Estimating Daily Streamlow at Ungaged Sites in NH: Negative - - PowerPoint PPT Presentation

Estimating Daily Streamlow at Ungaged Sites in NH: Negative Run-length Assessment

Neil M. Fennessey HYSR/Hydrologic Services

HYSRConsult@gmail.com

NH Department of Environmental Services Concord, NH August 27, 2018

Estimating Daily Flows at Ungaged Sites

 NH DES needs to determine PIFs at ungaged sites

in presently and future designated rivers

 NHDES needs a reliable way to estimate daily

streamflow time series at these ungaged sites

 The QPPQ Transform method is one approach  Conceived and developed by Fennessey (1994)  Has been used in the northeast and central US,

Canada and South Africa

Other Ways to Estimate Daily Flows at Ungaged Sites: Pluses & Minuses

 Watershed Area Ratio (WAR) method  Very simple to use  Assumes that all watersheds are exactly alike

except for the size of the watershed

 Rainfall-Runoff Models (HSPF, Sacramento Soil

Moisture Accounting Model)

 Physically based but parameter-intensive  Used by NOAA for flood forecasting  Requires long stream gage period-of-record for

site calibration and validation

 The QPPQ Transform method has been adopted

• r is being considered by several US Geological

Survey district offices:

 Massachusetts and Rhode Island USGS  Pennsylvania USGS  Iowa USGS  Minnesota USGS  New York USGS  New Hampshire and Vermont USGS

Does USGS Like QPPQ Transform?

 The QPPQ Transform method was extensively

tested by the USGS and bested 17 alternative methods for estimating daily flow time series at 182 gaged watershed sites located in

 West Virginia  Tennessee  North Carolina  South Carolina  Georgia  Alabama  Florida

Has the QPPQ Been Extensively Tested?

 The QPPQ Transform method has been adopted

by several state agencies:

 Massachusetts EOEEA  Massachusetts DEP Water Management Act

program

 Sustainable Water Management Initiative  The Delaware River Basin Authority  Maryland Dept. of the Environment  Pennsylvania Dept. of Environmental Protection  New York State Energy R & D Authority  Minnesota Pollution Control Agency

Have States or Agencies Adopted the QPPQ?

 The QPPQ Transform method has been used for  Many surface water reservoir system

assessments in Massachusetts and Connecticut

 River basin and watershed studies in

Massachusetts and Connecticut

 Systems analysis study of the Connecticut

River watershed

 Complex litigation in Connecticut

Examples of Local Applications

The QPPQ Transform Method

Four Steps to the QPPQ Transform Method

• 1. Chose a USGS Streamgage Site for the time

series: Qgage(t)

• 2. Use that time series to construct a streamflow

duration curve (FDC): Qgage(p)

• 3. Construct a FDC at the ungaged site using its

watershed’s soil characteristics, climate, and topography and a regional FDC model: Qungaged(p)

• 4. Generate a streamflow time series at the

ungaged site: Qungaged(t)

Minnesota

USGS

(Lorenz and Zigeweid, 2016)

Use Qgage(t) to Construct ‘observed’ FDC Qgage (p)

 Rank sort Qgage(t) “n” days from largest to smallest

Q(1) ≥ Q(2) ≥ Q(3) … Q(n-2) ≥ Q(n-1) ≥ Q(n)

POR 30 years x 365 = 10,950 days P[Q≥Q(n)]x100 =10,950/10951 x 100 = 99.991

 “i” is the rank-order i=1,n  “n” is number of days in the period of record  Use “plotting position” formula to estimate

exceedance probability p=P[Q≥q]

Step 1: Qgage(t) and Step 2: Qgage(p)

 Common factor between the flow date, time, and

the FDC probability, p, is Q = a 3-D relationship

 Knowing Q gage (t) & p(Qgage) => p(t)

Step 3: Regional FDC Model

 Construct a FDC for the ungaged site without

historic daily streamflow data

HYSR Phase 1 Study for NH DES: Update Regional FDC Model

 Fennessey (2018) developed a special

streamgage network of 133 USGS gaged watersheds in New England, NY, PA and NJ.

 Daily streamflow data: minimum 20-year POR in

1950-1990 base period.

 Research confirmed earlier Fennessey (1994)

finding that the 3-parameter Generalize Pareto (GPA) was the best among alternatives.

 L-moments used to estimate the 3 GPA

parameters: ζ, α, κ for each of 133 HYSR sites

 Gather watershed specific data for each of these

133 USGS gage sites in the HYSR network

 Use each of the 3 GPA parameters statistically

determined for each HYSR watershed from obs streamgage data as dependent variables

 Use OLS multivariate regression and each

watershed’s specific soil, climate, and topography data as candidate independent variables

 Keep only “statistically significant” soil, climate,

and topographic variables to create one equation for each GPA parameter; ζ, α and κ .

 To construct a FDC at the ungaged site, the analyst

must determine ten soil, climate, and topography characteristics for that watershed:

 Watershed area (mi2)  Avg annual precipitation (in/yr)  Avg annual temperature (oF)  Potential maximum soil moisture retention (in)  Percent of watershed area covered by HSG C soil (%)  Avg watershed elevation (ft)  Avg watershed aspect (degrees) 0-360O  Percent of watershed area covered by lakes, ponds and

reservoirs (%)

 Percent of watershed area covered by impervious surface (%)  Slope of main stream channel (ft/mile)

Compare FDCs and Assess “Goodness-of-Fit” HYSR Phase 1 (Fennessey, 2018)

 Construct “observed,” “fitted,” and regional “model”

FDCs at eleven NH USGS gages sites, watershed areas range: 12.1 mi2 – 622 mi2

 Graph all three FDCs for visual comparison  Estimate the Bias between the “observed” FDC and

the “fitted” FDC to see how suitable the GPA is as basis for the Reginal FDC Model

 Estimate the Bias between the “observed” FDC and

the regional “model” FDC to see how well it compared with the “fitted” FDC.

Baker River, near Rumney, NH

QPPQ Transform Steps 1 - 3

 Assume for Steps 2 and 3: p(Qungaged) = p(Qgaged)

Step 4: Generate Qungaged(t)

 Recall 3-D relationship between t, Q and p => p(t)  Use regional FDC model with p(t) instead of just p

HYSR Phase 2 Study for NHDES Negative Run-length QPPQ Transform Assessment

 In the Phase 2 study, HYSR focused on a negative

run-length analysis of daily flows during bioperiods determined by fishery and other specialists at five USGS streamgage sites in New Hampshire

 HYSR and NHDES reasoned that since Protected

Instream Flow rates, PIFs, are determined in part by sub-threshold event durations, a negative run- length analysis would be a good way to further assess the QPPQ Transform method for use in New Hampshire

A Negative Run-length

 A negative run-length event occurs when

streamflow, Q(t), falls below a designated threshold, QP, for one or more consecutive days.

Bioperiods

 The University of New Hampshire et al (2007) study

determined six PIF bioperiods for the Souhegan R.

No. Bioperiod Start End 1 Over-Wintering 15-Nov 28-Feb 2 Spring Flood 1-Mar 30-Apr 3 Shad Spawning 1-May 14-Jun 4 GRAF Spawning 15-Jun 14-Jul 5 Rearing & Growth 15-Jul 30-Sep 6 Salmon Spawning 1-Oct 14-Nov

 The participants of the New Hampshire et al (2007)

Souhegan River watershed study developed PIFs for each bioperiod

 HYSR used two of these PIFs for the present study

as negative run-length thresholds: Qcritical and Qrare.

 HYSR used Q85 and Q95 for the Diamond, Saco,

Oyster and Pemigewasset Rivers because these watersheds do not yet have PIF flow rates.

 Of particular interest to NHDES is the Rearing &

Growth bioperiod which begins mid-summer and ends in early fall, when yearly flows are usually the lowest and the aquatic ecosystem quite stressed.

 During the 1963-1968 Rearing & Growth bioperiod,

flows fell below Qcritical (yellow) and Qrare (red) for long runs of time.

Q (cfs)

1 10 100 1 10 100

Year

1964 1966 1968

 During all Rearing & Growth days of flow over the

entire POR, the FDC shows what fraction of time. Qcritical ≈ Q75,rearing & growth and Qrare ≈ Q90,rearing & growth

 By counting the number of nday-long negative run-

length events one constructs a frequency histogram for sub-Qcritical events during Rearing & Growth.

 Similarly, one constructs a frequency histogram for

sub-Qrare events that occurred over the POR during the 77 day-long Rearing & Growth bioperiod.

 If one wants to estimate the likelihood that a

negative run-length of nday or more given that flows are sub-Qcritical, one constructs a probability plot.

 Similarly, one may construct a probability plot for the

negative run-length of nday or more given that flows are sub-Qrare. Called a Run Duration Curve (RDC)

Further QPPQ Transform Assessment

 The QPPQ Transform is assessed by comparing

negative run-length, season-specific FDCs, frequency histograms and probability plot RDCs.

 Souhegan River assessment used Qcritical and Qrare

as event thresholds

 Diamond, Saco, Oyster and Pemigewasset River

assessments used Q85 and Q95 as event thresholds

 Tests at these five sites were run using two different

Index gages from the Fennessey (2018) 133 gage network as needed for Step 1 and 2 of the QPPQ Transform: HCDN NN and HCDN NN WA.

 The Obs. and QPPQ Transform HCDN NN Rearing

& Growth bioperiod FDC

 The Obs. and QPPQ Transform HCDN NN WA

Rearing & Growth bioperiod FDC

 The Obs. and QPPQ Transform HCDN NN and

HCDN NN WA Rearing & Growth bioperiod negative run-length histograms for Qcritical threshold HCDN NN HCDN NN WA

 The Obs. and QPPQ Transform HCDN NN and

HCDN NN WA Rearing & Growth bioperiod negative run-length histograms for Qrare threshold HCDN NN HCDN NN WA

 The Obs. and QPPQ Transform HCDN NN and

HCDN NN WA Rearing & Growth bioperiod negative run-length RDCs for Qcritical threshold HCDN NN HCDN NN WA

 The Obs. and QPPQ Transform HCDN NN and

HCDN NN WA Rearing & Growth bioperiod negative run-length RDCs for Qrare threshold HCDN NN HCDN NN WA

Still More QPPQ Transform Assessments

 The QPPQ Transform is further assessed by

comparing:

 the bioperiod specific total number of negative

run-length events over the entire POR

 the 95% confidence interval of the mean duration,

E[durat] of a negative run-length event

 the coefficient of determination, R2 between the

POR and season-specific obs. daily data and specific QPPQ transform Index gage daily flow pair for both HCDN NN and HCDN NN WA

 The Obs. and QPPQ Transform HCDN NN and

HCDN NN WA Rearing & Growth bioperiod number

• f negative run-length events for both Qcritical and

Qrare over the POR HCDN NN HCDN NN WA

 The Obs. and QPPQ Transform HCDN NN and

HCDN NN WA Rearing & Growth bioperiod 95% CI

• f the mean duration of negative run-length events,

E[durat] for both Qcritical and Qrare over the POR

Obs Qcrit QPPQ Qcrit Obs Qrare QPPQ Qrare

Event Duration (days)

5 10 15 20 Obs Qcrit QPPQ Qcrit Obs Qrare QPPQ Qrare

Event Duration (days)

5 10 15 20

HCDN NN HCDN NN WA

 The Obs. and QPPQ Transform HCDN NN and

HCDN NN WA POR and bioperiod R2 . Souhegan watershed area = 170.2 mi2

Obs and QPPQ NN Obs and QPPQ NN WA NN & NN WA Watershed areas: 129.5 mi2 147.4 mi2 Centroid-to-Centroid Distances: 24.6 miles 33.2 miles R2 (%) R2 (%) POR 34.2 68.5 Bioperiod Over-Wintering 30.7 74.5 Spring Flood 13.4 61.2 Shad Spawning 34.6 79.5 GRAF Spawning 57.0 63.4 Rearing & Growth 26.8 57.1 Salmon Spawning 27.1 57.1

The Final Tally

 HYSR examined the R2 tables and 178 graphs

found in the appendices of the Phase 2 study report.

 HYSR determined that by a 2 to 1 margin, the

HCDN NN was the better choice than the HCDN NN WA when using the QPPQ Transform to estimate daily flows at ungaged sites in New Hampshire

 With both phases of the HYSR study complete,

NHDES will be able to make an informed decision about adopting the QPPQ Transform method as the preferred way to estimate a long period time series

• f daily flows at ungaged sites in New Hampshire.

Summary

 The QPPQ Transform method was conceived and

developed by Fennessey (1994)

 The method has been widely adopted  The Step 3 regional FDC model was updated and

the QPPQ Transform broadly assessed for accuracy

 The goals of the QPPQ Transform remain the same:  Provide a method to generate a quality time

series of daily streamflow data at ungaged sites

 Allow the analyst the freedom to choose to do

whatever they want with the generated time series, Qungaged(t) as they would Qgage(t)

Questions and Comments?

Neil M. Fennessey Hydrologic Services 49 School Street South Dartmouth, MA 02748 HYSRConsult@gmail.com