90th Percentiles

Keypoint Data - Shadaken Tunnel Portal

24

Schoharie Reservoir ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Shandaken Tunnel

Date

Individual Traces

Median

10th and 90th Percentiles

Keypoint Data - Shadaken Tunnel Portal

25

Ashokan Reservoir – West Basin ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Ashokan West Basin Entering Dividing Weir Gate

Date

Individual Traces

Median

10th and 90th Percentiles

Keypoint Data - West Basin Gate House (middle level)

26

Ashokan Reservoir – West Basin ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Ashokan West Basin Entering Dividing Weir Gate

Date

Individual Traces

Median

10th and 90th Percentiles

Keypoint Data - West Basin Gate House (middle level)

27

Ashokan Reservoir – East Basin ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Turbidity Ashokan East Basin near Gate House

Date

Individual Traces

Median

10th and 90th Percentiles

Keypoint Data - East Basin Gate House

28

Ashokan Reservoir – East Basin ResultsOST Simulations of Dec. 20, 2011 vs. Keypoint Data

Turbidity Ashokan East Basin near Gate House

Date

Individual Traces

Median

10th and 90th Percentiles

Keypoint Data - East Basin Gate House

29

Esopus Creek Inflow

Esopus Creek Turbidity

Case Study – Late March 2010

Catskill Influent

Catskill Effluent

Delaware Effluent

Delaware Influent

(30, 40 or 50 NTU)(50, 100, 150 or 600 MGD)

(1050, 1000, 950 or 900 MGD)

(1.5 NTU)

(400 MGD)

(700 MGD)

• A large rain and snowmelt event occurred on March 22, leading to elevated turbidity in Ashokan Reservoir.

• Stop shutters were installed in the Catskill Aqueduct to permit reduced flows from Catskill into Kensico Reservoir.

• What aqueduct flow rates should be used based on potential turbidity inputs from the Ashokan diversion?

• Kensico CEQUAL-W2 reservoir model is used to determine effects of elevated turbidity at various Catskill Aqueduct flow rates on Kensico Reservoir effluent.

30

Reservoir Water Quality Model(CEQUAL-W2)

Meteorologic Input(Temp., RH, Solar Rad)

Flows Turbidity

Effluent Turbidity

Initial Conditions(Reservoir Water

Temperature/Turbidity Profiles)

Time Series Input (Forecast)

Initial Conditions (Snapshot)

Model

Results

Model Description – Application MethodKensico Sensitivity Model Application

31

Reservoir Water Quality Model(CEQUAL-W2)

Meteorologic Input(Temp., RH, Solar Rad)

Flows Turbidity

Effluent Turbidity

Initial Conditions(Reservoir Water

Temperature/Turbidity Profiles)

Time Series Input (Forecast)

Initial Conditions (Snapshot)

Model

Results

Model Description – Application MethodKensico Sensitivity Model Application

• Current reservoir water surface elevations

• Limnological surveys• Data from automated

profiles of water column temperature and turbidity

32

Reservoir Water Quality Model(CEQUAL-W2)

Meteorologic Input(Temp., RH, Solar Rad)

Flows Turbidity

Effluent Turbidity

Initial Conditions(Reservoir Water

Temperature/Turbidity Profiles)

Time Series Input (Forecast)

Initial Conditions (Snapshot)

Model

Results

Model Description – Application MethodKensico Sensitivity Model Application

• Forecast based on historical record or weather service analysis• Multiple traces representing many potential future input time series

33

Reservoir Water Quality Model(CEQUAL-W2)

Meteorologic Input(Temp., RH, Solar Rad)

Flows Turbidity

Effluent Turbidity

Initial Conditions(Reservoir Water

Temperature/Turbidity Profiles)

Time Series Input (Forecast)

Initial Conditions (Snapshot)

Model

Results

Model Description – Application MethodKensico Sensitivity Model Application

Prescribed constant time series based on expected turbidity and potential flow rates from upstream reservoirs (via aqueducts)

34

Reservoir Water Quality Model(CEQUAL-W2)

Meteorologic Input(Temp., RH, Solar Rad)

Flows Turbidity

Effluent Turbidity

Initial Conditions(Reservoir Water

Temperature/Turbidity Profiles)

Time Series Input (Forecast)

Initial Conditions (Snapshot)

Model

Results

Model Description – Application MethodKensico Sensitivity Model Application

For each combination of fixed turbidity and flow inputs the model produces multiple traces of turbidity results based on the meteorologic input traces

35

Kensico Analysis Results – March 25, 2010Simulated Delaware Effluent Turbidity

30 NTU

Catskill Aqueduct Influent TurbidityCatskillAqueduct

Inflow

50 MGD

100 MGD

200 MGD

50 NTUTu

rbid

ity (N

TU)

Turb

idity

(NTU

)Tu

rbid

ity (N

TU)

Date Date

36

Influent Turbidity

Inflows

Catskill Aqueduct Delaware Aqueduct

Kensico Results - ComparisonCatskill Aqueduct Input Chosen for Comparison: 30 NTU @ 50 MGD

Turb

idity

(NTU

)

Turb

idity

(NTU

)Fl

ow (M

GD

)

Flow

(MG

D)

Date Date

Data

Model Input

37

Influent Turbidity Load

Cumulative Influent Turbidity Load

Catskill Aqueduct Delaware Aqueduct

Kensico Results - ComparisonCatskill Aqueduct Input Chosen for Comparison: 30 NTU @ 50 MGD

Turb

idity

Loa

d (M

GD

*NTU

)

Turb

idity

Loa

d (M

GD

*NTU

)

Date Date

Turb

idity

Loa

d (M

GD

*NTU

)

Turb

idity

Loa

d (M

GD

*NTU

)

Data

Model Input

38

Effluent Turbidity

Catskill Aqueduct Delaware Aqueduct

Kensico Results - ComparisonCatskill Aqueduct Input Chosen for Comparison: 30 NTU @ 50 MGD

Turb

idity

(NTU

)

Turb

idity

(NTU

)

Date Date

Data

Model Output

39

Conclusions

• Uncertainties in modeling system results can arise from initial conditions, forecast inputs and model errors.

• DEP has developed an extensive suite of water quality and water system models to analyze turbidity transport within the Catskill System Reservoirs.

• DEP’s water quality models have been effectively used to help minimize turbidity within Kensico Reservoir.

• Model output uncertainty tends to increase as the length of the model run increases.

• Tracing simulation results by comparison with measured data can indicate:

• when model output no longer applies• when forecast may need to be redone• areas of improvement for the modeling system .