Niagara Power Project FERC No. 2216

 

UPPER NIAGARA RIVER TRIBUTARY BACKWATER STUDY

 

HTML Format.  Text only

 

Prepared for: New York Power Authority 

Prepared by: URS Corporation; Gomez and Sullivan Engineers, P.C.; and E/PRO Engineering & Environmental Consulting, LLC

 

 

August 2005

 

___________________________________________________

 

Copyright © 2005 New York Power Authority

 

 

aBBREVIATIONS

Agencies

FEMA              Federal Emergency Management Agency

FERC               Federal Energy Regulatory Commission

IJC                   International Joint Commission

INBC               International Niagara Board of Control

NOAA             National Oceanic and Atmospheric Administration

NYPA              New York Power Authority

USACE            United States Army Corps of Engineers

USGS               United States Geological Survey

Units of Measure

cfs                    cubic feet per second

El.                    elevation

fps                    feet per second

IGLD               International Great Lakes Datum

MW                 megawatt

NGVD             National Geodetic Vertical Datum

USLSD            U.S. Lake Survey Datum 1935

Miscellaneous

HEC-RAS        Hydraulic Engineering Center - River Analysis System

 

EXECUTIVE SUMMARY

As part of the relicensing effort for the Niagara Power Project, NYPA conducted an engineering analysis of surface water level and flow fluctuations in the Niagara River and presented the analysis in a report entitled Niagara River Water Level and Flow Fluctuation Study (URS et al. 2005).  

The upper Niagara River Tributary Backwater Study is a supplement to URS et al. 2005 and was done to provide additional information on the effect that water levels in the upper Niagara River can have on water levels in tributaries.  The study provides an estimate of the tributary reach lengths that could be influenced by water surface elevation fluctuations in the upper Niagara River.  It was used to determine the geographic scope of studies for other resources, including habitat, shoreline erosion, and water quality. 

Seven tributaries were selected based on field reconnaissance, the availability of existing hydraulic models, and the possible significance of the tributaries to the relicensing process for the Niagara Power Project.  The tributaries are Woods Creek, Gun Creek, and Spicer Creek, located on Grand Island, and Cayuga Creek, Bergholtz Creek, Tonawanda Creek, and Ellicott Creek, located on the U.S. mainland (Figure EX-1.). 

The study determined the length of each tributary that is influenced by fluctuations in water levels of the upper Niagara River.  The results are based on the median annual flow for each tributary and water levels for the upper Niagara River that included the annual minimum and maximum and the 5%, 50% (median), and 95% exceedance values for the period of record (1991- 2002). 

The tributary reach lengths influenced by upper Niagara River water levels are considered to be conservative estimates of the effects of U.S./Canadian hydropower generation for a median flow event because water level fluctuations are caused by other factors in addition to U.S./Canadian hydropower generation.  These factors include wind, natural flow and ice conditions, regional and long-term precipitation patterns that affect Lake Erie levels, control of the Niagara Falls flow for scenic purposes, and operation of power plants on the Canadian side of the river.  Water level fluctuations in the upper Niagara River, from all causes, are generally less than 1.5 feet per day.  As stated above, this study investigated a wide range of water levels, from the minimum to the maximum recorded elevations in the upper Niagara River (1991-2002), a difference that can span 4 to 6 feet.  Such a range in elevation does not represent a typical upper Niagara River daily water level fluctuation of 1.5 feet, or less.  Conditions other than Project operations (i.e., wind on Lake Erie), often drive the Niagara River to the extreme water levels such as the annual maximum and minimum elevations. 

Water levels in the upper Niagara River influence water levels in each of the 7 tributaries – Woods Creek, Gun Creek, Spicer Creek, Cayuga Creek, Bergholtz Creek, Tonawanda Creek, and Ellicott Creek.  Table EX-1 provides the estimated length of the study tributaries influenced by the Niagara River for the three highest study elevations: the median (50% exceedance), the 5% exceedance and the annual maximum water surface elevations.  The median Niagara River water level is representative of daily water levels due to U.S./Canadian hydropower generation.  Conversely, the annual maximum Niagara River water level occurred during a storm surge from Lake Erie and is not representative of the range of daily water level fluctuations due to U.S./Canadian hydropower generation.  For all of the tributaries studied except Gun Creek, the affected stream length varies depending on the water level of the upper Niagara River (see Table EX-1).  A steep profile in Gun Creek between 2,900 feet and 4,200 feet upstream of the mouth limits the upstream influence of Niagara River water level fluctuations on creek water levels. Figure EX-2 illustrates the tributary segments that were found to have water surface elevations influenced by the median upper Niagara River elevations.  

Analysis of water levels in Tonawanda and Ellicott Creeks is complicated by the fact that dredging and water diversions, for the purposes of navigation on the Erie (New York State Barge) Canal and flood control, have altered the hydraulics and hydrology in both tributaries and the relationship of these tributaries to the upper Niagara River.  An additional factor for Tonawanda Creek is that 11.6 miles of the tributary are part of the Erie (Barge) Canal system where water levels are influenced by canal operations.  In order to maintain a uniform canal width and depth and flat slope, the Tonawanda Creek channel has been dredged and maintained.  During the canal operational period (generally the first week of May through the last week of October) when the guard lock gate in Lockport is opened, flow in the creek often moves in an upstream (easterly) direction because the level of the Erie (Barge) Canal at Lockport is lower than that of the Niagara River at the City of Tonawanda.  Up to 1100 cfs of flow are diverted from the Niagara River for canal operations.  Accordingly, water levels in Tonawanda Creek are not solely influenced by upper Niagara River water levels. 

For the median Niagara River water level condition, the extent of influence of Niagara River water level was estimated to be 13.7 miles upstream of the mouth of Tonawanda Creek.  This estimate seems reasonable as the upstream end of the affected tributary has a more riverine character (as compared to sections of the Erie (Barge) Canal) and is near two riffles.  Field observations confirmed that two riffles, located at 13.6 miles and 14.1 miles from the mouth, act as hydraulic controls limiting the upstream influence of Niagara River water level fluctuations due to U.S./Canadian hydropower generation on Tonawanda Creek water levels (Gomez and Sullivan and E/PRO 2005).  For the annual maximum Niagara River water level, the extent of influence of Niagara River water levels is nearly 19 miles upstream from the Niagara River based on engineering judgment.  (The engineering judgment was made by extending the median and annual maximum water levels of the Niagara River to where they intersect the stream bottom of Tonawanda Creek.  The stream bottom profile was obtained from Flood Insurance Studies for towns and communities that are intersected by the creek.)

The analysis of Ellicott Creek found that flood control and dredging operations have also changed the hydraulic and hydrologic characteristics of the creek.  Based on engineering judgment, upper Niagara River water levels influence water levels in Ellicott Creek approximately 7 miles upstream of the mouth.  This distance coincides with a riffle that extends from 6.9 to 7.1 miles from the mouth, and for which field observations confirmed was a hydraulic control limiting the upstream influence of Niagara River water level fluctuations due to U.S./Canadian hydropower generation on creek water levels (Gomez and Sullivan and E/PRO 2005).  (The engineering judgment was made by extending the median and annual maximum water levels of the Niagara River to where they intersect the stream bottom of Ellicott Creek.  The stream bottom profile was obtained from Flood Insurance Studies for towns and communities that are intersected by the creek.)  To reduce flooding, both the U.S. Army Corps of Engineers (USACE) and local government entities have altered the channel by making it deeper and wider and by building a dam at Island Park near the Village of Williamsville.  Regulation occurs today by the seasonal manipulation of that dam and by intermittent pumping from stone quarries into stream. 

 

Table EX-1

Estimated Annual Median Flow, Approximate Length of a Tributary that was Studied, and Estimated Length of a Tributary Influenced by Upper Niagara River Water Surface Elevations

Study Tributary

Estimated Annual Median Flow (cfs)

Approximate Study Length of Tributary

Estimated Length of Tributary Influenced by Upper Niagara River (to nearest 100 feet)

At Median Niagara River Water Surface Elevations

At 5% Exceedance Niagara River Water Surface Elevations

At Maximum Niagara River Water Surface Elevations

Woods Creek

5.9

16,170 feet

9,000 feet

9,000 feet

11,000 feet

Woods Creek - Tributary 1

2.3

14,355 feet

1,300 feet

1,300 feet

2,400 feet

Woods Creek -Tributary 3

0.3

8,920 feet

No Influence

No Influence

No Influence

Spicer Creek

2.4

12,680 feet

2,600 feet

2,600 feet

4,600 feet

Gun Creek

2.7

7,460 feet

4,100 feet

4,100 feet

4,100 feet

Cayuga Creek

27.5

14,690 feet

8,300 feet

9,700 feet

10,100 feet

Bergholtz Creek

17.7

16,420 feet

9,000 feet

9,000 feet

10,900 feet

Ellicott Creek

77.2

11,000 feet

7.0 miles1

7.0 miles1

7.1miles 1

Tonawanda Creek

408.2

10,570 feet

13.7 miles1

18.8 miles1

18.9 miles1

1Based on engineering judgment from information provided in Flood Insurance Studies on the tributary profiles and the upper Niagara River water surface elevations

 

Figure EX-1

Location Plan – Upper Niagara River and Studied Tributaries

 

Figure EX-2

Tributary Segments that are Influenced by Upper Niagara River Median Water Levels

 

1.0     INTRODUCTION

The New York Power Authority (NYPA) is engaged in the relicensing of the Niagara Power Project in Lewiston, Niagara County, New York.  The 1,880-MW (firm power output) Niagara Power Project is one of the largest non-federal hydroelectric facilities in North America and was licensed to the Power Authority of the State of New York (now the New York Power Authority) in 1957.  Construction of the Project began in 1958, and electricity was first produced in 1961.

The present operating license for the plant expires in August 2007.  In preparation for the license submittal for the Niagara Project, NYPA is assembling information related to the ecological, engineering, recreational, cultural, and socioeconomic aspects of the Project. 

Water level fluctuations in the upper Niagara River are caused by a number of factors in addition to the operation of the Niagara Power Project.  These factors include wind, natural flow and ice conditions, regional and long-term precipitation patterns that affect Lake Erie levels, control of the Niagara Falls flow for scenic purposes, and operation of power plants on the Canadian side of the river.  Generally, water level fluctuations in the upper Niagara River, from all causes, are less than 1.5 feet per day. 

Currently, there are two regulatory constraints on flow and water level fluctuations – The Niagara River Water Diversion Treaty of 1950 and the 1993 Directive of the International Niagara Board of Control.  To balance the need for power with a desire to preserve the beauty of Niagara Falls, the Treaty provides for the regulation of the amount of water diverted for hydroelectricity production.  On average, more than 200,000 cubic feet per second (cfs), or 1.5 million gallons of water a second, flows from Lake Erie into the Niagara River.  The Treaty requires that at least 100,000 cfs of water be allowed to spill over Niagara Falls during the tourist season daylight hours, defined as April 1 through October 31.  The 100,000 cfs flow may be reduced to 50,000 cfs at night during this tourist season and throughout the day the rest of the year.  The remaining Niagara River flow is available for hydroelectric generation by Canada and the United States and is to be shared equally by the two nations.

Pursuant to the requirements of the 1993 Directive of the International Niagara Board of Control (INBC), water level fluctuations in the Chippawa-Grass Island Pool (in the upper Niagara River, i.e., above Niagara Falls) are limited to 1.5 feet per day within a 3-foot range for normal conditions (Figure 1.0-1).  For unusual conditions (high flow, low flow, ice, etc.), the allowable range of Chippawa-Grass Island Pool water levels is extended to 4 feet and the 1.5 feet daily fluctuation tolerance can be waived.  The water level fluctuation limits of the 1993 Directive are independent of the seasonal flow limits dictated by the 1950 treaty.

As part of the relicensing effort for the Niagara Power Project, NYPA conducted an engineering analysis of surface water level and flow fluctuations in the Niagara River and presented the analysis in a report entitled Niagara River Water Level and Flow Fluctuation Study (URS et al. 2005). 

The Upper Niagara River Tributary Backwater Study is a supplement to the above referenced study and is intended to provide additional information on the effect water levels in the upper Niagara River have on tributary water levels.  The study provides conservative estimates of tributary reach lengths that could be influenced by different elevations in the upper Niagara River by U.S./Canadian hydropower generation.  The estimates are conservative because water level fluctuations are caused by other factors in addition to U.S./Canadian hydropower generation.  Since this information was utilized to determine the geographic scope of studies for other resources such as habitat, shoreline erosion, water quality, etc., the other studies produced “conservative” estimates of impacts.  

Seven tributaries were selected for this study on the basis of a field reconnaissance of the tributary, the availability of existing hydraulic models, and possible significance of the tributary relative to the relicensing process for the Niagara Power Project.  The locations of the tributaries are shown on Figure 1.0-2 and they are Woods Creek, Gun Creek, and Spicer Creek, which are located on Grand Island, and Cayuga Creek, Bergholtz Creek, Tonawanda Creek, and Ellicott Creek, which are located on the U.S. mainland (eastern Niagara River shore).  Bergholtz Creek joins Cayuga Creek approximately 5,600 feet upstream of Cayuga Creek's confluence with the upper Niagara River; Ellicott Creek joins Tonawanda Creek approximately 1,600 feet upstream of Tonawanda Creek's confluence with the upper Niagara River.  

 

Figure 1.0-1

Regulation of the Chippawa-Grass Island Pool Water Levels as Specified by the INBC 1993 Directive

Elevation Datum: USLSD 1935.  To convert water levels in the upper Niagara River from USLSD 1935 to IGLD 1985 subtract 1.2 feet

Abnormal flow conditions are considered to exist when any four consecutive hourly mean Niagara River flows, as determined from levels at the Fort Erie gauge, are greater than 270,000 cfs or less than 150,000 cfs.

 

Figure 1.0-2

Upper Niagara River and Studied Tributaries

 

2.0     STUDY METHODOLOGY

The analysis of the seven tributaries selected for this study (see Section 1.0) began with the establishment of hydrology estimates and the development of hydraulic models for each tributary.  The following sections describe the methods used to develop the hydrologic and hydraulic components of this study. 

2.1         Hydrology

The flow (discharge) of water in a tributary will influence the water surface elevations found at points along the tributary.  When the physical nature of the stream channel (i.e., geometry, slope, channel roughness), applicable backwater effects, and the stream discharge are entered into a hydraulic model of the stream reach, the water surface profile for that reach can be determined. 

The discharge(s) to apply in the calculation of a water surface elevation should be related to the purpose(s) of the study.  For a study of water surface elevations found at locations along a stream during a severe flood, a flood discharge would be developed using statistical analyses of existing gauges and/or empirical models.  The calculated flood discharge would then be applied in a hydraulic model to determine the flood water surface elevations. 

The purpose of the upper Niagara River Tributary Backwater Study is to provide conservative estimates of the tributary reach lengths that could be influenced by fluctuations in the upper Niagara River.  Accordingly, the discharge to apply in this study would be one that is representative of the "normal" conditions found in the tributaries, as opposed to an extreme discharge event that would be applied in a flood study or low-flow application. 

For this study, the median annual flow or 50% exceedance flow was selected to be representative of the "normal" discharge conditions in the tributaries.  The median annual flow is a statistical value that is useful for describing the central tendency of flow over the year.  Thus, the discharge value applied to determine water surface elevations is one that would generally be found in the middle of a list of average daily flow records sorted for the year.  That is, one-half of the flows would generally be higher and one-half would generally be lower than the median flow value. 

Actual stream discharge records, which are used to determine the median annual flow at a specific locale in the tributary, are often not available and therefore, alternative engineering methods must be employed to estimate the median annual flow.  The methodology selection begins by investigating the availability of flow records within a given tributary.  If flow data are available for the tributary, such as from records taken by USGS stream flow gauges, and the records are of sufficient length to be representative of the hydrologic changes that can occur annually, those data would be used to determine the median annual flow. 

Of the seven tributaries investigated in this study, only two (Tonawanda Creek, Ellicott Creek) have stream flow data recorded by a stream flow gauge on the tributary.  The Tonawanda and Ellicott Creek stream flow gauges were located near the area considered for this study, and each have a record of sufficient length (over 25 years) to provide a normal distribution of flow.  Therefore, the gauge records were used to estimate the median annual flow for those tributaries.  Because the gauges were not located at the exact location of this study, the gauge records required adjustment using a ratio of drainage areas.

The drainage area ratio adjustment method is commonly used to estimate hydrologic conditions.  Drainage area is often the primary variable responsible for the quantity of discharge at a stream location in the watershed.  Accordingly, the drainage area ratio method employs records from the nearby gauging station and adjusts those records to the location of interest by the proportion of drainage areas:

Flow at Ungauged Location = Ay/Ax * Flow at Gauged Location

 

Where:           Ay = Drainage area at ungauged location,

Ax = Drainage area at gauged location.

The results of the median annual flow estimates for Tonawanda and Ellicott Creeks are provided in Section 3.0. 

For the five study tributaries without stream discharge records, an alternate engineering method was employed to estimate the annual median flow for each tributary.  The method used in this study was one that applied regional relationships of the median annual flow to the physical characteristics (drainage area, annual precipitation, etc.) of watersheds in the western New York region. 

The process began with the review of USGS flow gauges in the western New York region.  Factors considered when selecting a streamflow gauge included the length of gauge record available, the mean annual precipitation at the gauge, and geographical location.  The initial list of gauges was examined further to exclude gauges with a short or spotty gauge record, regulated stream flows, and/or a dissimilarity in watershed characteristics (e.g., unacceptable differences in basin area, percent forested, percent urbanized).  The final list of gauges that were used in this study to develop a median annual flow estimate for the five ungauged tributaries is shown in Table 2.1-1.  Figure 2.1-1 depicts the location of the gauges used in the analysis.

To reduce the probability for error in the multiple (regression) analysis, it was important that each sample set of regional flow gauge records be of equal size, or period of record.  As is shown in Table 2.1-1, the period of record is not common among the thirteen gauges selected in this analysis.  Therefore, to accurately develop a relationship to predict gauge flows, it was necessary to acquire missing data and/or extend the gauge records to a common period by developing a cross-correlation of flow between the records.  This study used an approach known as maintenance of variance extension (MOVE) to develop the correlations structure of the gauge records (Maidment 1992).  Additional information on the MOVE is provided in Appendix A. 

The multiple regression analysis was performed on the equally sized gauge records constructed for the 93 year period from January 1909 through December 2001.  The regression analysis was predicated with the fact that more than one independent variable was needed to estimate the median annual flow in a tributary.  The forward selection of variables found that two variables, drainage area and average annual precipitation, provided an effective model of the median annual tributary flow.  The analysis also found that the relationship of these independent variables to the median annual flow was best described with an exponential function of the variables.  Several computer models were developed to assess the relationship of these variables and those models found that the median annual flow was best estimated with the following equation: 

Q = C * D A * P B

where:    Q = Estimated Flow

C = Coefficient of Determination (0.001)

D = Drainage Area (square miles) of Tributary

P = Average Annual Precipitation (inches) within Tributary Watershed

A = Power factor to raise Drainage Area (0.95)      

B = Power factor to raise Average Annual Precipitation (1.88)

 

The median annual flows estimated for the five ungauged tributaries are presented in Section 3.0. 

Additional information on the regression analyses used for the development of hydrology in ungauged tributaries can be found in Maidment (1992) and in Appendix A. 

2.1         Hydraulic Models of Study Tributaries

The Hydrologic Engineering Center River Analysis System hydraulic program (HEC-RAS, version 3.1.1, May 2003) developed by the U.S. Army Corps of Engineers, Davis, CA, was used to analyze the effects of water levels in the upper Niagara River.  The models require, as input, detailed survey information, such as stream cross-sections, bridge and culvert geometry, the type of bridge or culvert, and the longitudinal profile of the streambed.  Those data were obtained from Federal Emergency Management Agency (FEMA) flood insurance studies that had been previously completed for the Town and/or City in which the tributary is located.  The following is a list of flood insurance studies referenced for the development of the tributary hydraulic models: 

·         FEMA Flood Insurance Study, Town of Grand Island, NY, July 1979

·         FEMA Flood Insurance Study, City of North Tonawanda, NY, July 1981

·         FEMA Flood Insurance Study, City of Tonawanda, NY, February 1979

·         FEMA Flood Insurance Study, City of Niagara Falls, NY, September 1990

·         FEMA Flood Insurance Study, Town of Wheatfield, NY, May 1991 

The original FEMA flood insurance studies were completed between 1979 and 1990.  Since that time, bridges and other structures were found to have changed from the original surveys.  Accordingly, the geometry of several bridges and culverts were checked in October 2002 and August 2003.  The modifications found at the structures were noted, and modifications that were considered to be significant in terms of hydraulic influence had the changes incorporated into the HEC-RAS models. 

Another parameter needed for hydraulic modeling is a coefficient of channel roughness (Manning's n).  The roughness coefficients for this study were taken from the HEC-2 models used in the FEMA studies. 

Generally, for each of the tributaries in this study, the channel slope is low gradient.  This fact, and the fact that a non-flooding flow (median annual flow) was studied, will result in tributary water surface elevations that are controlled by a downstream water surface elevation (i.e., sub-critical flow).  Because the study tributaries discharge into the Niagara River or into a Niagara River tributary near its mouth with the Niagara River, the Niagara River water surface elevations influence the water levels in those tributaries.  Therefore, the starting water surface elevation in the hydraulic models for all but two study tributaries is an upper Niagara River water level. 

The starting downstream water elevations were determined from duration analyses of hourly water levels at several locations within the upper Niagara River (URS et al. 2005).  These water level statistics were utilized in this study to determine water surface elevations for the annual 0% (maximum elevation of record), 5%, 50%, 95%, and 100% (minimum elevation of record) exceedance values at the point where a tributary enters the upper Niagara River.  Those elevations were entered into the HEC-RAS models as the starting water surface elevation (Table 2.2-1).  For Tonawanda Creek and Spicer Creek, the water level statistics at the Tonawanda gauge were used.  For Cayuga Creek, Woods Creek, and Gun Creek, an interpolation between the Tonawanda and LaSalle gauges was used.  The starting elevation for Ellicott Creek is the water surface elevation predicted by the HEC-RAS model at the junction with Tonawanda Creek.  The starting elevation for Bergholtz Creek is the water surface elevation predicted by the HEC-RAS model at the junction with Cayuga Creek. 

The results of the hydraulic model analyses performed on the seven tributaries in this study are provided in Section 3.0.

 

Table 2.1-1

USGS Stream Gauges in the Western New York Region

Gauge Location

USGS Gauge Number

Drainage Area

Period of Record

Little Tonawanda Creek at Linden, NY

No. 04216500

22.1 square miles

1912-1992

Oatka Creek at Warsaw, NY

No. 04230380

39.1 square miles

1964-2001

Murder Creek near Akron, NY

No. 04217750

58.8 square miles

1982-1999

Tonawanda Creek at Attica, NY

No. 04216418

76.9 square miles

1977-2001

Ellicott Creek below Williamsville, NY

No. 04218518

81.6 square miles

1972-2001

Cayuga Creek near Lancaster, NY

No. 04215000

96.4 square miles

1938-2001

Cazenovia Creek at Ebenezer, NY

No. 04215500

135 square miles

1940-2001

Buffalo Creek at Gardenville, NY

No. 04214500

142 square miles

1938-2001

Tonawanda Creek at Batavia, NY

No. 04217000

171 square miles

1944-2001

Tonawanda Creek near Alabama, NY

No. 04217500

231 square miles

1955-1989

Tonawanda Creek at Rapids, NY

No. 04218000

349 square miles

1955-2001

Cattaraugus Creek at Gowanda, NY

No. 04213500

436 square miles

1939-2001

Genesee River at Portageville, NY

No. 04223000

984 square miles

1902-2001

 

Note:  The Period of Record indicates the range of years for which data was recorded at the gauge.  The record may not be continuous through this period (records may be missing due to gauge malfunction, maintenance, etc.).

 

Table 2.2-1

Starting Water Surface Elevations for Tributaries of the Upper Niagara River

Tributary

Name of Downstream Water Body, and Geographic Location of Tributary Mouth

Tributary Starting Water Surface Elevation (USLSD 1935)

Annual Minimum

Annual 95% Exceedance

Annual 50% Exceedance

Annual 5% Exceedance

Annual Maximum

Cayuga Creek

Upper Niagara River (Chippawa-Grass Island Pool), City of Niagara Falls

562.22

563.24

563.96

564.69

566.60

Bergholtz Creek

Cayuga Creek, City of Niagara Falls

562.22

563.24

563.96

564.69

566.60

Gun Creek

Upper Niagara River (Chippawa-Grass Island Pool), Grand Island

562.85

564.17

564.99

565.77

568.05

Woods Creek

Upper Niagara River (Chippawa-Grass Island Pool), Grand Island

562.09

563.05

563.75

564.47

566.30

Tonawanda Creek

Upper Niagara River  (Chippawa-Grass Island Pool), City of Tonawanda

563.37

564.92

565.82

566.65

569.23

Ellicott Creek

Tonawanda Creek, City of Tonawanda

563.37

564.92

565.82

566.65

569.23

Spicer Creek

Upper Niagara River (Chippawa-Grass Island Pool), Grand Island

563.37

564.92

565.82

566.65

569.23

Note:  The starting elevation for Bergholtz Creek is the water surface elevation predicted by the HEC-RAS model at the junction with Cayuga Creek.  The starting elevation for Ellicott Creek is the water surface elevation predicted by the HEC-RAS model at the junction with Tonawanda Creek.

 

Figure 2.1-1

USGS Stream Gauges in the Western New York Region

 

3.0     ANALYSIS

3.1         Hydrology

As explained in Section 2.1, the median (50% exceedance) annual flow was used for development of the water surface elevations in each tributary.  The estimated median annual flows for the tributaries are presented in Table 3.1-1.

3.2         Hydraulics

As explained in Section 2.2, five different upper Niagara River water surface elevations were selected to examine the potential extent those Niagara River elevations have on tributary water surface elevations under median annual flow conditions. 

This study considered water surface profile differences of three inches (0.25 feet) or less to be an indication that water surface elevations in a tributary are no longer controlled by the upper Niagara River elevation.  For profiles where this difference occurred between two cross-sections in the model, the upstream limit was conservatively assumed to be at the upstream cross-section.  Note that the flow in a tributary may change the upstream limit such that a flow different from the annual median flow may lengthen or shorten the reach influenced by upper Niagara River water levels. 

A summary of the estimated tributary reach lengths influenced by the upper Niagara River water levels at the median, 5% exceedence and annual maximum water surface elevations can be found in Table 3.2-1.  The median upper Niagara River water level is representative of daily water levels due to U.S./Canadian hydropower generation.  Prior analyses found that the annual maximum elevation of the upper Niagara River was not driven by project operations (URS et al. 2005) but instead by an extreme storm event.  It was analyzed because it represents the “worst case” or is a conservative estimate for geographically scoping other studies that are potentially impacted by water level fluctuations in the upper Niagara River.   The 5% exceedance elevation, which is partially driven by project operations, is included in Table 3.2-1.  The following sections will present additional information on the analyses and results of the investigation for each of the tributaries studied in this report. 

3.2.1        Woods Creek and Woods Creek Tributaries

Woods Creek has three main branches identified as Tributary No. 1, No. 2, and Tributary No. 3.  Because hydraulic models did not pre-exist for Tributary No. 2, that tributary was not selected for this study. 

Woods Creek has an estimated median annual flow of 5.9 cfs at its confluence with the upper Niagara River.  The median flow reduces to 3.6 cfs upstream of Woods Creek's confluence with Tributary No. 1, and to 3.3 cfs upstream of Tributary No. 3. 

Figures 3.2.1-1 and 3.2.1-2 illustrate the stream reaches for Woods Creek and Tributary No. 1 that are influenced by upper Niagara River water surface elevations.  Water surface profiles for Woods Creek and Tributary No. 1 are provided on Figures 3.2.1-3 and 3.2.1-4, respectively. 

The extent that upper Niagara River water levels affect Woods Creek water levels varies for the median annual flow.  For the annual minimum water level condition, the extent of influence is 6,400 feet as compared to 11,000 feet for the annual maximum water level condition.  The creek length of approximately 11,000 feet upstream of the mouth corresponds to a location that is between Baseline and Stony Point Roads and is downstream of Tributary No. 3.  Tributary No. 1 enters Woods Creek approximately 6,900 feet upstream of the upper Niagara River.  Water levels in the upper Niagara River were found to influence Woods Creek Tributary No. 1 water levels for approximately 2,400 feet upstream of the tributary's confluence with the mainstem of Woods Creek.  If the annual maximum Niagara River water level was excluded, the affected length in the Woods Creek mainstem would be reduced to approximately 9,000 feet for Woods Creek (as opposed to 11,000 feet), and 1,300 feet for Tributary No.1 (as opposed to 2,400 feet).

Tributary No. 2 enters Woods Creek approximately 9,000 feet upstream of the upper Niagara River.  As noted above, Tributary No. 2 was not analyzed in this study.  However, from the Woods Creek analysis it appears that Tributary No. 2 will be influenced by water levels of the upper Niagara River for Niagara River water levels higher than the annual minimum.  Tributary No. 3 enters Woods Creek upstream of the location where upper Niagara River water levels were found to influence Woods Creek.  Therefore, Tributary No. 3 is not influenced by upper Niagara River water levels. 

3.2.2        Gun Creek

Figure 3.2.2-1 illustrates Gun Creek tributary reaches that are influenced by the different upper Niagara River water surface elevations.  Water surface profiles are provided on Figure 3.2.2-2. 

Gun Creek is located in the Town of Grand Island and has an estimated median annual flow of 2.7 cfs at the mouth.  For the median annual flow and all five upper Niagara River water level conditions, the hydraulic analysis indicates that Niagara River water levels could influence Gun Creek water levels for approximately 4,100 feet upstream of the mouth, a point downstream of the Ransom Road crossing.  The upstream influence of Niagara River water level fluctuations on Gun Creek water levels is limited by a steep profile in Gun Creek between 2,900 feet and 4,200 feet upstream of the mouth.

3.2.3        Spicer Creek

Figure 3.2.3-1 illustrates the reaches of Spicer Creek that are influenced by upper Niagara River water surface elevations.  Water surface profiles are provided on Figure 3.2.3-2. 

Spicer Creek is located in the Town of Grand Island.  The creek has an estimated median annual flow of 2.4 cfs at the mouth.  At the median flow, the length of Spicer Creek influenced by upper Niagara River water level varies from a short distance of 600 feet for the annual minimum condition to 4,600 feet for the annual maximum condition.  The hydraulic model estimated for the median and the 5% exceedance upper Niagara River water levels that the affected Spicer Creek reach had a length of approximately 2,600 feet.  This distance is slightly more than half of the maximum extent of 4,600 feet influenced by the annual maximum upper Niagara River water level, a location that is near Whitehaven Road. 

3.2.4        Cayuga Creek

Figure 3.2.4-1 illustrates the reaches of Cayuga Creek that are influenced by upper Niagara River water surface elevations.  Water surface profiles are provided on Figure 3.2.4-2. 

The mouth of Cayuga Creek is located in the City of Niagara Falls.  The creek also flows in the Towns of Niagara, Wheatfield, and Lewiston.  Cayuga Creek's largest tributary, Bergholtz Creek, was also studied (see Section 3.2.5) and enters Cayuga Creek approximately 5,600 feet upstream of the upper Niagara River. 

Cayuga Creek has an estimated median annual median flow of 27.5 cfs at its confluence with the Niagara River and an estimated median annual flow of 10.7 cfs upstream of the Bergholtz Creek confluence. 

At the median annual flow rate, the extent of influence varies depending on the upper Niagara River water level.  For the annual minimum condition, the extent of influence is 7,000 feet as compared to 10,100 feet for the annual maximum condition, a location near the Juron Drive and Lozina Drive intersection.  The hydraulic analysis found the influence was up to 7,400 feet of Cayuga Creek for the median water level condition.

3.2.5        Bergholtz Creek

Figure 3.2.5-1 illustrates the reaches of Bergholtz Creek that are influenced by upper Niagara River water surface elevations.  Water surface profiles are provided on Figure 3.2.5-2.

Bergholtz Creek enters Cayuga Creek approximately 5,600 feet upstream of Cayuga Creek's confluence with the upper Niagara River, or just downstream of the Cayuga Drive Bridge in the City of Niagara Falls.  Reaches of the Bergholtz Creek are also located in the Towns of Wheatfield and Cambria, Niagara County, New York.  The estimated median annual flow for Bergholtz Creek is 17.7 cfs. 

Since the mouth of Bergholtz Creek is within the reach of Cayuga Creek influenced by upper Niagara River water levels, it too is affected by upper Niagara River water levels.  The analysis for Bergholtz Creek at the median annual flow rate shows that the length of Bergholtz Creek influenced varies from a distance of 6,800 feet for the annual minimum upper Niagara River water level condition to 10,900 feet for the annual maximum upper Niagara River water level condition.  For the median and the 5% exceedance upper Niagara River water levels, the affected Bergholtz Creek reach had a length of approximately 9,000 feet.  The length of approximately 10,900 feet would be near to, or just downstream of, the Walmore Road bridge crossing. 

3.2.6        Tonawanda Creek

Figure 3.2.6-1 illustrates the study reach for Tonawanda Creek.  Water surface profiles are provided on Figure 3.2.6-2.

The mouth of Tonawanda Creek is located in the City of Tonawanda and the City of North Tonawanda.  The reach of the creek considered in this study ends approximately 10,570 feet upstream of the upper Niagara River, a location that is near Mayors Park and Dematteo Drive in the City of North Tonawanda.  The entire reach of study for Tonawanda Creek is part of the Erie (Barge) Canal system. 

The median annual flow for Tonawanda Creek is estimated to be 401 cfs.  With that flow rate, the upper Niagara River was found to influence all 10,570 feet of the study reach.  Therefore, the upstream extent of the Niagara River’s influence on water levels in Tonawanda Creek was not determined from the hydraulic model developed in this study. 

Because much of Tonawanda Creek is part of the Erie (Barge) Canal System, the upstream extent of influence may extend for several miles upstream of the Niagara River.  Tonawanda Creek, from the Niagara River to Pendleton, a distance of approximately 11.6 miles, has been altered by the dredging and diversions necessary for canal operations.  These alterations change this section of the creek to a regulated elevation, standard width and a flat hydraulic slope.  Based on an engineering analysis of the creek profile, it appears that the dredged channel may cause the extent of the upper Niagara River influence to be approximately 13.7 miles for the median Niagara River water level which is representative of water levels due to U.S./Canadian hydropower generation.  This estimate seems reasonable because the creek at this upstream location is more riverine in character (as compared to sections of the Barge Canal) and is near two riffles which act as hydraulic controls limiting the upstream influence of Niagara River water level fluctuations due to U.S./Canadian hydropower generation on Tonawanda Creek water levels.  The two riffles are located 13.6 and 14.1 miles upstream of the mouth (Gomez and Sullivan and E/PRO 2005).  For the annual maximum water level which resulted from extreme storm conditions, the influence extends nearly 19 miles upstream of the upper Niagara River.  The affected creek lengths shown in Table 3.2-1 for various upper Niagara River water levels were estimated by extending the various water levels of the upper Niagara River in Tonawanda to where they intersect the stream bottom of Tonawanda Creek.  The stream bottom profile for this analysis was taken from Flood Insurance Studies for the towns and communities that are intersected by the creek. 

Another important item to consider for Tonawanda Creek is that during the canal operational period (generally the first week of May through the last week of October) when the guard gate in Lockport is opened, flow in the creek often moves in an "upstream" (easterly) direction.  Up to 1,100 cfs are diverted from the Niagara River for canal operations.  The hydraulic and hydrologic alterations to Tonawanda Creek therefore cause the upper Niagara River's influence to relocate, depending on flow direction and water level controls for the canal. 

3.2.7        Ellicott Creek

Figure 3.2.7-1 illustrates the reach of study for Ellicott Creek.  Water surface profiles are provided on Figure 3.2.7-2. 

Ellicott Creek is a tributary of Tonawanda Creek and enters Tonawanda Creek approximately 1,600 feet upstream of the upper Niagara River.  The reach of Ellicott Creek considered for this study ends approximately 11,000 feet upstream of the Tonawanda Creek confluence at a location near the Colvin Boulevard bridge crossing in the Town of Tonawanda. 

The median average daily flow for Ellicott Creek was estimated to be 100 cfs.  With that flow rate, the upper Niagara River was found to influence all 11,000 feet of study reach.  Therefore, the upstream extent of the Niagara River’s influence on water levels in Ellicott Creek was not determined from the hydraulic model developed in this study. 

The analysis of Ellicott Creek found that flood control efforts have also changed the hydraulic and hydrologic characteristics of the creek.  Similar to the situation in Tonawanda Creek, dredging and flow diversions have changed Ellicott Creek to a condition that could make the upper Niagara River water levels influence approximately 7 miles of the creek (see Table 3.2-1).  This distance coincides with a riffle that extends from 6.9 to 7.1 miles from the mouth, and for which field observations confirmed was a hydraulic control limiting the upstream influence of Niagara River water level fluctuations due to U.S./Canadian hydropower generation on creek water levels (Gomez and Sullivan and E/PRO 2005).  These estimates were determined by extending the median and annual maximum water levels of the Niagara River to where they intersect the stream bottom of Ellicott Creek.  The stream bottom profile was taken from Flood Insurance Studies for communities that lie along the creek.

 

Table 3.1-1

Estimated Annual Median Flow Calculated for Study Tributaries

Study Tributary

Drainage Area (square miles)

Average Annual Precipitation (inches/year)1

Flow Calculation Method

Estimated Annual Median Flow (cfs)

Woods Creek at Mouth

7.50

36.50

Regression Method

5.9

Woods Creek Tributary 1 at Confluence with Woods Creek

2.81

36.50

Regression Method

2.3

Woods Creek Tributary 3 at Confluence with Woods Creek

0.30

36.50

Regression Method

0.3

Spicer Creek at Mouth

2.97

36.50

Regression Method

2.4

Gun Creek at Mouth

3.28

36.50

Regression Method

2.7

Cayuga Creek at Mouth

38.6

36.25

Regression Method

27.5

Bergholtz Creek at Confluence with Cayuga Creek

24.3

36.25

Regression Method

17.7

Ellicott Creek at Confluence with Tonawanda Creek

110

37.00

Drainage Area Ratio Method

100

Tonawanda Creek at Mouth

635

37.00

Drainage Area Ratio Method

401

1 Source:  USDA 1998.

 

Table 3.2-1

Estimated Tributary Reach Length Influenced by Upper Niagara River Water Surface Elevations

Study Tributary

Estimated Annual Median Flow (cfs)

Approximate Study Length of Tributary

Estimated Length of Tributary Influenced by Upper Niagara River (to nearest 100 feet)

At Median Niagara River Water Surface Elevations

At 5% Exceedance Niagara River Water Surface Elevations

At Maximum Niagara River Water Surface Elevations

Woods Creek

5.9

16,170 feet

9,000 feet

9,000 feet

11,000 feet

Woods Creek - Tributary 1

2.3

14,355 feet

1,300 feet

1,300 feet

2,400 feet

Woods Creek -Tributary 3

0.3

8,920 feet

No Influence

No Influence

No Influence

Spicer Creek

2.4

12,680 feet

2,600 feet

2,600 feet

4,600 feet

Gun Creek

2.7

7,460 feet

4,100 feet

4,100 feet

4,100 feet

Cayuga Creek

27.5

14,690 feet

8,300 feet

9,700 feet

10,100 feet

Bergholtz Creek

17.7

16,420 feet

9,000 feet

9,000 feet

10,900 feet

Ellicott Creek

77.2

11,000 feet

7.0 miles1

7.0 miles1

7.1miles 1

Tonawanda Creek

408.2

10,570 feet

13.7 miles1

18.8 miles1

18.9 miles1

1Based on engineering judgment from information provided in Flood Insurance Studies on the tributary profiles and the upper Niagara River water surface elevations

 

Figure 3.2.1-1

Woods Creek - Stream Reaches Influenced by Water Levels in Upper Niagara River

 

Figure 3.2.1-2

Woods Creek Tributary No. 1 – Stream Reaches Influenced by Water Levels in Upper Niagara River

 

Figure 3.2.1-3

Woods Creek – Water Surface Profile

 

Figure 3.2.1-4

Woods Creek Tributary No. 1 – Water Surface Profile

 

Figure 3.2.2-1

Gun Creek – Stream Reaches Influenced by Water Levels in Upper Niagara River

 

Figure 3.2.2-2

Gun Creek – Water Surface Profile

 

Figure 3.2.3-1

Spicer Creek – Stream Reaches Influenced by Water Levels in Upper Niagara River

 

Figure 3.2.3-2

Spicer Creek – Water Surface Profile

 

Figure 3.2.4-1

Cayuga Creek – Stream Reaches Influenced by Water Levels in Upper Niagara River

 

Figure 3.2.4-2

Cayuga Creek – Water Surface Profile

 

Figure 3.2.5-1

Bergholtz Creek – Stream Reaches Influenced by Water Levels in Upper Niagara River

 

Figure 3.2.5-2

Bergholtz Creek – Water Surface Profile

 

Figure 3.2.6-1

Tonawanda Creek – Stream Reaches Influenced by Water Levels in Upper Niagara River

 

Figure 3.2.6-2

Tonawanda Creek – Water Surface Profile

 

Figure 3.2.7-1

Ellicott Creek – Stream Reaches Influenced by Water Levels in Upper Niagara River

 

Figure 3.2.7-2

Ellicott Creek - Water Surface Profile

 

4.0     SUMMARY

This study found that water surface elevations of the upper Niagara River will hydraulically influence the water surface elevations of the Niagara River tributaries.  Of the tributaries investigated in this study, those found to be influenced by Niagara River water levels are:  Woods Creek, Woods Creek Tributary No. 1, Gun Creek, Spicer Creek, Cayuga Creek, Bergholtz Creek, Tonawanda Creek, and Ellicott Creek. 

A tributary flow rate equal to the annual median flow was analyzed for the tributaries in this study.  The median annual flow rate was selected because it was deemed representative of the tributary flow over the year.  However, tributary flows can vary significantly over the year and the extent of the reach influenced by upper Niagara River elevations may change according to those flow changes. 

It should be noted that this study investigated a wide range of water levels, from the minimum to the maximum-recorded water levels in the upper Niagara River, a difference that spans between four to six feet.  Such a range in elevation does not represent a typical upper Niagara River daily water level fluctuation of 1.5 feet, or less.  Because a wide range of Niagara River elevations were investigated, the reach length of streams influenced by the upper Niagara River elevations are considered to be very conservative estimates for a median flow event.  Furthermore, because the annual maximum upper Niagara River water surface elevation in this study is driven by unusual natural events, and not hydropower project operations, it is important to focus on the influence of the other Niagara River water surface elevations studied in this report, such as the median water surface elevation which is more representative of the range of water level due to U.S./Canadian hydropower generation. 

It is important to note that dredging and water diversions, for the purposes of navigation on the Erie(Barge) Canal and flood control, have occurred in both Tonawanda and Ellicott Creeks.  Such operations have altered the hydraulics and hydrology in the tributaries and the relationship of the tributaries to the upper Niagara River.  An additional factor for Tonawanda Creek is that 11.6 miles of the tributary are part of the Erie (Barge) Canal system and water levels are therefore influenced by canal operations.   In order to maintain a uniform canal width and depth and flat slope, the Tonawanda Creek channel has been dredged and maintained.  During the canal operational period (generally the first week of May through the last week of October) when the guard lock gate in Lockport is opened, flow in the creek often moves in an upstream (easterly) direction because the level of the Erie (Barge) Canal at Lockport is lower than that of the Niagara River at the City of Tonawanda.  Up to 1100 cfs of flow are diverted from the Niagara River for canal operations. 

The analysis of Ellicott Creek found that flood control and dredging operations have also changed the hydraulic and hydrologic characteristics of the creek. Both the U.S. Army Corps of Engineers (USACE) and local government entities have made channel improvements consisting of deepening and widening the stream channel.  In the 1930’s Ellicott Creek was enlarged and a dam (water control structure) was built at Island Park 2.4 miles upstream of the United States Geological Survey (USGS) gauge near the Village of Williamsville to help control flooding.  Regulation occurs today by the seasonal manipulation of that dam and by intermittent pumping from stone quarries into stream.  In 1965, Erie County also completed construction of a diversion channel in Ellicott Creek Park in the Town of Tonawanda, from Ellicott Creek to Tonawanda Creek for the purpose of controlling floods upstream in the Town of Amherst. The USACE has constructed the North Diversion Channel, Pfohl Park Diversion Channel and Upper Diversion Channel along Ellicott Creek through Amherst and has completed major improvements to the stream channel (FEMA 1992).

Accordingly, water levels in Tonawanda Creek and Ellicott Creek are not solely influenced by upper Niagara River water levels.  For Tonawanda Creek and Ellicott Creeks, the upstream extent of the Niagara River’s influence was not determined by the hydraulic models in this study.    The extent of the Niagara River’s influence on the water levels in these two creeks may extend for several miles upstream of the Niagara River. 

Water level data collected by NYPA in 2003 for one location on Tonawanda Creek and three locations on Ellicott Creek (see Figure 4.0-1) confirm that the Niagara River’s influence extends beyond that of the study reaches.  The similarity observed between the pattern of daily water level fluctuation at the station on Tonawanda Creek, located 4-miles from the mouth, and the pattern at the permanent gauge station on the Niagara River located near the mouth is indicative that there is an influence from the Niagara River at that point.  A similar finding was observed for Ellicott Creek where the farthest upstream station monitoring water level was located 3.3-miles upstream of its mouth (URS and Gomez and Sullivan 2005). 

Dredging and water diversions for navigation and flood control may cause the extent of influence of the median Niagara River water level to be 13.7 miles for Tonawanda Creek and 7 miles for Ellicott Creek, based on engineering judgment.  (The estimates for Tonawanda and Ellicott Creeks were determined by extending the median water level of the Niagara River to where it intersects the stream bottom of the creeks.  The stream bottom profiles were obtained from Flood Insurance Studies for towns and communities that are intersected by the creeks.) 

 

Figure 4.0-1

Water Level Gauge Locations for Tonawanda and Ellicott Creeks

 

REFERENCES

R1019215231 \ Text Reference: FEMA 1992 \ Federal Emergency Management Agency.  1992.  Flood Insurance Study: Town of Amherst, New York. 

R1019216357 \ Text Reference: Gomez and Sullivan and E/PRO 2005 \ Gomez and Sullivan Engineers, P.C., and E/PRO Engineering & Environmental Consulting, LLC.  2005.  Mapping of Aquatic and Riparian Habitats of Ellicott and Tonawanda Creeks, and Tributaries to Tonawanda Creek.  Prep. for the New York Power Authority. 

R1019216396 \ Text Reference: Hirsch 1979 \ Hirsch, R.M.  1979.  An evaluation of some record reconstruction techniques.  Water Resources Research 15(6):1780-81.

R1019216397 \ Text Reference: Hirsch 1982 \ Hirsch, R.M.  1982.  A comparison of four stream flow record extension techniques.  Water Resources Research 18(4):1081-88.

R1019215951 \ Text Reference: Maidment 1992 \ Maidment, David R.  1992.  Handbook of Hydrology.  McGraw-Hill.

R1019215989 \ Text Reference: USDA 1998 \ U.S. Department of Agriculture.  1998.  New York Average Annual Precipitation, 1961-1990.  Natural Resources Conservation Service. 

R1019215958 \ Text Reference: URS and Gomez and Sullivan 2005 \ URS Corporation and Gomez and Sullivan Engineers, P.C.  2005.  Surface Water Quality of the Niagara River and its U.S. Tributaries, prep. for the New York Power Authority. 

R1019215119 \ Text Reference: URS et al. 2005 \ URS Corporation, Gomez and Sullivan Engineers, P.C., and E/PRO Environmental & Engineering Consulting, LLC.  2005.  Niagara River Water Level and Flow Fluctuation Study, prep. for the New York Power Authority.

 

 

APPENDIX A

Methodology for Development of Hydrologic Data

This appendix provides additional information on the engineering methods used to develop hydrology estimates for ungauged locations in the upper Niagara River Tributary Backwater Study (Issue #23).  Specifically, information is provided for the following topics:

·         Gauge record extension/augmentation using Maintenance of Variance Extension (MOVE) techniques;

·         Multiple regression analysis for the development of median annual flow estimates.

Record Extension Technique (MOVE)

The task began with a review of USGS flow gauges in the western New York region (Table A-1).  This review found thirteen gauges with record lengths and watershed characteristics suitable to estimate median annual flows for ungauged tributaries.  All the gauge data used in this study were from USGS average daily stream flow records with a minimum of 19 years of data at the gauge location. 

Because the stream flow records did not have similar calendar dates, and/or have missing records due to equipment failure and inaccurate data records, an information transfer technique known as maintenance of variance extension, or MOVE, was applied to develop relationships among the gauges and eventually augment the data so that all thirteen gauges had similar period of records (Maidment 1992). 

Maidment (1992) describes the use of record extension methods as follows:

In conducting studies of the probability of water-supply shortage in the reservoir, one would like to be able to simulate operations over some long hydrologic record.  Suppose n were 5 years and N were 100 years and the events of interest (extreme, long-term droughts) may be 0.05 or 0.1 probability events.  It is clear that having a 100-year record would provide for much more accurate results than could be achieved with only 5 years of record.

Thus one wishes to exploit the correlation between X (nearby stream with a very long record, Xi, i = 1, 2, . . . , N) and Y (reservoir inflow gauge near the location of study with a limited record, Yi, i =  1, 2, . . . , n) to extend the length of the Y record by estimating a set of N - n flows for the period during which there was an X record but no Y record.  The obvious approach to this problem is to use regression to estimate Y as a function of X.  If the relationship of X and Y meets the assumptions for regression (linear and homoscedastic) then the regression estimate of Yi is the lowest-variance estimator of Yi.  However, it is not a particular value of Yi that is important, but rather the full collection of estimates of Y.  What is needed is a set of Y values that possess the correct statistical properties of Y.  The properties include the mean and variance of Y and may also include its serial correlation structure. 

The procedure used for this analysis is a class of methods known as maintenance of variance extension, or MOVE.  Pioneered by Hirsch (1979, 1982), MOVE procedures are used by hydrologists to extend the length of stream flow records, while maintaining the distribution shape and serial correlation structure of two gauging stations found in similar hydrologic settings 1. 

Multiple Regression Analysis For the Development of Median Annual Flow Estimates

Once the record extension technique was used to develop similar periods of records for the local streamflow data, analyses were conducted to develop relationships for the gauged tributaries.  In hydrology, a well-established and physically plausible relationship can frequently be made between the dependent variable of flow rate and the independent variable of drainage area.  For the thirteen flow gauges in this analysis, a scatter plot of the median annual flow rate versus the drainage area for the site shows a linear relationship may exist between the median annual flow and drainage area (Figure A-1).  Although it can be stated that drainage area is the principal explanatory variable to define median annual flow for the ungauged tributaries, the variation in the dependant variable (median annual flow) can sometimes be further defined with the use of other variables applicable to the watershed.  Starting with a list of these explanatory variables, a series of tests were performed to find those variables that applied to the model results.  It was found that median annual flow in a tributary could be best estimated with a function that incorporated two drainage basin characteristics:  drainage area and annual average precipitation. 

The power function took the following form: 

 

 

Median Annual Flow: Q = c DA PB

where:        c = Regression Coefficient

                D = Drainage area, in square miles

                A = Power Factor to raise Drainage Area

                P = Average Annual Precipitation, in inches

                B = Power Factor to raise Average Annual Precipitation

The derivation of the above equation utilized computer models to evaluate the regression coefficients and power factors needed to yield the highest possible coefficient of multiple determination, or R2.  The value of R2 represents the proportion of the total variation in the values of median annual flow that can be accounted for or explained by the above power function.  For example, a R2 value of 0.92 would indicate that 92% of the variation in median annual flow is explained by the variables in the power function.  A perfect R2 value of 1.0 would indicate that all the variation in median annual flow is explained by the variation in drainage area and precipitation. 

The analysis performed on the power function found the following coefficients would result in the highest R2 value (R2 = 0.9833): 

Regression Coefficient, c = 0.001

Power Factor to raise Drainage Area , A = 0.95

Power Factor to raise Average Annual Precipitation, B = 1.88

The above values were applied to the power equation.  Thereafter, the average annual median flow for the ungauged tributaries to the upper Niagara River could be estimated with the applicable independent variables (ungauged tributary's drainage area and annual precipitation).

 

Table A-1

USGS Gauges in the Western New York Region

USGS Gauge Number

Gauge Name

Median Annual Flow1

Drainage Area (square miles)

Average Annual Precipitation (inches)

04216500

LITTLE TONAWANDA CREEK AT LINDEN NY

9.6

22.1

35.0

04230380

OATKA CREEK AT WARSAW NY

24.7

39.1

35.0

04217750

MURDER CREEK NEAR AKRON NY

25.2

58.8

36.0

04216418

TONAWANDA CREEK AT ATTICA NY

50.6

76.9

36.5

04218518

ELLICOTT CREEK BELOW WILLIAMSVILLE NY

60.0

81.6

35.0

04215000

CAYUGA CREEK NR LANCASTER NY

43.0

96.4

36.0

04215500

CAZENOVIA CREEK AT EBENEZER NY

92.0

135.0

39.0

04214500

BUFFALO CREEK AT GARDENVILLE NY

82.0

142.0

38.0

04217000

TONAWANDA CREEK AT BATAVIA NY

88.6

171.0

35.5

04217500

TONAWANDA CREEK NEAR ALABAMA NY

120.3

231.0

35.0

04218000

TONAWANDA CREEK AT RAPIDS NY

175.0

349.0

34.5

04213500

CATTARAUGUS CREEK AT GOWANDA NY

409.9

436.0

41.5

04223000

GENESEE RIVER AT PORTAGEVILLE NY

602.0

984.0

36.5

 

1 Based on adjusted records

 

Figure A-1

Scatter Plot of Drainage Area vs. Median Annual Runoff