Text

Response to Request for Additional Information Related to License Amendment Request to Revise Technical Specifications to Adopt Risk Informed Completion Times TSTF-505, Revision 2, Provide Risk-Informed Extended Completion
	ML24131A079
Person / Time
Site:	Quad Cities
Issue date:	05/10/2024
From:	Humphrey M; Constellation Energy Generation
To:	; Office of Nuclear Reactor Regulation, Document Control Desk
References
	RS-24-042
	Download: ML24131A079 (1)
	v • d • e

4300 Winfield Road Warrenville, IL 60555 630 657 2000 Office 10 CFR 50.90 RS-24-042 May 10, 2024 U.S. Nuclear Regulatory Commission ATTN: Document Control Desk Washington, DC 20555-0001 Quad Cities Nuclear Power Station, Units 1 and 2 Renewed Facility Operating License Nos. DPR-29 and DPR-30 NRC Docket Nos. 50-254 and 50-265

Subject:

Response to Request for Additional Information Related to License Amendment Request to Revise Technical Specifications to Adopt Risk Informed Completion Times TSTF-505, Revision 2, "Provide Risk-Informed Extended Completion Times - RITSTF Initiative 4b" and Application to Adopt 10 CFR 50.69, "Risk-Informed Categorization and Treatment of Structures, Systems, and Components for Nuclear Power Reactors"

References:

1. Letter from P.R. Simpson (Constellation Energy Generation LLC) to U.S.

NRC, "License Amendment Request to Revise Technical Specifications to Adopt Risk Informed Completion Times TSTF-505, Revision 2, 'Provide Risk-Informed Extended Completion Times - RITSTF Initiative 4b,'" dated June 8, 2023 (ADAMS Accession No. ML23159A249)

2. Letter from P.R. Simpson (Constellation Energy Generation LLC) to U.S.

NRC, "Application to Adopt 10 CFR 50.69, 'Risk-Informed Categorization and Treatment of Structures, Systems, and Components for Nuclear Power Reactors,'" dated June 8, 2023 (ADAMS Accession No. ML23159A253)

3. Email from R. Kuntz (U.S. NRC) to R. Steinman (Constellation Energy Generation, LLC), "Request for Additional Information RE: Quad Cities TSTF-505 and 50.69 amendments," dated April 10, 2024 (ADAMS Accession No. ML24102A242)

In References 1 and 2, Constellation Energy Generation, LLC (CEG) requested an amendment to Renewed Facility Operating License Nos. DPR-29 and DPR-30 for Quad Cities Nuclear Power Station (QCNPS) Units 1 and 2. The proposed amendment requested in Reference 1 would modify the Technical Specifications (TS) requirements to permit the use of Risk Informed Completion Times (RICTs) in accordance with TSTF-505, Revision 2, "Provide Risk-Informed Extended Completion Times - RITSTF Initiative 4b." The proposed amendment requested in Reference 2 would modify the QCNPS licensing basis, by the addition of a License Condition, to

May 10, 2024 U.S. Nuclear Regulatory Commission Page 2 implement the provisions of 10 CFR 50.69, "Risk-informed categorization and treatment of structures, systems and components for nuclear power reactors."

In Reference 3, the NRC requested additional information needed to support the NRC review of References 1 and 2. In response to this request, CEG is providing the attached information.

CEG has reviewed the information supporting the findings of no significant hazards consideration, and the environmental considerations, that were previously provided to the NRC in References 1 and 2. The additional information provided in this submittal does not affect the bases for concluding that the proposed license amendments do not involve a significant hazards consideration. In addition, the information provided in this submittal does not affect the bases for concluding that neither an environmental impact statement nor an environmental assessment needs to be prepared in connection with the proposed amendments.

CEG is notifying the State of Illinois of this supplement to a previous application for a change to the operating license by sending a copy of this letter and its attachments to the designated State Official in accordance with 10 CFR 50.91, "Notice for public comment; State consultation,"

paragraph (b).

There are no regulatory commitments contained in this letter. Should you have any questions concerning this letter, please contact Mr. Ken Nicely at (779) 231-6119.

I declare under penalty of perjury that the foregoing is true and correct. Executed on the 10th day of May 2024.

Respectfully, Mark Humphrey Sr. Manager Licensing Constellation Energy Generation, LLC Attachments:

1. Response to EXHB Request for Additional Information

2. Aterra Solutions, "Probabilistic Flood Hazard Assessment Report for the Mississippi River, Quad Cities Nuclear Generating Station," dated December 7, 2021 cc:

Regional Administrator - NRC Region III NRC Senior Resident Inspector - QCNPS NRC Project Manager, NRR - QCNPS Illinois Emergency Management Agency - Division of Nuclear Safety Humphrey, Mark D.

Digitally signed by Humphrey, Mark D.

Date: 2024.05.10 09:59:36 -05'00'

ATTACHMENT 1 Quad Cities Nuclear Power Station Docket Nos. 50-254 and 50-265 Facility Operating License Nos. DPR-29 and DPR-30 Response to Request for Additional Information Related to License Amendment Request to Revise Technical Specifications to Adopt Risk Informed Completion Times TSTF 505, Revision 2, "Provide Risk-Informed Extended Completion Times - RITSTF Initiative 4b" and Application to Adopt 10 CFR 50.69, "Risk-Informed Categorization and Treatment of Structures, Systems, and Components for Nuclear Power Reactors" Response to EXHB Request for Additional Information

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 1 of 16 Docket Nos. 50-254 and 50-265 REQUEST FOR ADDITIONAL INFORMATION BY THE OFFICE OF NUCLEAR REACTOR REGULATION QUAD CITIES NUCLEAR POWER STATION -

LICENSE AMENDMENT REQUESTS TO ADOPT TSTF-505, RISK INFORMED COMPLETION TIMES, RITSTF INITIATIVE 4b AND 10 CFR 50.69, RISK-INFORMED CATEGORIZATION AND TREATMENT OF STRUCTURES, SYSTEMS AND COMPONENTS FOR NUCLEAR POWER PLANTS EXHB RAI 01 Regulatory Guide (RG) 1.200, titled "Acceptability of Probabilistic Risk Assessment Results for Risk-Informed Activities" (ML090410014), section C.1.1, explains that for external hazards such as floods, the risk to a facility can be evaluated qualitatively, quantitatively, or both, albeit in a simplified manner. Section C.1.2.8 (Technical Elements for External Flood, At-Power Probabilistic Risk Assessment) of this guide further details that external flood hazard analysis determines the frequency of external floods at a site through site-specific probabilistic hazard analysis. It ensures that uncertainties in models and parameters must be appropriately considered to derive a mean hazard curve from the family of hazard curves obtained.

Appendix A, titled "Quad Cities Multi-Site Stochastic Daily Weather Simulation," of the PFHA report, describes that 49 out of 250 weather gauging stations were used for stochastic modeling. The PFHA report states that this was simplified to 10 subbasins for hydrologic modeling, covering approximately 88,600 square miles. The PFHA assumed that for the hydrologic model precipitation depths within each subbasin fluctuate over time but stay constant across space. However, this simplification could result in underestimating uncertainties in flood hazard estimates, especially at low-frequency levels. Also, the staff lacks clarity on the handling of missing weather data, if any, prior to modeling.

To address these concerns:

a) Discuss the potential reduction in variability of daily precipitation and temperature due to weather field simplification. Elaborate on the uncertainties introduced by using simplified weather fields represented by 10 subbasins for hydrologic modeling.

b) Explain the methodology employed to handle missing weather data in stochastic modeling.

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 2 of 16 Docket Nos. 50-254 and 50-265 Constellation Response to EXHB RAI 01 a) The Quad Cities Probabilistic Flood Hazard Assessment (PFHA) is provided in. The U.S. Army Corps of Engineers (USACE), Hydrologic Engineering Center, Hydrologic Modeling System (HEC-HMS) model for the Quad Cities PFHA, which simulates the precipitation-runoff response over the 88,600-mi2 upper Mississippi River watershed, was originally developed for the Probable Maximum Flood (PMF) hydrology as part of the post-Fukushima Flood Hazard Reevaluation Report (FHRR)

(Analysis No. QDC-0085-S-1990, Revision 0). The original model contained 10 subbasins, ranging in size from 286 mi2 to 19,286 mi2. The 10-subbasin structure leverages locations of USGS streamflow gages, which are needed for the long-term simulation calibration. The model is based on the use of "lumped" techniques (i.e., unit hydrograph theory) for transforming runoff volume to flow-time functions (hydrographs) at observation (gage) locations, where transformation parameters can be refined, and uncertainties quantified. Despite having 49 precipitation stations, a higher resolution model (more subbasins) would not provide significant benefit since the headwater basins could not be calibrated.

The 10-subbasin delineations used in the hydrologic model are considered to be meteorologically homogeneous since there is no significant topography within each subbasin and the meteorological and climatological settings are the same for purposes of stochastic weather evaluations. The selection of the 10 basins were based on local meteorology and climatology and available streamflow records.

Temperatures and precipitation in the region are controlled by the same general synoptic patterns and seasonality, with variations related to latitude and distance from the Great Lakes. Therefore, the use of a subset of station observations with spatial coverage over each basin provides adequate inputs for stochastic weather generation that reflects the spatial variation of temperature and precipitation throughout the region. The range of temperature and precipitation patterns are adequately covered in the stations used and the period of record applied. This does not mean that every aspect of a particular individual event was captured, but instead that the general range and combination of events was captured as needed for stochastic weather generation input. For temperatures, this is easier to capture because the spatial change/gradient in temperature throughout this region covers a large region at any given time. Therefore, this is well captured by the stations as the general air masses associated with a given temperate regime covers 100's to 1000's of square miles at any given time. For precipitation there is more spatial variability for any given event, especially for localized storms. However, precipitation events that are important for this type of analysis are controlled by long duration, large area size storms that cover 1000's of square miles and often last several days. Therefore, the spatial cover of the stations used would capture these storms adequately. Finally, Applied Weather Associates'(AWA's) storm database was investigated to confirm all extreme events that have been analyzed in the region were captured by one or more of the stations. Adding more stations would not improve the relationships developed. The final set of stations, used for stochastic modeling, provided good spatial representation from north to south and east to west while also providing a consistent and long period of record to realistically model the spatial correlations of daily rainfall, temperature, and snow water equivalent (SWE). These

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 3 of 16 Docket Nos. 50-254 and 50-265 locations were chosen based on data availability, data quality, period of record, and spatial distribution requirement.

b) The station screening started with 250 stations, which was reduced to 49 stations to provide good spatial representation from north to south and east to west while also providing a consistent and long period of record. The final stations were selected based on station period of record, data availability, and spatial distribution requirements for the calibration period identified (1949-2019). For the stations selected, the median data missing, both precipitation and temperature, was less than 1-percent (Figure 1). A complete daily time series was needed at each station for stochastic simulation; the missing data were estimated and filled in based on nearest neighbor's station data.

Figure 1. Daily Data Percent Missing for the 1949-2019 Calibration Period EXHB RAI 02 RG 1.200 delineates, in part, the quality of probabilistic risk assessments for external hazards relevant to plant safety analysis in terms of scope, detail level, and technical adequacy.

The licensee employed the RMAWGEN model for spatial multi-site stochastic generation of daily time series of temperature and precipitation. Cordano and Eccel (Cordano E. and E. Eccel

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 4 of 16 Docket Nos. 50-254 and 50-265 E. (2016), "Tools for Stochastic Weather Series Generation in R Environment," Italian Journal of Agrometeorology, doi:10.19199/2016.3.2038-5625.031), which is one of the key references to RMAWGEN, states that the vector auto-regressive model (VARs) implemented in the RMAWGEN works correctly for normally distributed variables. However, unlike minimum and maximum temperatures, precipitation data in the PFHA report, even after Gaussian transformation, does not exhibit normal distribution (see Figure 2.c). This may lead to underestimation of flood hazards and their uncertainty bounds especially on low frequency flood analysis. Moreover, the licensee states that they generated synthetic weather data using a general linear model in RMAWGEN following Wilks' approach for spatial correlation. However, the report lacks clarity regarding the implementation of Wilks' approach for spatial correlations in RMAWGEN.

The staff requests clarification on:

a) Whether RMAWGEN modeling adequately addresses non-normally distributed precipitation data and its impact on low frequency flood analysis.

b) The process and results of Wilks' approach implemented in RMAWGEN.

c) How well the generated data preserve statistics of measured weather data compared to generated data.

Constellation Response to EXHB RAI 02 a) Regarding precipitation, the RMAWGEN model utilizes VAR models, a precipitation occurrence based on Generalized Additive Models (GAMs); multi-site generation of random values based on Wilks (1998) correlation matrix; random generation of precipitation amount for each site using simultaneous precipitation occurrences as predictors. Precipitation amounts are simulated using an empirical cumulative distribution function (default) or a fitted cumulative exponential distribution, while maintaining quantile frequency magnitudes. Whereas temperature data tend to be normally distributed, then the RMAWGEN method utilizes VAR models and Gaussian transformation.

Wilks, D.S., 1998. Multisite generalization of a daily stochastic precipitation generation model, Journal of Hydrology, Volume 210, Issues 1-4.

https://doi.org/10.1016/S0022-1694(98)00186-3.

b) Wilks (1998) describes a method to simulate rainfall occurrences through generation of Gaussian random variable combinations and establishes a relationship for each pair of rain gauges between Gaussian variables correlation and binary precipitation occurrence values. See Wilks (1998) for additional details.

A daily precipitation model is based on a first-order, two-state Markov process governing daily precipitation occurrence, with serially independent precipitation amounts on wet days. A precipitation model is created from precipitation time series taken at several correlated sites. A wet day is defined as a minimum precipitation threshold, this threshold was set equal to or greater than 1 mm. If precipitation is less than the threshold or zero, the day is defined as dry. If precipitation is greater than or equal to the threshold it is defined as a dry day.

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 5 of 16 Docket Nos. 50-254 and 50-265 For wet day periods, a first-order Markov chain model follows from the assumption that the probability of a wet day is defined fully by whether precipitation occurred or did not on the previous day. This probability is conditioned to the state of the previous day(s) and is calculated by a logistic auto-regression with the predictors: (i) precipitation occurrence in the previous day(s) (ii) daily temperature if uses as exogenous variables, and (iii) the day of the year. Precipitation amounts are simulated using an empirical cumulative distribution function (default) or a fitted cumulative exponential distribution, while maintaining quantile sample frequency.

c) The final calibrated VAR precipitation (Ppt) and VAR maximum temperature (Tmax), and VAR minimum temperature (Tmin) models were compared to the observed and simulated input data. The observed versus simulated data were compared at quantiles from 0.1 to 0.99 at a 0.01 incremental level, the average correlation among the observed and simulated values were excellent with all station's correlation being greater than 0.98 (Section 3.2, Appendix A, of the PFHA Report).

In addition to the observed versus simulated correlation statistics, daily precipitation quantile statistics for the observed data and the simulated precipitation data are provided for each basin in Table 1 and monthly mean precipitation for the observed data and the simulated data are provided for each basin in Table 2. The daily quantile statistics and monthly statistics for each basin were simulated well, the one exception is for the Wapsipinicon basin 100-percentile daily statistic. The Wapsipinicon basin difference in the upper quantiles is attributed to the sampling and interpolation from an extreme precipitation event that occurred on July 22, 2017 (Figure 2). The localized precipitation was captured by the station data, this was used to interpolate and estimate the basin average daily precipitation. The average precipitation (5.35 inches) looks to be overestimated using the station point data when compared to the spatial pattern and basin average estimate based on National Weather Service (NWS) Stage IV radar estimate of (3.25 inches) which fits simulated quantile data better (Figure 2).

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 6 of 16 Docket Nos. 50-254 and 50-265 Table 1. Observed and Simulated Daily Precipitation Quantile Statistics (inches)

Table 2. Observed and Simulated Monthly Basin Mean Precipitation (inches)

Quantile 0%

5%

25%

50%

75%

95%

99%

100%

Observed 0.8 0.9 1.2 1.4 1.7 2.1 3.0 3.9 Simulated 0.9 1.0 1.3 1.5 1.8 2.6 3.0 3.9 Observed 0.9 1.2 1.3 1.6 2.0 2.5 3.4 3.6 Simulated 1.0 1.1 1.3 1.7 2.0 2.5 3.3 3.3 Observed 1.0 1.3 1.7 1.9 2.3 2.8 3.9 4.0 Simulated 1.2 1.2 1.6 1.9 2.4 3.3 3.9 4.0 Observed 1.0 1.0 1.3 1.5 1.9 2.4 2.6 2.6 Simulated 0.8 1.0 1.3 1.5 2.1 2.4 2.6 2.6 Observed 0.6 0.9 1.1 1.3 1.7 2.0 2.4 2.5 Simulated 0.8 0.9 1.2 1.4 1.7 2.3 2.4 2.5 Observed 1.0 1.4 1.7 2.1 2.5 3.6 4.4 4.5 Simulated 1.3 1.4 1.6 2.1 2.6 4.4 4.5 4.5 Observed 0.8 1.0 1.3 1.6 1.8 2.5 3.3 4.2 Simulated 0.9 1.1 1.4 1.7 1.9 2.2 2.8 2.9 Observed 0.9 1.2 1.5 1.9 2.2 2.8 3.8 5.4 Simulated 1.2 1.4 1.6 1.9 2.3 2.8 3.2 3.2 Observed 1.0 1.2 1.3 1.7 2.3 3.0 3.7 3.7 Simulated 0.9 1.1 1.3 1.8 2.3 2.7 3.0 3.4 Observed 0.8 1.0 1.2 1.4 1.7 2.1 3.1 3.8 Simulated 0.9 1.0 1.2 1.5 1.8 2.8 3.8 3.8 Wapsipinicon Winona Wisconsin Chippewa Clinton McGregor Minnesota Quad Cities St. Croix Anoka

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 7 of 16 Docket Nos. 50-254 and 50-265 Figure 2. Daily Precipitation Estimates for July 22, 2017 EXHB RAI 03 RG 1.200 defines the quality of probabilistic risk assessment in terms of scope, detail level, and technical adequacy.

Reviewing Appendix A to the PFHA report, the staff identified the following specific concerns related to setup and calibration of RMAWGEN:

Lack of detailed information on the setup and its results of the stochastic models described in Section 3.2.

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 8 of 16 Docket Nos. 50-254 and 50-265 Unclear variation of calibrated model parameters with different year data stated in Section 4.0.

Weather models built using data only within the watershed basin, contradicting regulatory requirements of General Design Criteria 2 of Part 50.

To address these concerns:

a) Describe the process and results of VAR model calibration, including selection of the order (time-lag) of VAR models, utilization of exogenous variables, goodness-of-fit statistics, and variance of white noises.

b) Explain how calibrated parameters vary with different year calibration sets and their incorporation into final VAR model.

c) Discuss the potential impacts of using historical data both inside and outside the basin based on the concept of stochastic storm transposition on evaluating low frequency floods and their uncertainties.

Constellation Response to EXHB RAI 03 a) Precipitation generation was tested with auto-regression time-lags of 1-day (P01), 3-day (P03), 1-day time-lag with Gaussian Principal Component Analysis (P01_GPCA), and 3-day time-lag with Gaussian Principal Component Analysis (P03_GPCA). Longer durations, such as 7-day lag, were not considered based on typical storm event durations in the region. Individual storms in the region typically occur within a 1-to 3-day window and therefore the 1-day and 3-day lag times tested follow the synoptic patterns that occur in the region.

The four VAR precipitation models (P01, P03, P01_GPCA, P03_GPCA) were used to evaluate goodness-of-fit measures based on model residuals. The model residuals were examined using two diagnostic tests: i) Multivariate Portmanteau and Breusch-Godfrey (Lagrange Multiplier tests) to verify the absence of time-autocorrelation of the VAR residuals, and ii) the Jarque-Bera and multivariate skewness and kurtosis test to validate the VAR residuals are multi-normally distributed.

The precipitation goodness-of-fit results (significance p-values of 0.05) indicate that both normality and serial tests were successful/passed for the GPCA with autoregression lag-time equal to 3-days (P03_GPCA), the P03_GPCA was considered the best choice for simulating precipitation among those tested (Table 3).

Table 3. Results of Normality and Seriality Tests for Precipitation Var Model Serial Test Normality Test P01 Reject Reject P03 Pass Reject P01_GPCA Reject Pass P03_GPCA Pass Pass

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 9 of 16 Docket Nos. 50-254 and 50-265 For temperature data (Tmax and Tmin), VAR models were investigated with auto-regression time-lags of 1-day (T01), 3-day (T03), and 1-day with Gaussian Principal Component Analysis (T01_GPCA), and 3-day with Gaussian Principal Component Analysis (T03_GPCA). In addition, the best fit precipitation VAR (P03_GPCA) data were used as exogenous variables for temperature simulation screening, i.e., more cloud cover and rainfall and affect temperature.

The four VAR temperature models applied goodness-of-fit measures to test model residuals. The residuals were examined using the same two diagnostic tests for normality and seriality in precipitation data. The test results (significance p-values of 0.05) indicate that both normality and serial tests were successful/pass for the GPCA with autoregression lag-time equal to 1-and 3-days (T01_GPCA; T03_GPCA), the T03_GPCA was considered the best choice for simulating temperature among those tested (Table 4).

Table 4. Results of Normality and Seriality Tests for Temperature VAR Model Serial Test Normality Test T01 Reject Reject T03 Pass Reject T01_GPCA Pass Pass T03_GPCA Pass Pass The final stochastic simulations for precipitation were based on GPCA with an autoregression time-lag 3-days (P03_GPCA) and no exogenous variables. The final stochastic simulations for temperature (Tmax and Tmin) were based on GPCA with an autoregression time-lag 3-days (T03_GPCA) and precipitation (P03_GPCA) as exogenous variables.

In RMAWGEN, the stochastic white noise is randomly generated based on the variability of the residuals. If the VAR residuals do not fit well, i.e., do not pass the serial and normality tests, the white noise would need to be manually inserted.

b) Section 4.0, Appendix A of the PFHA Report discusses the long-term calibration while Sections 3.2 and 6.1 discuss the stochastic modeling methods. Section 4.0 discusses the use of PRISM daily data to aid in identifying the appropriate calibration years to perform a detailed Storm Precipitation Analysis System (SPAS) analysis to estimate basin average inputs for hydrologic model calibration. The water years investigated were based on streamflow records, the top three wet years were 1986, 1993, and 2001 and the top three dry years were 1987, 1988, and 2009.

The daily PRISM precipitation data were used to assimilate water year metrics both spatially and by magnitude. The intent was to identify normal, wet, and dry water year to capture potential differences in soil moisture and evapotranspiration for hydrologic model calibration. Based on the PRISM sub-basin precipitation screening, the 1993 water year was selected to represent a wet year calibration, and the 1988 water year was selected to represent a dry year calibration. The 1991 period was selected to represent a normal

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 10 of 16 Docket Nos. 50-254 and 50-265 precipitation year based on streamflow records and because it bracketed the dry year (1988) and wet year (1993) calibration periods.

The daily station data for water years 1988, 1991, and 1993 were used in SPAS to provide daily estimates of sub-basin average precipitation as input into the hydrologic model for calibration. The PRISM data were not used to calibrate the stochastic VAR model. The daily station data used to calibrate the stochastic VAR model were used as input into SPAS for water years 1988, 1991, and 1993 calibration using the same basin average format as the stochastically generated data.

c) Many probabilistic approaches have been developed for extreme flood estimation, ranging from local or regional frequency analysis of flood peaks or volumes to rainfall-runoff stochastic simulation methods. Most of these probabilistic estimate approaches can be distinguished in two different ways: i) event-based simulation, and ii) continuous simulations. Event based simulations are when the rainfall-runoff model is used to simulate a design, reference, or storm event of a defined probability. Continuous simulations are when the rainfall-runoff model is driven by historical or synthetic rainfall records to generate a continuous streamflow series. Both approaches can be used to estimate flood frequencies, but advantages for continuous simulation approach compared to the event-based have been identified in literature. The main advantage is that the soil moisture conditions are continuously accounted for by the rainfall-runoff model, while in the event-based approaches, assumptions must be made regarding the pre-storm (antecedent) conditions (or initial loss) in the catchment.

The Stochastic Storm Transposition (SST) method is an event-based simulation, whereas stochastic weather generators are a continuous simulation based on historical data. SST includes several key elements: i) defining a transposition domain; ii) developing an extreme storm catalog; iii) randomly transposing storms in a region over a watershed; and iv) estimating rainfall or flood probabilities. In doing so, new realizations of extreme rainfall are created for a watershed of area that resides within this transition region. On key aspect of the SST is that the spatial and time structure of storm events, including intensities, areas, and movement is preserved. However, significant judgment is applied in the SST process by assuming that a storm would behave the same way as observed in the original location as the new location. Factors such as moisture access, topography differences, and atmospheric dynamic differences are assumed to be the same, which is not always the case.

A hybrid option, which AWA has implemented, is the SCHADEX method which is termed a semi-continuous simulation method that incorporated SST to account for extreme precipitation events beyond the observed historical dataset in the basin of interest (Paquet et al, 2013; Lawrence et al, 2014). The SCHADEX method uses continuous rainfall stochastic generation along with synthetic SST events inserted into the continuous historical rainfall records. This process allows the simulation of floods generated by precipitations representing a full range of return periods and which embrace all possible hydrological conditions for a given catchment while maintaining soil moisture conditions continuously described by the rainfall-runoff model based on the historical record such that correct seasonality and realistic dry/wet sequences are reproduced.

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 11 of 16 Docket Nos. 50-254 and 50-265 Both stochastic weather generators and SST methods are statistical models that simulate realistic or plausible random sequences of atmospheric variables such as temperature and precipitation. These synthetic sequences provide a set of alternate realizations that can be used for risk and reliability assessment in the design and operation of agricultural, water resource and environmental systems. Both methods have their own strengths and limitations, which range from data availability, number and type of observed storm events in record, data processing, computation efficiency, and statical representation of flood simulations. For example, stochastic simulations that use rainfall data to generate a continuous streamflow series allow streamflow/flood statistics to be directly calculated whereas the SST/event models typically relate the rainfall input information (e.g. storm duration, total and maximum rain depths) with the basins soil moisture to described statistical patterns of streamflow (Paquet et al, 2013).

Lawrence, D., E. Paquet, J. Gailhard, and A.K. Fleig., 2014. Stochastic semi-continuous simulation for extreme flood estimation in catchments with combined rainfall-snowmelt flood regimes, Nat. Hazards Earth Syst. Sci 14, 1283-1298.

https://doi.org/10.5194/nhess-14-1283-2014 Paquet, E., F. Garavaglia, R. Garçon, and J Gailhard, 2013. The SCHADEX method: A semi-continuous rainfall-runoff simulation for extreme flood estimation, Journal of Hydrology, 495:23-37. https://doi.org/10.1016/j.jhydrol.2013.04.045.

EXHB RAI 04 Guidance C.1.2.8 in RG 1.200 recommends a probabilistic hazard analysis incorporating recent site-specific information and up-to-date databases. Section 6, Appendix A of the PFHA report discusses generating site-specific weather data for climate change projection scenarios based on the Intergovernmental Panel on Climate Change, fifth assessment report (IPCC, AR5).

However, it's unclear to the staff why the climate change-based precipitation projection scenarios which could potentially increase the magnitude and frequency of future floods are not used and how weather data generation with climate change scenarios was performed.

To address these concerns:

a) Justify the exclusion of certain climate change-based precipitation projection scenarios from hydrologic simulation and flood frequency analysis, despite evidence from IPCC reports indicating increased heavy precipitation events.

b) Clarify the use of about 100-year monthly weather projections for IPCC's Representative Concentration Pathways (RCPs) 4.5 and 8.5 to generate 12 future 1000-year daily climate scenarios with RMAWGEN.

Constellation Response to EXHB RAI 04 a) The Intergovernmental Panel on Climate Change, sixth assessment report (IPCC, AR6) that utilizes Coupled Model Intercomparison Project Phase 6 (CMIP6) Shared Socioeconomic Pathways (SSP) 4.5 and 8.5 were not available at the time this study was conducted. The climate change analysis was based on the current IPPC AR5 CMIP5 Representative Concentration Pathways (RCPs) 4.5 and 8.5.

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 12 of 16 Docket Nos. 50-254 and 50-265 The ClimateNA data, developed by the University of British Columbia, contains gridded monthly, seasonal, and annual data for more the 50 variables was used to quantify monthly and annual future climate scenarios. The ClimateNA precipitation and temperature datasets were used to estimate monthly and annual climatologies and statistics for RCP 4.5 and RCP 8.5 climate scenarios. Short duration events (1-day and 3-day) climate change data were not investigated for this study. In other studies that are based on CMIP6 data, short duration events (1-day and 3-day), monthly, seasonal, and annual statics were applied to the stochastic model. Furthermore, short-duration events do not impact flooding on relatively large watersheds with flat terrain, which produce long lag times.

For reference, AWA has completed more than 70 climate change studies (as of April 2024). Studies completed before 2021 used CMIP5 (RCP 4.5 and 8.5 scenarios) while studies since that time are based on CMIP6 (SSP 4.5 and 8.5 scenarios). Results of these studies demonstrate several key findings that can lead to potential impacts and uncertainty on the analysis if it were to use CMIP6:

1) First, the number of climate models used to generate climate change scenarios:

CMIP5 had fewer than 8 models whereas CMIP6 has more than 30 models. A CMIP6 analysis will provide more robust statistics due to sample size and provide a better representation of the ensemble mean. In addition, the CMIP6 data used a longer "historical" period (1950-2014) compared to CMIP5 period (1950-2005).

2) Temperature - when looking and daily, monthly, annual, and frequency statistics there is less variability in temperature statistics and the CMIP5 versus CMIP6 estimates compare quite well. For example, maximum daily temperature change by the year 2100 for RCP/SSP 4.5 scenarios are typically between 2.0C to 3.0C and RCP/SSP 8.5 scenarios are typically between 5.0C to 6.0C.

3) Precipitation - results have more variability; this can be attributed to the non-linear relationship of temperature to precipitation. The variability in precipitation for the CMIP5 daily, monthly, and annual frequency statistics tend to be greater than CMIP6 statistics for similar locations. Also, the CMIP6 analysis provides a more robust ensemble mean estimate that is based on a larger suite of climate models and longer calibration period.

Table 5 shows two example locations that are based on the same statistical analysis methods but differ in the inputs which are CMIP5 RCP8.5 and CMIP6 SSP8.5 climate scenario. The table highlights the general findings for temperature and precipitation as presented above on CMIP5 versus CMIP6 analysis.

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 13 of 16 Docket Nos. 50-254 and 50-265 Table 5. Comparison of Climate Change Analysis based on CMIP5 RCP8.5 and CMIP6 SSP8.5 Climate Scenarios for Two Locations b) The CMIP6 Shared Socioeconomic Pathways (SSP) 4.5 and 8.5 were not available at the time of this project, so the climate change analysis was based on the available CMIP5 Representative Concentration Pathways (RCPs) 4.5 and 8.5.

In RMAWGEN, the calibrated VAR stochastic model can be used to generate a time series with specific probability distribution for future climate change scenarios. For this study, a matrix was created that summarized monthly mean daily precipitation for each future RCP scenario. The ClimateNA data, developed by the University of British Columbia, contains gridded monthly, seasonal, and annual data for more the 50 variables was used to quantify monthly and annual future climate scenarios. The ClimateNA precipitation and temperature datasets were used to estimate monthly and Example 1 Median Min Max Median Min Max Temperature 1-Day; C 5.5 3.9 5.9 6.0 3.4 7.8 Temperature 1-Day Dry; C 5.3 4.1 6.2 6.1 3.3 8.0 Temperature 1-Day Wet PF; C 5.5 3.7 5.9 6.1 3.2 7.8

+Precipitation 1-Day PF; %

28

-19 145 5

-23 63 Precipitation 1-Day Dry PF; %

63 4

176 0

0 0

Precipitation 1-Day Wet PF; %

-5

-21 8

5

-23 63

+Precipitation 3-Day PF; %

26

-4 146 13

-8 77 Precipitation 3-Day Dry PF; %

116 12 201 0

0 0

Precipitation 3-Day Wet PF; %

1

-8 27 13

-8 77 Precipitation Annual PF; %

37 20 64 26

-16 140 Example 2 Median Min Max Median Min Max Temperature 1-Day; C 4.9 4.4 5.6 5.9 2.9 8.4 Temperature 1-Day Dry; C 4.9 4.4 5.6 5.9 2.9 8.4 Temperature 1-Day Wet PF; C 5.1 4.5 6.7 6.6 3.3 9.7

+Precipitation 1-Day PF; %

59 12 73 2

-28 21 Precipitation 1-Day Dry PF; %

75 14 80 3

-21 27 Precipitation 1-Day Wet PF; %

7

-22 34 3

-21 39

+Precipitation 3-Day PF; %

65 6

71 10

-15 58 Precipitation 3-Day Dry PF; %

72 9

94 5

-27 66 Precipitation 3-Day Wet PF; %

38

-7 47 10

-21 88 Precipitation Annual PF; %

20 2

38 8

-18 43 RCP85 SSP85 RCP85 SSP85

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 14 of 16 Docket Nos. 50-254 and 50-265 annual climatologies and statistics for RCP 4.5 and RCP 8.5 climate scenarios. The estimated RCP 4.5 and RCP 8.5 monthly climatologies, estimated from monthly and annual statistics in the climate change datasets, were used with the calibrated VAR to generate future climate scenario time series. The 12 future scenarios of 1000-year daily data used the calibrated VAR stochastic model, the updated RCP 4.5 and RCP 8.5 monthly climatology matrices, and 12 random generated scenarios.

EXHB RAI 05 RG 1.200 sets standards for the quality of flood hazard analysis for PRA in terms of scope, detail level, and technical adequacy.

Section 4.3 of the PFHA report mentions recalibrating hydrologic model for continuous simulation. However, this section lacks sufficient detail for NRC staff to conclude the adequacy of model setup and recalibration.

To address these concerns:

a) Describe changes in key model parameters from recalibration and their impacts on estimating peak flow events, if any.

b) Discuss the adequacy of using 1991 calibration parameters for flood frequency analyses at the 1E-6 annual exceedance probability.

Constellation Response to EXHB RAI 05 a) As discussed previously, the HEC-HMS model for the PFHA study was originally used to develop hydrology for the Probable Maximum Flood (PMF) as part of the post-Fukushima Flood Hazard Reevaluation (Analysis No. QDC-0085-S-1990, Revision 0).

The original model contained 10 sub-basins, ranging in size from 286 mi2 to 19,286 mi2, and was developed for single events. It used a constant loss approach for abstractions/retention, synthetic Clark Unit Hydrograph for runoff transformation, and the Muskingum method for river reach routing. The initial loss and percent impervious were all set at 0.0. The PMF HEC-HMS model parameters were calibrated using observed data from past single events.

For the Quad Cities PFHA, a continuous simulation approach was selected to reduce uncertainties related to antecedent conditions and soil-moisture recovery. While setup and calibration of the PMF HEC-HMS model was considered adequate for single event analyses, the PFHA HEC-HMS model warranted a long term, continuous approach (for reasons previously discussed). It was, therefore, necessary to develop and calibrate for a continuous simulation. The model must have the capability to "recover" during periods of no rainfall and simulate rainfall-runoff processes over an annual cycle.

There are twelve (12) different loss methods available in HEC-HMS with five (5) having continuous simulation capabilities. Based on a review of the original model, available data, and experience, the deficit and constant approach was selected as the loss method. Similar to the original PMF model, the Muskingum method was used for river reach routing and the Clark Unit Hydrograph method was used for runoff transformation.

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 15 of 16 Docket Nos. 50-254 and 50-265 For long term simulations, it was necessary to add several other components that essentially allow the model to keep track of precipitation as it moves through the system to the outlet, specifically canopy, surface, and alternate groundwater methods were incorporated into the model.

The Simple Canopy method was selected to represent a plant layer as a single entity.

This approach uses several parameters to represent the effects of the vegetation in each subbasin, including initial storage, maximum storage, a crop coefficient, evapotranspiration, and soil uptake. Each of these parameters and approaches were used during the calibration phase. A Simple Surface was used to represent the soil surface where water may be held in storage after a rain event. It is a relatively small portion of the overall annual water budget. Conversely, evapotranspiration plays a major role in the overall water budget. For the continuous simulations, a Monthly Evapotranspiration approach was used. Evapotranspiration is an input to the meteorological model within the HEC-HMS framework. Lastly, a Groundwater Reservoir approach was used, with each subbasin being assigned three (3) groundwater reservoirs. This is also known as the Linear Reservoir Baseflow method within the HEC-HMS framework. As the name implies, water entering the subsurface is routed through a series of reservoirs. This approach allows water to be stored and returned to the system at various times. The key target in calibrating the model by adjusting these parameters was the annual maximum flow, which was extracted for use in the frequency analysis.

b) The objective in calibrating the HEC-HMS model was to provide a reasonably realistic representation of the watershed's response to annual rainfall patterns, with an emphasis on matching peaks as closely as possible. 1988 was an atypical year and, therefore, not used to "fine tune" the input parameters. Parameters calibrated to the 1991 event provided the most representative watershed response for a range of precipitation magnitudes and was viewed as being most suitable for the continuous simulation. The strategy was to avoid excessive over or under estimation of flows, particularly at the peak, since the flow-stage-frequency results would contain conservatisms by using upper-bound flows from the Monte Carlo simulations and climate scenarios.

EXHB RAI 06 Guidance C.1.2.8 in RG 1.200 states that external flood hazard analysis estimates the frequency of external flood at the site using a site-specific probabilistic hazard analysis and that uncertainties in the models and parameter values are properly accounted for and fully propagated to allow the derivation of a mean hazard curve from the family of hazard curves obtained.

Section 7 of the PFHA report addresses uncertainties in flood frequency analysis considering the following:

For aleatory variability, the licensee used a Monte Carlo sampling and hydrologic simulations, to cope with the variability of input data, assumptions, parameters, weather scenarios, and others.

QCNPS Requests to Adopt TSTF-505 and 10 CFR 50.69 Response to EXHB Request for Additional Information Page 16 of 16 Docket Nos. 50-254 and 50-265 For epistemic uncertainty in flood frequency estimates, they fitted different probability distribution functions using the L-moment method, analyzed stage-discharge curves using different calibration years, and considering different combined flood events.

However, the licensee does not account for uncertainties associated with weather field simplification and different weather projection scenarios. To address these concerns:

a) Discuss mean flood frequency curve with appropriate uncertainty bounds, either quantitively, qualitatively, or mixed.

b) Discuss how uncertainties in flood frequency estimation impact suggested risk-informed flood protection or mitigation measures.

Constellation Response to EXHB RAI 06 a) As indicated in the RAI, the family of flow-stage-frequency curves contained in Section 7.4 of the PFHA Report (Results of Uncertainty Analysis) represent the mean curve for each respective scenario. The 90% confidence interval curves (which represent theoretical upper and lower bounds around the mean when fitting a distribution function to computed moments from each of the 12,000-year annual maximum flow dataset) were intentionally excluded from the results. Proposing results at the upper confidence limit would be a mischaracterization of uncertainties and inappropriate for use in risk-informed decision making. The analysis considered 132,000 years of synthetic data by computing a family of 11 curves, each with 12,000 years of annual maximum flows in the dataset. Rather than creating one 132,000-year trace, which would dilute the Monte Carlo results, each of the 11 curves were plotted separately to identify the upper and lower bounds for the uncertainty analysis. Using only the most conservative climate scenario (normal) from Table 10 in the PFHA Report, Table 12 in the PFHA Report shows a relatively narrow range in water surface elevations (WSEs)

(595.9 to 596.5) from the Monte Carlo simulations. Overall, the approach taken provides uncertainty bounds while maintaining a "demonstrably conservative" conclusion, which is most appropriate for use in this type of risk-informed application.

b) The WSE associated with screening the external flooding hazard is approximately 596.5 ft (1 ft above finished floor elevation). This is the WSE where CDF is shown to be below 1E-6/YR (screening limit). Uncertainties, including combined events, were evaluated to demonstrate that the available physical margin (APM) from the top of this WSE and top of barriers (599.0 ft) is approximately 2.5 feet. The margin present is adequate to confirm that conclusions from the screening analysis will remain unchanged given the relatively narrow range of WSEs computed in the uncertainty analysis for the hazard curve. The protection measures installed at Quad Cities would continue to have margin even considering elevated WSEs from the uncertainty evaluation. Additionally, there are no parameters that would change the timing associated with the scenario.

Given that the time margin is so large to install the protective gates, a shortened warning time would not change the reliability of installation or carrying out the strategy.

ATTACHMENT 2 Quad Cities Nuclear Power Station Docket Nos. 50-254 and 50-265 Facility Operating License Nos. DPR-29 and DPR-30 Response to Request for Additional Information Related to License Amendment Request to Revise Technical Specifications to Adopt Risk Informed Completion Times TSTF 505, Revision 2, "Provide Risk-Informed Extended Completion Times - RITSTF Initiative 4b" and Application to Adopt 10 CFR 50.69, "Risk-Informed Categorization and Treatment of Structures, Systems, and Components for Nuclear Power Reactors" Aterra Solutions, "Probabilistic Flood Hazard Assessment Report for the Mississippi River, Quad Cities Nuclear Generating Station," dated December 7, 2021

www.aterrasolutions.com PROBABILISTIC FLOOD HAZARD ASSESSMENT REPORT FOR THE MISSISSIPPI RIVER Quad Cities Nuclear Generating Station December 7, 2021

December 7, 2021 Page i Aterra Solutions, LLC Quad Cities - Flood-Frequency Analysis

©2021 All Rights Reserved FLOOD-FREQUENCY ANALYSIS FOR MISSISSIPPI RIVER Quad Cities Nuclear Generating Station TABLE OF CONTENTS 1.0 PURPOSE........................................................................................................................................ 1 2.0 STUDY AREA.................................................................................................................................. 1 3.0 STOCHASTIC WEATHER MODELING.......................................................................................... 2 3.1 Overview............................................................................................................................. 2 3.2 Calibration........................................................................................................................... 4 4.0 HYDROLOGIC MODELING............................................................................................................ 6 4.1 Overview............................................................................................................................. 6 4.2 HEC-HMS Model................................................................................................................. 8 4.3 Hydrologic Model Calibration.............................................................................................. 8 4.4 HEC-HMS Thousand-Year Simulations............................................................................ 12 5.0 FLOW-FREQUENCY ANALYSIS................................................................................................. 13 6.0 STAGE-FREQUENCY ANALYSIS................................................................................................ 16 7.0 UNCERTAINTY ANALYSIS.......................................................................................................... 17 7.1 Aleatory Variability............................................................................................................ 17 7.1.1 Climate Change................................................................................................... 17 7.1.2 Variability in the Hydrologic Inputs....................................................................... 17 7.1.3 Natural Variation in River System........................................................................ 19 7.1.4 Land Use Changes.............................................................................................. 19 7.2 Epistemic Uncertainty....................................................................................................... 19 7.2.1 Probability Distribution Function.......................................................................... 19 7.2.2 Potential Error in Stage-Discharge Relationship................................................. 20 7.2.3 Event Combinations............................................................................................. 20 7.3 Summary of Approach to Quantifying Uncertainties......................................................... 21 7.4 Results of Uncertainty Analysis........................................................................................ 22

8.0 CONCLUSION

S............................................................................................................................. 26 9.0 IDENTIFICATION OF COMPUTER PROGRAMS........................................................................ 27

10.0 REFERENCES

............................................................................................................................... 27

December 7, 2021 Page ii Aterra Solutions, LLC Quad Cities - Flood-Frequency Analysis

©2021 All Rights Reserved LIST OF FIGURES Figure 1. QCNGS and Drainage Basin........................................................................................................ 1 Figure 2. Daily Meteorological Stations across the Quad Cities Basins (stations selected for stochastic model calibration and simulation are shown in red).............................................................................. 3 Figure 3. Representative Concentration Pathway (RCP) Trajectories from IPCC AR5.............................. 4 Figure 4 - Water Years Investigated for Model Calibration (precipitation in inches)..................................... 5 Figure 5. Normal Water Year (1991) Investigated for Model Calibration (precipitation in inches).............. 6 Figure 6. Hydrologic Modeling Sub-Basins.................................................................................................. 7 Figure 7. Location of USGS Stream Gauge Locations used in Calibration................................................. 9 Figure 8. Calibration Results @ Clinton, IA for 1988 Water Year (dry)..................................................... 11 Figure 9. Calibration Results @ Clinton, IA for 1991 Water Year (Normal)............................................... 11 Figure 10. Calibration Results @ Clinton, IA for 1993 Water Year (wet)................................................... 12 Figure 11. Example of 1,000-year Daily Flow Output at QCNGS with Enhanced Area............................ 13 Figure 12. Example of Annual Maximum Flows for 1,000-year Trace at QCNGS.................................... 14 Figure 13. Flow-Frequency Plot for 12,000-year "Normal Climate Scenario............................................ 15 Figure 14. Rating Curve Developed for QCNGS....................................................................................... 16 Figure 15 - Water Surface Elevations Associated with AEP Flows............................................................ 17 Figure 16. Example of Uncertainty Analysis Output Illustrating Min, Max, and Mean Daily Flows........... 18 Figure 17. L-Skewness vs. L-Kurtosis for a typical 1,000-year Simulation................................................ 20 Figure 18. Annual Exceedance Probability Flow Rates (cfs) for Normal, RCP45, RCP85, & Enhanced ET Analyses.............................................................................................................................................. 22 Figure 19. Annual Exceedance Probabilities WSELs for Normal, RCP45, RCP85, & Enhanced ET Analyses.............................................................................................................................................. 23 Figure 20. Annual Exceedance Probabilities Values (cfs) for Monte Carlo Simulations........................... 24 Figure 21. Annual Exceedance Probabilities Values (WSELs) for Monte Carlo Simulations..................... 25 LIST OF TABLES Table 1. Water Year Average Sub-Basin Total Precipitation for each Calibration Year Investigated......... 5 Table 2. Sub-Basins and Drainage Areas.................................................................................................... 7 Table 3. HEC-HMS Model Elements........................................................................................................... 8 Table 4. Mean Aerial Precipitation for Wet, Average, and Dry Years.......................................................... 9 Table 5. Calibration Statistics for Water Years.......................................................................................... 10 Table 6. Flow-Frequency Analysis Results for 12,000-year "Normal Climate" Scenario.......................... 15 Table 7. Water Surface Elevations Associated with AEP Flows................................................................ 16 Table 8. HEC-HMS Parameter and Sub-Basin Components Used in Uncertainty Analysis..................... 18 Table 9. Annual Exceedance Probability Flow Rates (cfs) for Normal, RCP45, RCP85, & Enhanced ET Analyses.............................................................................................................................................. 22 Table 10. Annual Exceedance Probabilities WSELs for Normal, RCP45, RCP85, & Enhanced ET Analyses.............................................................................................................................................. 23 Table 11. Annual Exceedance Probabilities Values (cfs) for Monte Carlo Simulations............................. 24 Table 12. Annual Exceedance Probability Water Surface Elevations for Monte Carlo Simulations.......... 25 Table 13. Flow Comparisons at Higher Frequencies................................................................................. 26 APPENDICES Appendix A: Applied Weather Associates (AWA), Quad Cities Multi-Site Stochastic Daily Weather Simulation Report

December 7, 2021 Page iii Aterra Solutions, LLC Quad Cities - Flood-Frequency Analysis

©2021 All Rights Reserved LIST OF ACRONYMS AND ABBREVIATIONS AEP Annual Exceedance Probability ARC Antecedent Runoff Conditions AWA Applied Weather Associates cfs Cubic Feet per Second CN Curve Number EMA Expected Moments Algorithm ft Foot/Feet HEC-HMS USACE Hydrologic Engineering Centers Hydrologic Modeling System HEC-SSP USACE Hydrologic Engineering Centers Statistical Software Package LIP Local Intense Precipitation MSL Mean Sea Level NGVD29 National Geodetic Vertical Datum of 1929 NOAA National Oceanic and Atmospheric Administration NSE Nash-Sutlcliff Estimator PMF Probable Maximum Flood PMP Probable Maximum Precipitation QCNGS Quad Cities Nuclear Generating Station USACE United States Army Corps of Engineers USGS United States Geological Survey WSEL Water Surface Elevation

December 7, 2021 Page 1 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved 1.0 PURPOSE The purpose of this calculation is to develop Probabilistic Flood Hazard Assessment (PFHA) for Mississippi River flooding at Exelons Quad Cities Nuclear Generating Station (QCNGS). The objective of the PFHA is to characterize flood-stage-frequencies at QCNGS to an annual exceedance probability (AEP) of 10-6. It should be noted that the flood-frequency and annual exceedance probability (AEP) are often used interchangeably, as the AEP is the inverse of the flood-frequency return period. For example, a 100-year flood frequency return period has an AEP of 1/100 or 10-2 percent of occurring in any given year. Previously, Aterra Solutions, LLC (Aterra) completed a Flood Frequency Analysis, per the Nuclear Energy Institute (NEI) guidance 16-05 (Nuclear Energy Institute, 2016), which characterized flood frequencies to an AEP of 10-4 based on methods described in the Federal Advisory Committee on Water Information, Bulletin 17C (England, 2018). In this study, Aterra was supported by Applied Weather Associates (AWA) in conducting stochastic weather modeling to produce future daily weather scenarios, which were used as input to a continuous hydrologic simulation.

2.0 STUDY AREA The QCNGS is located along the upper reach of the Mississippi River approximately 10 miles south of Clinton, IA. The drainage basin is approximately 88,600 mi2, located in the upper midwest region of the United States covering much of Wisconsin and Minnesota and small portions of South Dakota, Iowa, and Illinois (Figure 1).

Figure 1. QCNGS and Drainage Basin

December 7, 2021 Page 2 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved 3.0 STOCHASTIC WEATHER MODELING 3.1 Overview AWA previously completed a site-specific Probable Maximum Precipitation (PMP) (Kappel W. H., 2012) and Local Intense Precipitation (LIP) analysis (Kappel W. H., 2014) for this basin and location. Therefore, AWA had a detailed understanding of the weather and climate of the region, spatial and seasonal variations, and available data sets.

Important for stochastic weather modeling, this region is covered with a large number of historical observations, many with long period of record. An investigation of daily meteorological stations available resulted in more than 250 daily stations initially identified within the basin (Figure 2). The topography and meteorological characteristics across the basin can be considered generally homogenous, with minimal climatological variations from south to north and east to west. The large distance across the basin generally results in temperatures cooling from south to north, annual precipitation decreasing from south to north, and annual snowfall increasing from south to north.

To reduce computational time, while still providing adequate data to realistically model the spatial correlations of daily rainfall, temperature, and snow water equivalent (SWE), the initial list of stations identified were consolidated into representative list. This resulted in 49 station locations shown in Figure

2. These stations provided good spatial representation from north to south and east to west while also providing a consistent and long period of record. These locations were chosen based on data availability, data quality, period of record, and spatial distribution requirements. Several discussions occurred to confirm the data were sufficient to produce a high quality, long duration meteorological data output required for detailed stochastic weather generation and provided the input needed for hydrologic evaluations.

The analysis utilized a multisite stochastic modeling approach using daily observations of precipitation, minimum and maximum temperatures, and SWE from the 49 sites located across the basin. Stochastic weather modeling utilized the Multi-site Auto-regressive Weather GENerator (RMAWGEN))

framework. Stochastic weather generators are statistical models that simulate realistic or plausible random sequences of atmospheric variables such as temperature and rainfall. The stochastic weather generators attempt to reproduce the spatial and temporal dynamics and correlation structures of the variables of interest. The synthetic sequences provide a set of alternate meteorological realizations that were provided to Aterra to develop runoff simulations and discharge-stage-frequency relationships at QCNGS.

December 7, 2021 Page 3 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

Observed precipitation, temperature, and SWE records were used to calibrate RMAWGEN model for the 1949-2019 period. Validation was performed on the calibration period data. Results of the calibrated model demonstrate that the model was able to capture spatiotemporal characteristics of observed precipitation, temperature, and SWE converted to snowmelt fields, such as inter-site and inter-variable correlation, and regional variations present in observed data. The calibrated stochastic model was used to generate twelve (12) 1,000-year iterations of daily weather sequences of precipitation, minimum and maximum temperatures, and SWE.

In addition to the observational data, two climate change projection scenarios of daily weather sequences of precipitation, minimum and maximum temperatures, and SWE were evaluated. These climate change projections utilized regional downscaled model output for the region driven by representative concentration pathway (RCP) 4.5 and 8.5 from CMIP5 global climate model output. These model projections were used to generate twelve (12) 1,000-year iterations representing possible future climate scenarios. The Intergovernmental Panel on Climate Change (IPCC) fifth assessment report (AR5) (IPCC, 2017) contains representative concentration pathways (RCPs), a greenhouse gas concentration trajectory; often referred to as emission scenarios. The pathways describe different climate futures, all of

December 7, 2021 Page 4 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved which are considered possible depending on the volume of greenhouse gases (GHG) emitted in the years to come (Figure 3). The RCPs investigated; RCP45 and RCP85; represent a mid-level mitigation and no mitigation to limit radiative forcing (IPCC, 2017). The thirty-six (36) 1,000-year simulations equated to 36,000-years of simulated plausible sequences of precipitation, daily temperature, and SWE. These 1,000-year traces were then used as meteorological inputs to the hydrologic model, discussed further below.

Figure 3. Representative Concentration Pathway (RCP) Trajectories from IPCC AR5 3.2 Calibration AWA utilized the Storm Precipitation Analysis System (SPAS) to analyze rainfall over the Quad Cities basin (Hultstrand, 2017). Three water years were selected for calibration: a wet, a dry, and a normal precipitation year. Based on streamflow records, the top three wet years were 1986, 1993, and 2001 and the top three dry years were 1987, 1988, and 2009. AWA used the daily PRISM gridded dataset (Daly, 1997) to assess the precipitation spatial component and basin magnitude component for each water year (Figure 4 and Table 1). Based on sub-basin frequency counts, the 1993 water year was selected to represent the wet calibration period, and the 1988 water year was selected to represent the dry calibration period. The 1991 period was selected to represent the normal precipitation period because of its streamflow record and because it bracketed the dry year (1988) and wet year (1993) calibration periods (Figure 5). Daily values of precipitation, temperature, and SWE were generated for each of the Quad Cities sub-basins utilizing the 49 meteorological stations identified in Section 3.1.

December 7, 2021 Page 5 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

Table 1. Water Year Average Sub-Basin Total Precipitation for each Calibration Year Investigated Basin Name Area (mi2)

Wet Years Dry Years 1986 1993 2001 1991 1987 1988 2009 Quad Cities 330 46.5 52.8 33.7 29.7 32.5 20.4 44.2 Minnesota 17,065 34.6 37.9 30.6 34.0 19.0 18.7 25.3 Chippewa 9,079 39.1 34.0 35.4 36.1 24.4 24.5 23.3 St. Croix 7,727 39.3 30.3 35.6 37.3 19.9 24.1 24.3 Anoka 19,905 34.8 30.8 30.9 31.4 19.9 21.0 26.2 Wapsipinicon 2,333 41.9 53.1 36.7 34.8 31.2 21.9 38.9 McGregor 7,882 38.1 43.7 34.6 33.5 29.4 24.5 28.6 Winona 5,969 41.7 42.6 35.1 34.1 27.4 23.1 26.1 Clinton 7,892 41.4 50.9 36.5 34.5 31.1 22.4 36.3 Wisconsin 10,419 37.4 39.3 33.1 33.0 25.1 26.2 27.2 Wet/Dry Count 3

7 0

3 6

1 Three wet years investigated Three dry years investigated

December 7, 2021 Page 6 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved Figure 5. Normal Water Year (1991) Investigated for Model Calibration (precipitation in inches) 4.0 HYDROLOGIC MODELING 4.1 Overview The U.S. Army Corps of Engineers (USACE), Hydrologic Engineering Center, Hydrologic Modeling System (HEC-HMS) was used to perform the hydrologic simulations. For the purposes of this study, the overall watershed was sub-divided into 10 smaller sub-basins (Figure 6). Table 2 provides the sub-basin names and drainage areas.

December 7, 2021 Page 7 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

Quad Cities 330 Minnesota 17,065 Chippewa 9,079 St.Croix 7,727 Anoka 19,905 Wapsipinicon 2,333 McGregor 7,882 Winona 5,969 Clinton 7,892 Wisconsin 10,419

December 7, 2021 Page 8 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved 4.2 HEC-HMS Model A continuous simulation approach was used for this study, as opposed to an event-based model. An event-based simulation typically uses a single rainfall event, and the primary output of interest is flow exiting the watershed. In these situations, there is often no need for modeling certain components of the hydrologic cycle, such as evapotranspiration and groundwater storage, since these have little to no impact on flows during a single event lasting a few hours to a few days. As previously noted, this study seeks to model the watersheds using stochastically generated weather inputs for long-term periods (1,000 years each).

In the continuous simulation, the fate of the input precipitation is modeled at the sub-basin scale. Each sub-basin is represented by a series of elements or methods and associated parameters that simulate the various hydrologic processes. The sub-basin elements (methods) in this study include: a canopy method, a surface method, a loss method, transformation method, and a baseflow method. Flow in the rivers must also be modeled in the various river reaches using a reach routing technique. The HEC-HMS User's Manual (USACE, 2000) provides in-depth descriptions of each element and process used, summarized in Table 3.

Table 3. HEC-HMS Model Elements Element/Method Chosen Method Canopy Method Simple Canopy Surface Method Simple Surface Infiltration / Loss Deficit & Constant Transformation Clark Unit Hydrograph Baseflow Linear Reservoir (3 Reservoirs)

Reach Routing Muskingum 4.3 Hydrologic Model Calibration The HEC-HMS model was based on the model developed for the 2013 QCNGS post-Fukushima 50.54(f)

Flood Hazard Reevaluation and is associated with QCNGS Calculation Number QDC-0085-S-1990 (Exelon Nuclear, 2012). The model uses the same 10 sub-basins illustrated above, as well as a number of river reaches and junctions.

In calibrating the HEC-HMS model, three (3) different years were chosen that represented what were deemed a normal flow year, a dry year, and a wet year. These were selected by visually examining flow data at the USGS River Gauge at Clinton, IA, as well as other gauges located in the overall drainage area. Several years were selected, and then additional evaluations were made using PRISM (Daly, 1997) monthly data to get a better sense of the spatial distribution of precipitation for each Water Year.

Based on the combination of streamflow and mean aerial precipitation data, the years 1988 (dry), 1991 (average) and 1993 (wet) were selected for calibration. Table 4 illustrates the mean aerial precipitation for each sub-basin water year. The locations of stream gauges used for the calibration are illustrated in Figure 7.

December 7, 2021 Page 9 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved Table 4. Mean Aerial Precipitation for Wet, Average, and Dry Years Basin Name Wet Avg Dry WY1993 1991 WY1988 Quad Cities 52.8 30.0 20.4 Minnesota 37.9 33.2 18.7 Chippewa 34.0 40.6 24.5 St. Croix 30.3 36.3 24.1 Anoka 30.8 32.6 21.0 Wapsipinicon 53.1 33.4 21.9 McGregor 43.7 36.1 24.5 Winona 42.6 32.8 23.1 Clinton 50.9 35.1 22.4 Wisconsin 39.3 34.1 26.2 Figure 7. Location of USGS Stream Gauge Locations used in Calibration

December 7, 2021 Page 10 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved In the calibration process, sub-basins were initially parameterized using the aforementioned HEC-HMS model for most elements. However, for a continuous simulation model, additional elements were added (e.g., canopy, linear reservoirs, etc.). Prior experience with calibration and visually analyzing the timing and response of sub-basins to precipitation events allowed for initial parameter estimation. Subsequent model runs led to refinements of the parameters using observed streamflow data. The basins were calibrated starting at the headwaters and moving downstream to QCNGS. There is no streamflow gauge at QCNGS, so the final calibration location was the streamflow gage at Clinton, IA; approximately 11 miles upstream. It should be noted that not all stream gauges used for calibration were at the actual sub-basin outlet, creating some discrepancies in the simulated and observed flows. The HEC-HMS model also includes the ability to perform an Uncertainty Analysis. In this process, various parameters are allowed to vary within a specified range, randomly sampled from a prescribed probability distribution function. This allows the user to gain additional knowledge and confidence in the calibration process.

Representative calibration results of observed vs. simulated flows are provided in Figure 8 thru Figure 10.

The calibration process provided estimates of the various parameters to be used in the final simulations.

Calibration statistics (percent bias (%Bias) and Nash-Sutlcliff Estimator (NSE)) are provided in Table 5.

Based on several metrics and visual analysis, the 1988 water year resulted in a poor calibration, while water years 1991 and 1993 were both considered very good. As previously noted, the top three dry years were 1987, 1988, and 2009. The 1988 water year was in the middle of a relatively low four to five-year period where river flows were below normal. In fact, 1987 through 1989 were the lowest three-year period of river flow since 1980. This multi-year low flow period made the calibration more difficult.

Scaling issues were much more prevalent and exaggerated and groundwater or sub-surface flows were difficult to match. This ultimately resulted in poor agreement with the observed flows and a lack of confidence in using the calibrated parameters for any analysis. After considering the three (3) calibration years, it was decided to use the 1991 (average year) calibration data for the final analyses.

Table 5. Calibration Statistics for Water Years Water Year Percent Bias (Computed vs. Observed)

NSE Comments 1988 (Dry)

-8.01 %

0.251 Unacceptable (NSE <0.5 NSE) 1991 (Normal) 0.67%

0.823 Very Good (NSE > 0.75) 1993 (Wet)

-4.95%

0.865 Very Good (NSE > 0.75)

December 7, 2021 Page 11 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

Figure 9. Calibration Results @ Clinton, IA for 1991 Water Year (Normal)

December 7, 2021 Page 12 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved Figure 10. Calibration Results @ Clinton, IA for 1993 Water Year (wet) 4.4 HEC-HMS Thousand-Year Simulations As discussed in Section 3.0, thirty-six (36), 1,000-year daily weather sequences were created. The daily precipitation and snow water equivalent (SWE) were the meteorological inputs to the calibrated HEC-HMS model. The thirty-six (36) sequences or traces were divided in to three distinct categories (12 traces each): Normal (or current) climate, RCP45, and RCP85 (representing future potential climate scenarios).

Each of the 1,000-year traces were run through the HEC-HMS model, producing 1,000-year traces of daily flow values at the QCNGS, as well as all sub-basins within the watershed. The 1,000-year daily flow plots are cluttered and hard to read; however, zoomed in sections reveal the typical annual and seasonal cycles that are expected in the hydrologic response of this basin.

December 7, 2021 Page 13 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved Figure 11. Example of 1,000-year Daily Flow Output at QCNGS with Enhanced Area 5.0 FLOW-FREQUENCY ANALYSIS To compute the annual exceedance probabilities (AEPs), statistical analyses was conducted on an annual maximum series created from the 1,000-year HEC-HMS traces. For each climate scenario, the twelve (12) 1,000-year daily flow data sets were seamed together to create a single 12,000-year record.

The annual maximum flows for each of the 12,000 years were then extracted for use in the statistical analysis. An example of a single 1,000-year annual maximum series is illustrated in Figure 12.

December 7, 2021 Page 14 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved Figure 12. Example of Annual Maximum Flows for 1,000-year Trace at QCNGS Bulletin 17C (England, 2018) provides updated flow-frequency guidelines published in earlier Bulletins 17, 17A and 17B, and it addresses the limitations of the earlier versions and employs new statistical and computational approaches, including the Expected Moments Algorithm (EMA) to analyze available flood data in a single, uniform and consistent framework that does not require the introduction of additional algorithms to adjust the flood frequency curve to incorporate or account for the presence in the dataset of historic information, zero "ows, or low outliers as is the case with Bulletin 17B.

Bulletin 17C recommends the use of Pearson Type III distribution with log transformation of the flood data (log-Pearson Type III or LPIII) for defining the annual flood series and the method of moments with the EMA for estimating the parameters of the distribution. The detailed procedures, including approaches for plotting positions, flood distribution, parameter estimation, and EMA, are described in Bulletin 17C.

Therefore, the detailed procedures and corresponding mathematical expressions will not be described in this report.

Initially, the flood flow-frequency analysis was to be performed using the USACE HEC-SSP software (USACE, 2019), which is one of the two software packages recommended in Bulletin 17C. However, due to the size of the data sets, the USACE HEC-SSP was not used for the final analysis on the 12,000 years of data. A series of computer scripts were developed to accomplish all necessary data mining and analyses, including the log-Pearson Type III (LP-III or LP3) analysis. These scripts were tested multiple times against the HEC-SSP package using smaller data sets to confirm the accuracy and precision. The equations used in the scripts are based on several well documented approaches (Hosking J., 2015)

Figure 13 illustrates the AEP output analysis for the 12,000-year normal climate analysis. The tabular results are illustrated in Table 6 (where E represents the exponent of 10). The flow-frequency relationships in Figure 13 and Table 6 are considered the mean frequency curves for the normal climate scenario. The uncertainty analysis, discussed in Section 7.0, developed a family of curves above the mean curve to establish an upper-bound curve.

December 7, 2021 Page 15 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved Figure 13. Flow-Frequency Plot for 12,000-year "Normal Climate Scenario Table 6. Flow-Frequency Analysis Results for 12,000-year "Normal Climate" Scenario Return Period AEP Flow (cfs) 2 0.5 151,321 50 0.02 267,341 100 0.01 288,652 200 0.005 309,714 500 0.002 337,411 1,000 1E-03 358,376 10,000 1E-04 428,924 100,000 1E-05 501,931 1,000,000 1E-06 578,283

December 7, 2021 Page 16 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved 6.0 STAGE-FREQUENCY ANALYSIS The water surface elevations (WSELs) associated with the AEP flow values were determined from a stage-discharge rating curve that was created from a USACE one-dimensional unsteady HEC-RAS model developed for calculation QDC-0085-S-1991 (Exelon Nuclear, 2012), associated with the post-Fukushima flood hazard re-evaluation for QCNGS. Calculation QDC-0085-S-1991 provides additional details regarding the HEC-RAS model development and calibration. The rating curve is provided in Figure 14.

Figure 14. Rating Curve Developed for QCNGS All AEP flows were converted to WSEL using this rating curve. For reference, plant grade elevation is understood to be 594.5 ft mean sea level (MSL). Table 7 lists the WSELs that correspond to the various AEP flows. For ease of reference, all WSELs equal to or exceeding plant grade elevation are in red.

The WSELs are also illustrated in Figure 15.

Table 7. Water Surface Elevations Associated with AEP Flows Return Period (YR)

AEP NORMAL WSEL 2

0.5 578.4 50 0.02 584.6 100 0.01 585.6 200 0.005 586.6 500 0.002 587.8 1,000 1E-03 588.7 10,000 1E-04 591.2 100,000 1E-05 593.6 1,000,000 1E-06 595.9 575 580 585 590 595 600 605 0

100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000 900,000 Water Surface Elevation (ft)

Flow (cfs)

December 7, 2021 Page 17 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved Figure 15 - Water Surface Elevations Associated with AEP Flows 7.0 UNCERTAINTY ANALYSIS The discussion on uncertainties is addressed in two parts: 1) aleatory variability and 2) epistemic uncertainty. Aleatory variability refers to the natural and sampling variability in the hydrologic and/or hydraulic processes. Epistemic uncertainty refers to the uncertainty in the knowledge or modeling of the hydrologic and/or hydraulic processes.

7.1 Aleatory Variability 7.1.1 Climate Change Uncertainties related to the effects of climate change on the flood-frequency relationship is addressed by hydrologically simulating the three climate scenarios discussed in Section 3.1.

7.1.2 Variability in the Hydrologic Inputs An analysis was conducted to quantify variability in inputs used to simulate the hydrologic response.

Within the hydrologic modeling framework, there are many assumptions and parameters that may introduce errors in the simulated flows. The complexity of the continuous simulation model makes it difficult, if not impossible, to attribute overall errors in output to any one component. In addition, errors in some components may be compensated for by errors in other components.

The uncertainty analysis in HEC-HMS uses a Monte Carlo approach, in which multiple model simulations are completed allowing for various parameters to be altered for each simulation. An automatic sampling procedure is used for selected parameters within each simulation. Parameters were assumed to have a 578.0 580.0 582.0 584.0 586.0 588.0 590.0 592.0 594.0 596.0 598.0 1.0E-06 1.0E-05 1.0E-04 1.0E-03 1.0E-02 1.0E-01 1.0E+00 WSEL Annual Exceedance Probability

December 7, 2021 Page 18 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved normal distribution within a given range. The range of each parameter was partially based on values obtained in the calibration phase. To determine the most appropriate parameters for use in the uncertainty analysis, a series of simulations were performed on the original calibration analyses.

Parameters were tested individually and analyzed for sensitivity in the resulting peak flows and total runoff volumes. The parameters in Table 8 were selected for the final uncertainty analyses based on their potential to impact the final outcomes.

Table 8. HEC-HMS Parameter and Sub-Basin Components Used in Uncertainty Analysis Parameter Sub-Basin Component Groundwater Coefficient 1 Baseflow Groundwater Coefficient 2 Baseflow Groundwater Coefficient 3 Baseflow Constant Loss Rate Loss (Deficit & Constant)

Maximum Deficit Loss (Deficit & Constant)

Percent Impervious Loss (Deficit & Constant)

A minimum of twelve (12) samples were run for each uncertainty analysis with each sub-basin being treated independently. Each sub-basin was sampled independently with no dependence on any other sub-basin. The only output for each uncertainty analysis was flow at QCNGS. The minimum, mean, and maximum value of flow for each day and each simulation were recorded. Figure 16 illustrates a typical analysis and the responses. Using the same data analysis and statistical approach as previously described, the AEPs were calculated for the maximum (green) flow values in each uncertainty analysis.

For AEP analyses, only the maximum output was used.

Figure 16. Example of Uncertainty Analysis Output Illustrating Min, Max, and Mean Daily Flows

December 7, 2021 Page 19 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved 7.1.3 Natural Variation in River System The flood events used in the unsteady HEC-RAS hydraulic model calibration (which is the basis for the stage-discharge relationship discussed in Section 6.0) were both cool-season (early spring) and warm season events, with varying vegetation growth levels, so the Manning n-values should represent a full spectrum of seasonal roughness conditions. Furthermore, the Mississippi River is a navigable waterbody that is dredged to maintain a navigable channel; therefore, channel system is considered stable with negligible hydraulic effects of bathymetric changes. See calculation QDC-0085-S-1991 (Exelon Nuclear, 2012).

7.1.4 Land Use Changes A study of hydrologic trends in the Upper Mississippi River watershed (Knapp, 1994) concluded that while hydrologic modifications impacting average flows and flooding are detected for the major tributaries, there is relatively little impact on the flows in the Mississippi River. These results suggest that watershed scale has a significant influence on the impacts of human-induced modification. Watershed changes must be widespread to cause modification in large river flows, and these modifications may often be attenuated by the substantial storage associated with large watersheds. But two impacts are noted.

Reforestation in Wisconsin appears to have a mild impact on average flow in the Mississippi River at Clinton, Iowa and flood control reservoirs in the Missouri River watershed appear to produce a 10 percent reduction in the average flood peak and flood volume for the Mississippi River at St. Louis, Missouri.

The study further concluded that most of the changes in streamflow are directly associated with climate variability with a strong correlation between average annual precipitation and average annual streamflow.

Therefore, land use changes were considered to have negligible effects on extreme floods along the Mississippi River.

7.2 Epistemic Uncertainty 7.2.1 Probability Distribution Function Uncertainty in the LPIII representation of the streamflow-frequency relationship was addressed by applying alternative probability distribution functions to the streamflow data set and using moments of the distributions to test goodness-of-fit. Moments characterize the shape of the probability distribution function and include mean (centroid), standard deviation or variance (dispersion about the mean),

skewness (measure of the asymmetry), and kurtosis (measure of sharpness of the probability distribution). Moments can be estimated using sample data but are, by nature, not unbiased. Hosking and Wallis (Hosking J. a., 1997) indicate that skewness and kurtosis may be severely biased and have bounds that depend on sample size. Hosking and Wallis (Hosking J. a., 1997) propose using L-moments to provide an alternative system of describing the shapes of probability distributions and arose as modifications of the probability weighted moments.

Multiple 1,000-year traces were analyzed for goodness-of-fit using the computer program Hydro-FIT, developed by HydroMetriks, LLC. The L-skewness vs L-kurtosis plot for a typical 1,000-year trace is illustrated in Figure 17. Each probability distribution has a known relation between L-skewness and L-kurtosis, shown on Figure 17 as lines for 3-parameter distributions (location, scale, and shape) and points for 2-parameter distributions (location and scale). The lines and points are compared to the L-skewness and L-kurtosis from the site/sample data for each distribution to assess goodness-of-fit. Note that the exact relationship between L-skewness and L-kurtosis for the 4-parameter distributions is complicated and difficult to plot and, therefore, is not presented in this study. The plot indicates that Log-Normal (LNO), Generalized Extreme Value (GEV), and Pearson Type III (PE3) are all consistent and acceptable.

Therefore, it was concluded that the LP-III distribution used in the Bulletin 17C/HEC-SSP analysis

December 7, 2021 Page 20 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

Figure 17. L-Skewness vs. L-Kurtosis for a typical 1,000-year Simulation 7.2.2 Potential Error in Stage-Discharge Relationship Flow data were converted to WSELs using a stage-discharge relationship developed from a calibrated unsteady flow one-dimensional HEC-RAS model, created for the Probable Maximum Flood (PMF) water surface profile in Calculation QDC-0085-S-1991 (Exelon Nuclear, 2012). The HEC-RAS model was calibrated against the five largest floods since the 1900s that resulted in the five highest peak WSELs at QCNPS, as follows: 1965, 2001, 1993, 1965, and 1997. The range of differences between the modeled and observed results was +0.7 ft to -0.5 ft. The HEC-RAS hydraulic model was also tested for sensitivity to downstream boundary condition. The results of the sensitivity analysis showed a minimum effect of the downstream boundary condition on the peak WSEL. Therefore, since the median error in the HEC-RAS model is negligible (at approximately +0.1 ft), no further adjustments were made to the stage-discharge relationship.

7.2.3 Event Combinations Two potential event combinations with precipitation-induced flooding along the Mississippi River were considered: upstream dam failure and wind-wave runup. A detailed probabilistic assessment of upstream dam failure was not conducted. Instead, the post-Fukushima dam breach calculation QDC-0085-S-2032 (Exelon Nuclear, 2012) was reviewed to assess the potential effects of dams failing concurrently with a large flood. With over 1,500 dams in the watershed upstream of QCNGS, the possible flood-dam failure event combinations are enormous. Consequently, calculation QDC-0085-S-2032 considered two approaches to evaluating upstream dam failure: 1) failure of a subset of dams within the watershed (screening dams that are small and/or remote) and 2) assume all dams fail in the watershed by lumping a sub-set of smaller upstream dams into larger hypothetical dams. For this study, Approach 2 was not

December 7, 2021 Page 21 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved considered plausible, even for a 10-6 AEP flood at QCNGS, primarily because the watersheds to each of the 1,500+ dams (excluding locks and dams located along the main stem) would, individually, not experience a flood of the same magnitude as at QCNGS. Following Approach 1, calculation QDC-0085-S-2032 identified a failure scenario of larger dams and dams near QCNGS that could influence flood levels: domino (or cascading) failure of Lock and Dams No. 11, 12, and 13, and individual failures of Lake Carroll Dam, Apple Canyon Lake Dam, Smallpox Creek Dam, and Eau Galle Reservoir Dam. This dam failure scenario produces a 0.4 ft and 0.7 ft increase in WSEL during the PMF and 1/2 PMF, respectively, at QCNGS. Therefore, it was concluded that simultaneous dam failure during a precipitation-induced flood could produce an increase in flood depth at QCNGS of between 0.0 and 0.7 ft for very low AEP flood events.

Another potential event combination with the Mississippi River flooding are the effects of wind-generated waves. Calculation QDC-0085-S-2032 (Exelon Nuclear, 2018) produced estimates of deepwater (fetch-limited) wave heights in the main river channel. However, with stillwater elevations less than 2 feet above plant grade at a 10-6 AEP, deepwater waves will break and become depth-limited at the plant. A depth-limited wave height is commonly understood to be approximately 0.8 x flood depth. Therefore, wave runup is expected not to exceed 0.8 foot.

7.3 Summary of Approach to Quantifying Uncertainties In assessing the variability and uncertainty factors described above, the approach to quantifying uncertainties included developing AEPs for three climate scenarios (Section 7.1.1) and a Monte Carlo analysis of the hydrologic inputs (Section 7.1.2).

Simulations were performed for the following climate scenarios:

1. Normal Climate Scenario - Simulations assuming no climate change

2. RCP45 Climate Scenario - Simulations assuming moderate climate change

3. RCP85 Climate Scenario - Simulations assuming more severe climate change

4. RCP45 Climate Scenario Enhanced Evapotranspiration - Additional Evapotranspiration (ET) was added earlier in the year to account for slightly increased temperatures.

5. RCP85 Climate Scenario Enhanced Evapotranspiration - Similar to RCP 45 analysis, additional ET was added earlier in the year.

The Monte Carlo analysis was performed within HEC-HMS by running a minimum of 12 samples of the normal distribution function for each basin and variable (listed below).

1. Groundwater 1 Coefficient

2. Groundwater 2 Coefficient

3. Groundwater 3 Coefficient

4. Constant Loss Rate

5. Percent Impervious

6. Maximum Canopy Storage The maximum daily flow output values at QCNGS were used for the statistical analysis to compute flow-frequencies for a large number of scenarios. Uncertainty was further addressed by testing the accuracy of the results at higher AEPs (e.g., 0.01) that can be reliably quantified using observed annual maximum stream flow records, fitted to a LPIII probability distribution, based on methods in Bulletin 17C (England, 2018).

December 7, 2021 Page 22 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved 7.4 Results of Uncertainty Analysis Figure 18, Figure 19, Table 9, and Table 10 illustrate the outcomes (in both flows and WSEL) of the five (5) climate scenarios, with the normal climate simulation producing the highest flow rates at lower AEPs.

The RCP45 and RCP85 generally produced smaller values. This is primarily due to these climate scenarios producing precipitation and subsequent melt earlier in the year. In effect, while the atmosphere is getting slightly warmer, the precipitation has the effect of being more spread out and with fewer large peaks. As expected, the Enhanced ET simulations show an even further decrease in the AEP values.

The increased ET has the effect of removing more precipitation from the runoff, thus causing lower daily flows. After evaluating these simulation outcomes, it was decided to remove the RCP45, RCP85, and Enhanced ET outcomes from further analyses.

Figure 18. Annual Exceedance Probability Flow Rates (cfs) for Normal, RCP45, RCP85, &

Enhanced ET Analyses Table 9. Annual Exceedance Probability Flow Rates (cfs) for Normal, RCP45, RCP85, & Enhanced ET Analyses AEP NORMAL RCP45 RCP85 RCP45 ET RCP85 ET 0.5 151,321 147,592 151,296 131,722 133,125 0.02 267,341 258,004 268,076 239,012 237,737 0.01 288,652 277,665 288,978 258,745 256,725 0.005 309,714 296,926 309,479 278,242 275,423 0.002 337,411 322,009 336,208 303,869 299,906 1E-03 358,376 340,817 356,271 323,254 318,360

December 7, 2021 Page 23 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved AEP NORMAL RCP45 RCP85 RCP45 ET RCP85 ET 1E-04 428,924 403,059 422,777 388,387 379,985 1E-05 501,931 465,916 490,086 455,596 443,024 1E-06 578,283 530,148 558,989 525,652 508,208 Figure 19. Annual Exceedance Probabilities WSELs for Normal, RCP45, RCP85, & Enhanced ET Analyses Table 10. Annual Exceedance Probabilities WSELs for Normal, RCP45, RCP85, & Enhanced ET Analyses AEP NORMAL RCP45 RCP85 RCP45 ET RCP85 ET 0.5 578.4 578.2 578.4

N/A

N/A 0.02 584.6 584.2 584.6 583.2 583.2 0.01 585.6 585.1 585.6 584.2 584.1 0.005 586.6 586.0 586.6 585.1 585.0 0.002 587.8 587.1 587.8 586.3 586.1 1E-03 588.7 588.0 588.6 587.2 587.0 1E-04 591.2 590.3 591.0 589.7 589.4 1E-05 593.6 592.5 593.2 592.1 591.7 1E-06 595.9 594.5 595.3 594.4 593.8

December 7, 2021 Page 24 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved Using the normal climate scenario, the remaining uncertainty simulations were then performed in a similar manner. Figure 20 and Table 11 illustrate the results of these AEP analyses. Note that the AEP was only conducted on the annual maximum series for the higher flow curves in the Monte Carlo simulation (as shown in Figure 16). Therefore, Figure 20 and Table 11 only represent the mean (for the normal climate scenario), the family of curves above the mean, and the upper bound curve. The family of curves below the mean (for the normal climate scenario) and the lower bound curve were not developed.

The rather tight family of curves provides additional confidence the overall analysis. In the tabular results, the values of the 10-6 flow ranges from 578,283 cfs (mean) to 600,161 cfs (upper bound).

Figure 20. Annual Exceedance Probabilities Values (cfs) for Monte Carlo Simulations Table 11. Annual Exceedance Probabilities Values (cfs) for Monte Carlo Simulations AEP NORMAL GW1 GW2 GW3 Constant Loss

% Imp Max Canopy Storage 0.5 151,321 155,054 156,702 152,159 151,958 158,397 153,579 0.02 267,341 274,374 276,038 268,348 270,359 279,115 269,801 0.01 288,652 296,393 297,865 289,671 292,322 301,222 291,390 0.005 309,714 318,184 319,413 310,742 314,088 323,054 312,801 0.002 337,411 346,882 347,713 338,444 342,797 351,737 341,066 1E-03 358,376 368,636 369,107 359,409 364,589 373,431 362,541 1E-04 428,924 442,023 440,948 429,934 438,294 446,323 435,300 1E-05 501,931 518,246 515,061 502,883 515,122 521,595 511,361 1E-06 578,283 598,241 592,345 579,146 596,021 600,161 591,693 Using the rating curve in Figure 14, flow rates in Table 11 were converted to WSELs. For ease of reference, all WSELs that equal or exceed plant grade elevation are in red.

December 7, 2021 Page 25 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved Figure 21. Annual Exceedance Probabilities Values (WSELs) for Monte Carlo Simulations Table 12. Annual Exceedance Probability Water Surface Elevations for Monte Carlo Simulations AEP NORMAL GW1 GW2 GW3 Constant Loss

% Imp Max Canopy Storage 0.5 578.4 578.7 578.8 578.5 578.5 578.9 578.6 0.02 584.6 584.9 585.0 584.7 584.7 585.1 584.7 0.01 585.6 585.9 586.0 585.6 585.8 586.2 585.7 0.005 586.6 587.0 587.0 586.6 586.8 587.2 586.7 0.002 587.8 588.2 588.3 587.9 588.1 588.4 588.0 1E-03 588.7 589.0 589.1 588.7 588.9 589.2 588.8 1E-04 591.2 591.7 591.6 591.2 591.5 591.8 591.4 1E-05 593.6 594.2 594.1 593.7 594.1 594.3 593.9 1E-06 595.9 596.5 596.3 595.9 596.4 596.5 596.3 The flow-frequency relationship produced from this study compares well at higher AEP flows with those produced by a LPIII analysis from observed annual maximum flows (published by the USGS at streamflow gage station 05420500 at Clinton, IA). See the flood-frequency analysis in calculation number QDC-0085-S-2332 (Exelon Nuclear, 2018). The comparison in Table 13 shows that flows are comparable, with flows from the current study being consistently higher. Ignoring historical observations and perception thresholds for the paleo-period, the systematic annual maximum streamflow record used in calculation QDC-0085-S-2332 spans 155 years. Flow-frequencies are considered reliable for return periods approximately equal to the period of record. Therefore, good agreement between flows from the

December 7, 2021 Page 26 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved current study and calculation QDC-0085-S-2332 at return periods of 200 years and less confirms that the flow-frequency relationship produced by the current study is reliable at higher frequencies.

Table 13. Flow Comparisons at Higher Frequencies Return Period AEP Flow (cfs) from Current Study Flow (cfs) from Calculation QDC-0085-S-2332 Difference

(%)

2 0.5 151,321 143,000 5.5 50 0.02 267,341 257,000 3.9 100 0.01 288,652 274,000 5.1 200 0.005 309,714 290,000 6.4 500 0.002 337,411 309,000 8.4

8.0 CONCLUSION

S The following conclusions were derived from the flood-frequency analysis at QCNGS:

The current (normal) climatology generates the highest flows for very low AEPs. Two representative RCPs (a greenhouse gas concentration trajectory) studied, with and without Enhanced Evapotranspiration, produce smaller flow rates particularly at very low AEPs. From this, it can be deduced that a warmer atmosphere may produce more frequent high-intensity and short-duration rainfall events but, with melt occurring earlier in the year, less severe floods in the Mississippi River.

The mean flood-frequency curve is conservative since it only included the normal climate scenario, which produced the highest flows. Introducing flows from the other climate scenarios would have pulled the mean for the overall family of curves lower.

Monte Carlo simulations for key hydrologic inputs showed low variability in peak flow rates at QCNGS. For example, at the 10-6 AEP, the full range of flows is 21,900 cfs; only 3.7% of the normal climate simulation (corresponding to a difference in WSEL of 0.6 foot).

In evaluating adequate margin, the cumulative effect of flood event combinations (i.e., upstream dam failure and wind-generated waves), approximately between 0 and 1.5 ft, should be considered.

The flood-frequency analysis produced by this study agrees well with the Bulletin 17C analysis of observed annual maximum flows at higher AEPs, with the results from this study being slightly higher; indicating that the results are conservative.

It is worth noting how the flow-frequency results from the current study compares to the PMF peak flow rate computed in calculation QDC-0085-S-1990 (Exelon Nuclear, 2012), associated with the post-Fukushima Flood Hazard Reevaluation for QCNPS. Calculation QDC-0085-S-1990 produced peak flow rates at QCNPS for two PMF scenarios:

o All-Season PMP with 500-year Antecedent Rainfall (552,000 cfs) o Cool-Season PMP with 100-year Snowpack (745,000 cfs)

December 7, 2021 Page 27 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

©2021 All Rights Reserved The results from this study shown in Table 11 shows that the All-Season PMP produces flows with AEPs approximately between 10-5 and 10-6. Whereas, by visually extrapolating the plot on Figure 20, the Cool-Season PMP/100-year Snowpack combination has an AEP less than 10-7.

9.0 IDENTIFICATION OF COMPUTER PROGRAMS

1. Microsoft Excel 2019 MSO (16.0.10377.20023

2. USACE HEC-HMS version 4.8

3. USACE HEC-SSP version 2.1

4. HydroMetriks Frequency Intensity Tool Hydro-FIT

10.0 REFERENCES

Alword, J. a. (1915). Report of the Rivers and Lakes Commission on the Illinois River and Its Bottom Lands.

Daly, C. T. (1997). The PRISM Approach to Mapping Precipitation and Temperature. 10th Conference on Applied Climatology, American Meteorology Society, 10-12. Reno, NV.

England, J. J. (2018). Guidelines for Determining Flood Flow Frequency - Bulletin 17C: U.S. Geological Survey Techniques and Methods, Book 4, Chap. B5, 143 p. U.S. Geological Survey. Retrieved from https://doi.org/103133/tm4B5.

Exelon Nuclear. (2012). Calculation of Probable Maximum Flood Water Surface Elevation: Evaluation of Riverine Hydraulics of the Upper Mississippi River at QCNGS, Analysis No. QDC-0085-S-1991, Revision 0.

Exelon Nuclear. (2012, December 18). Probable Maximum Flood (PMF) for the Upper Mississippi River Watershed Contributory to QCNGS, Analysis No. QDC-0085-S-1990, Revision 0.

Exelon Nuclear. (2012). Upstream Dam Failure Flood Evaluation at QCNGS, Analysis No. QDC-0085-S-2032, Revision 0.

Exelon Nuclear. (2013). Combined Events Flood Assessment, Analysis No. QDC-0085-S-2034, Revision

0.

Exelon Nuclear. (2018, May). Flood Frequency Analysis per NEI 16-05, Analysis No. QDC-0085-S-2332, Revision 0.

Hosking, J. (2015). L-moments. R package, version 2.5. Retrieved from http://CRAN.R-project.org/package=lmom Hosking, J. a. (1997). Regional Frequency Analysis, An Approach Based on L-Moments. 224. Cambridge University Press.

Hultstrand, D. a. (2017). The Storm Precipitation Analysis System (SPAS) Report. Nuclear Regulatory Commission (NRC) Inspection Report No 99901474/2016-201.

IPCC. (2017). Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5)

Observed Climate Change Impacts Database, Version 2.01. Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC). https://doi.org/10.7927/H4FT8J0X.

Kappel, W. H. (2012). Site-Specific Probable Maximum Precipitation (PMP) Study for the Quad Cities Nuclear Generating Station. Quad Cities, IA.

Kappel, W. H. (2014). Site-Specific Local Intense Precipitation (LIP) Study for the Quad Cities Nuclear Station.

December 7, 2021 Page 28 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

199-206. DOI: 10.1080/02508069408686230.

Knox, J. (1996). Late Quartenary Upper Mississippi River Alluvial Episodes and their Significance to the Lower Mississippi River System. Engineering Geology, 45, 263-285.

Munoz, S. e. (2015). Cahokias Emergence and Decline Coincided with Shifts of Flood Frequency on the Mississippi River,. PNAS, 112(20), 6319-6324.

Nuclear Energy Institute. (2016). External Flooding Assessment, NEI 16-05 [Revision 1].

State of Illinois. (1929, June). Flood Control Report. An Engineering Study of the Flood Situation in the State of Illinois.

University of Illinois. (2017). Illinois State Water Survey, State Climatologist Office for Illinois. Retrieved from http://www.isws.illinois.edu/atmos/statecli/Roses/wind_climatology.htm USACE. (1999, July). Cold Regions Research & Engineering Laboratory. Flow Control to Manage River Ice.

USACE. (2000, March). Hydrologic Engineering Center, Hydrologic Modeling System (HEC-HMS).

Technical Reference Manual.

USACE. (2003, April). Application of Paleohydrology to Corps Flood Frequency Analysis, RD-47.

USACE. (2004, January). Upper Mississippi River System Flow Frequency Study, Final Report.

USACE. (2015, September 30). Coastal Engineering Manual. EM 1110-2-1100 (Part II), Change 4.

USACE. (2017). National Levee Database. Retrieved from http://nld.usace.army.mil/egis/f?p=471:32:7650842244696::NO USACE. (2019, June). HEC-SSP Statistical Software Package, Version 2.2. Hydrologic Engineering Center.

USDOI. (1936). Floods in the United States, Magnitude and Frequency. Water-Supply Paper 771.

USGS. (2004). Estimating Flood-Peak Discharge Magnitudes and Frequencies for Rural Streams in Illinois. Scientific Investigations Report 2004-5103.

USGS. (2017). National Water Information System: Web Interface, USGS 05420500 Mississippi River at Clinton, IA, Surface-Water: Peak streamflow. Retrieved from https://nwis.waterdata.usgs.gov/ia/nwis/peak/?site_no=05420500&agency_cd=USGS

December 7, 2021 Aterra Solutions, LLC Quad Cities - Probabilistic Flood Hazard Assessment Report

Quad Cities Multi-Site Stochastic Daily Weather Simulation Report

PO Box 175 Monument, CO 80132 (719) 488-4311 http://www.appliedweatherassociates.com Quad Cities Multi-Site Stochastic Daily Weather Simulation Report Prepared for Prepared by Applied Weather Associates, LLC Monument, Colorado Doug Hultstrand, Ph.D., Senior Hydrometeorologist Bill Kappel, Chief Meteorologist/ Project Manager Kristi Steinhilber, Staff Meteorologist November 2021

NOTICE This report was prepared by Applied Weather Associates (AWA). The results and conclusions in this report are based upon our best professional judgment using currently available data. Due to the uncertainty associated with this type of work, neither AWA nor any person acting on behalf of AWA can (a) make any warranty, express or implied, regarding future use of any information or method shown in the report or (b) assume any future liability regarding use of any information or method contained in the report. The results contained in this report are based on the professional judgment of the experts in this subject field at AWA. The included report is conservative and accurate to the best of our knowledge at the time of its preparation based on available information, methodology, and data.

Stochastic Weather Generation Report Page 3 of 24 11/3/2021 Applied Weather Associates TABLE OF CONTENTS ACRONYMS and ABBREVIATIONS....................................................................................... 5

1.0 INTRODUCTION

............................................................................................................... 6 2.0 STUDY LOCATION........................................................................................................... 6 3.0 REGIONAL STOCHASTIC WEATHER GENERATION............................................ 9 3.1 Precipitation and Temperature Data........................................................................................... 11 3.2 Precipitation and Temperature Model Calibration...................................................................... 11 3.3 1000-year Simulation.................................................................................................................. 12 3.4 Reformat Script........................................................................................................................... 12 4.0 LONGTERM CALIBRATION........................................................................................ 13 5.0 SNOW WATER EQUIVALENT MODELLING........................................................... 15 6.0 CLIMATE CHANGE........................................................................................................ 18 6.1 1000-year Simulation.................................................................................................................. 20 7.0 RESULTS AND CONCLUSIONS................................................................................... 20

8.0 REFERENCES

................................................................................................................... 22 LIST OF FIGURES Figure 1: Daily meteorological stations across the Quad Cities basins (blue), stations selected for stochastic model calibration and simulation are shown in red....................................................... 7 Figure 2: Example of the Anoka basin VAR models accounting for seasonally changing weather variables of precipitation (a); maximum temperature (b); and minimum temperature (c)........... 10 Figure 3: Example of the ten subbasins precipitation Q-Q plot.................................................... 12 Figure 4: Example of McGregor basin 100-year simulated data from stochastic simulation number 6..................................................................................................................................................... 13 Figure 5: Water years investigated for model calibration, i) top row represents the three wet years investigated 1986, 1993, and 2001; ii) bottom row represents the three dry years investigated 1987, 1988, and 2009. Image shows precipitation in inches.................................................................. 14 Figure 6: Normal water year (1991) investigated for model calibration, precipitation in inches. 15 Figure 7: Example of simulated SWE calibration for the McGregor subbasin for a) water year 1988, b) water year 1991, and c) water year 1993........................................................................ 17 Figure 8: Representative Concentration Pathway (RCP) Trajectories from IPCC AR5.............. 18 Figure 9: Quad Cities basin monthly climate normal data compared to future projection climate normal data based on RCP45 and RCP85 climate projections..................................................... 19 Figure 10: The monthly change in temperature and precipitation from historical period (observation period) to climate projections based on RCP45 and RCP85 scenarios................... 20

Stochastic Weather Generation Report Page 4 of 24 11/3/2021 Applied Weather Associates LIST OF TABLES Table 1: List of stations used for regional stochastic weather analysis.......................................... 8 Table 2. Water year average sub-basin total precipitation for each calibration year investigated.

........................................................................................................ Error! Bookmark not defined.

Stochastic Weather Generation Report Page 5 of 24 11/3/2021 Applied Weather Associates ACRONYMS and ABBREVIATIONS AWA Applied Weather Associates POR Period of Record Ppt Precipitation RMAWGEN Multi-Site Auto-Regressive Weather GENerator Tmax Maximum Temperature Tmin Minimum Temperature QC Quality Control RCP Representative Concentration Pathway SWE Snow Water Equivalent VAR Vector AutoRegressive Model

Stochastic Weather Generation Report Page 6 of 24 11/3/2021 Applied Weather Associates

1.0 INTRODUCTION

AWA previously completed a site-specific Probable Maximum Precipitation (PMP) (Kappel et al.,

2012) and Local Intense Precipitation (LIP) analysis (Kappel et al., 2014) for this Quad Cities Nuclear Station basin. Therefore, AWA has a detailed understanding of the weather and climate of the region, spatial and seasonal variations of important meteorological variables, and has evaluated meteorological data sets utilized in this analysis. AWA extracted and analyzed daily precipitation and daily maximum and minimum temperature data for input into a stochastic weather model, RMAWGEN: Multi-Site Auto-Regressive Weather GENerator (Cordano and Eccel, 2016; Cordano and Eccel, 2017). The RMAWGEN model is used for spatial multi-site stochastic generation of daily time series of temperature and precipitation. The model makes use of Vector AutoRegressive models (VARs). The weather generator model is then saved as an object and is calibrated by daily instrumental "Gaussian" time series through the 'vars' package tools (Cordano and Eccel, 2016; Cordano and Eccel, 2017).

2.0 STUDY LOCATION The Quad Cities basin, approximately 88,600 mi2, is located in the upper midwest region of the United States covering much of Wisconsin and Minnesota and small portions of South Dakota, Iowa, and Illinois (

Figure 1). Important for stochastic weather modeling, this region is covered with a large number of meteorological observations, many with long period of record. An investigation of daily meteorological stations available resulted in more than 250 daily stations initially identified within the basin (Figure 1). The topography and meteorological characteristics across the basin can be considered generally homogenous, with minimal climatological variations caused by topography from south to north and east to west. The large distance across the basin generally results in temperatures cooling from south to north, annual precipitation decreasing from south to north, and annual snowfall increasing from south to north.

To reduce computational time, while still providing adequate data to realistically model the spatial correlations of daily rainfall, temperature, and snow water equivalent (SWE), the initial list of stations identified were consolidated into representative list. This resulted in 49 station locations shown in Figure 1 and listed in Table 1. These stations provided good spatial representation from north to south and east to west while also providing a consistent and long period of record. These locations were chosen based on data availability, data quality, period of record, and spatial distribution requirements. Several discussions occurred to confirm the data were sufficient to produce a high quality, long duration meteorological data output required for detailed stochastic weather generation and provided the input needed for hydrologic evaluations.

Stochastic Weather Generation Report Page 7 of 24 11/3/2021 Applied Weather Associates Figure 1: Daily meteorological stations across the Quad Cities basins (blue), stations selected for stochastic model calibration and simulation are shown in red.

Stochastic Weather Generation Report Page 8 of 24 11/3/2021 Applied Weather Associates Table 1: List of stations used for regional stochastic weather analysis StationID Name Latitude Longitude Elevation USC00113312 GALENA, IL 42.3995

-90.3860 822 USC00115901 MOUNT CARROLL, IL 42.0980

-89.9841 640 USC00131635 CLINTON NUMBER 1, IA 41.7947

-90.2639 585 USC00131954 CRESCO 1 NE, IA 43.3894

-92.0938 1255 USC00132235 DE WITT, IA 41.8108

-90.5405 685 USC00132864 FAYETTE, IA 42.8826

-91.8303 1145 USC00135131 MAQUOKETA, IA 42.0771

-90.6647 668 USC00135952 NEW HAMPTON, IA 43.0452

-92.3123 1148 USC00210939 BRAINERD, MN 46.3433

-94.2086 1180 USC00210981 BRICELYN, MN 43.5440

-93.8423 1170 USC00211227 CAMBRIDGE, MN 45.5565

-93.2348 965 USC00211263 CANBY, MN 44.7184

-96.2696 1210 USC00211691 COLLEGEVILLE ST. JOHN S, MN 45.5794

-94.3920 1210 USC00212721 FARIBAULT, MN 44.3092

-93.2660 970 USC00213311 GRANITE FALLS, MN 44.8135

-95.5517 930 USC00213567 HASTINGS DAM 2, MN 44.7597

-92.8689 680 USC00214438 LAKE CITY, MN 44.4363

-92.2793 700 USC00214652 LEECH LAKE DAM, MN 47.2467

-94.2228 1302 USC00214861 LONG PRAIRIE, MN 45.9646

-94.8892 1340 USC00215598 MOOSE LAKE 1 SSE, MN 46.4378

-92.7578 1110 USC00215615 MORA, MN 45.8776

-93.3147 1018 USC00215638 MORRIS WEST CENTRAL, MN 45.5901

-95.8745 1140 USC00217460 SANDY LAKE DAM LIBBY, MN 46.7953

-93.3211 1234 USC00470239 ANTIGO, WI 45.1604

-89.1130 1519 USC00470855 BLACK RIVER FALLS SEWAGE, WI 44.2903

-90.8538 810 USC00471578 CHIPPEWA FALLS, WI 44.9277

-91.4081 820 USC00471923 CUMBERLAND, WI 45.5331

-92.0226 1240 USC00473405 HANCOCK EXPERIMENTAL FARM, WI 44.1187

-89.5359 1092 USC00473511 HAYWARD RANGER STATION, WI 46.0002

-91.5074 1200 USC00474383 LAC VIEUX DESERT, WI 46.1211

-89.0758 1760 USC00474391 LADYSMITH 3 W, WI 45.4651

-91.1236 1130 USC00474546 LANCASTER 4 WSW, WI 42.8278

-90.7889 1040 USC00475164 MATHER 3 NW, WI 44.1746

-90.3482 978 USC00475178 MAUSTON 1 SE, WI 43.7899

-90.0597 865 USC00475364 MERRILL, WI 45.1785

-89.6615 1250 USC00475516 MINOCQUA, WI 45.8865

-89.7322 1601 USC00475563 MONDOVI, WI 44.5647

-91.6719 830 USC00475808 NEILLSVILLE 3 ESE, WI 44.5378

-90.5350 1080 USC00476398 PARK FALLS DNR HQ, WI 45.9336

-90.4506 1525 USC00476827 PRAIRIE DU CHIEN, WI 43.0515

-91.1349 658 USC00477092 REST LAKE, WI 46.1222

-89.8772 1612 USC00477113 RHINELANDER, WI 45.5985

-89.4508 1572 USC00477226 RIVER FALLS, WI 44.8544

-92.6122 940 USW00014920 LA CROSSE, WI 43.8788

-91.2527 652 USW00014922 MINNEAPOLIS ST. PAUL, MN 44.8831

-93.2289 872 USW00014925 ROCHESTER INTERNATIONAL, MN 43.9041

-92.4916 1304 USW00014992 REDWOOD FALLS MUNICIPAL, MN 44.5483

-95.0804 1021 USW00094967 PARK RAPIDS MUNICIPAL, MN 46.9006

-95.0678 1434

Stochastic Weather Generation Report Page 9 of 24 11/3/2021 Applied Weather Associates 3.0 REGIONAL STOCHASTIC WEATHER GENERATION Stochastic weather generators are statistical models that can simulate realistic or plausible random sequences of atmospheric variables such as temperature and rainfall (Wilks and Wilby, 1999). The stochastic weather generators attempt to reproduce the spatial and temporal dynamics and correlation structures of the variables of interest (Ailliot et al., 2015). The synthetic sequences provide a set of alternate realizations that can be used for risk and reliability assessment in the design and operation of agricultural, water resource and environmental systems (Mehrotra et al.,

2006).

The analysis utilized a multisite stochastic modeling approach using daily observations of precipitation, minimum and maximum temperatures, and SWE from the 49 sites located across the basin. Stochastic weather modeling utilized the Multi-site Auto-regressive Weather GENerator (RMAWGEN)) framework. The R package RMAWGEN was built to generate daily temperature and precipitation time series at several sites by using the theory of VARs (Cordano and Eccel, 2016; Cordano and Eccel, 2017). RMAWGEN was selected for the Quad Cities basin because it is able to maintain the temporal and spatial correlations among the station data series (Cordano and Eccel, 2017). In particular, observed time series of daily maximum and minimum temperature and precipitation are used to calibrate the parameters of a VAR model. The VAR model, coupled with monthly mean weather variables (i.e., station mean, climate mean and/or climate projection means), can be used to generate stochastic daily scenarios.

The structure of the RMAWGEN package consists of functions that transform precipitation and temperature time series into Gaussian-distributed random variables through deseasonalization and Principal Component Analysis (Cordano and Eccel, 2016; Cordano and Eccel, 2017). VAR models are calibrated on transformed time series. The time series generated by VAR are then inversely re-transformed into precipitation and/or temperature series. For seasonally-changing weather variables such as temperature and precipitation in this region, the Gaussian process can take into account its dependency on a monthly or season time frame. For example, Figure 2 illustrates the monthly seasonalized precipitation and temperature for the Anoka basin.

For this study, daily weather time series recorded at 49 locations were used to calibrate the RMAWGEN model to the period 1949-2019, the calibrated precipitation and temperature models were used to simulation ten 1000-years of daily precipitation, maximum temperature, and minimum temperature. The data methods used are listed in the following sub-sections.

Stochastic Weather Generation Report Page 10 of 24 11/3/2021 Applied Weather Associates Figure 2: Example of the Anoka basin VAR models accounting for seasonally changing weather variables of precipitation (a); maximum temperature (b); and minimum temperature (c).

Stochastic Weather Generation Report Page 11 of 24 11/3/2021 Applied Weather Associates 3.1 Precipitation and Temperature Data The datasets used in this study include daily total precipitation (Ppt), and minimum and maximum temperatures (Tmin; Tmax) for the 1949-2019 period from a network of 49 stations, obtained from NOAA Global Historical Climatology Network Daily(GHCN) (

Figure 1).

3.2 Precipitation and Temperature Model Calibration Daily precipitation was generated for the reference period 1949-2019 through a random generation with an auto-regression based on generalized linear models (Chambers and Hastie, 1992) implemented in the RMAWGEN package following Wilks approach for spatial (inter-station) correlations. After generation of precipitation occurrences, Gaussianized precipitation is calculated through a simple linear regression of occurrences making use of the observed frequency distribution with the addition of a random white noise. Finally, precipitation depth is obtained through an inverse Gaussianization with the use of the monthly non-parametric distribution from the observed samples.

The daily minimum and maximum temperature were generated for the reference period 1949-2019 through a random generation with an auto-regression based on generalized linear models (Chambers and Hastie, 1992) implemented in the RMAWGEN package following Wilks approach for spatial (inter-station) correlations. The VAR precipitation model was used as an exogenous parameter to aid in the prediction of stochastically estimated temperature data. After generation of precipitation occurrences, Gaussianized temperature is calculated through a simple linear regression of occurrences making use of the observed frequency distribution with the addition of a random white noise (Cordano and Eccel, 2016; Cordano and Eccel, 2017). For additional details on the calibration process readers are referred to these citations: Cordano and Eccel, 2016; Cordano and Eccel, 2017; Cordano 2017; Ailliot et al., 2015.

The final calibrated VAR precipitation (Ppt) and VAR maximum temperature (Tmax), and VAR minimum temperature (Tmin) models were compared to the observed and simulated input data.

The observed versus simulated data were compared at quantiles from 0.1 to 0.99 at a 0.01 incremental level, the average correlation among the observed and simulated values were excellent with all stations correlation being greater than 0.98. As an example, the ten subbasins Q-Q plot are used to illustrate the basin precipitation correlations (Figure 3).

Stochastic Weather Generation Report Page 12 of 24 11/3/2021 Applied Weather Associates Figure 3: Example of the ten subbasins precipitation Q-Q plot 3.3 1000-year Simulation The calibrated VAR models based on 1949-2019 data for Ppt, Tmax, and Tmin were used to simulate daily precipitation, maximum temperature, and minimum temperature for twelve 1000-year periods. The 1000-year period was randomly selected to start on January 1, 2000 and simulate daily data through December 31, 3000. The simulated daily time series included leap years. To save space, the 1000-year daily generated time series at the 49 stations were packaged together (i.e., the Ppt data for all 49 stations were in one file, the Tmax data for all 49 stations were in one file, and the Tmin data for all 49 stations were in one file) for each of the twelve simulations.

3.4 Reformat Script An R-script was created and provided as a deliverable to reformat the twelve 1000-year stochastically generated deliverables (Ppt, Tmax, Tmin). The R-script reads in the grouped Ppt, Tmax, Tmin files and generates individual station input files that contain each stations daily Ppt, Tmax, and Tmin along with the needed formatting and headers for the hydrological model. The reformatted files contain the station identification number and the simulation number in the generated file name (i.e., Anoka_Sim_data_no1.txt will be the Anoka basin for simulation number 1). The reformatted induvial input data files can be imported and plotted using various software.

Figure 4 provides a graphical example of 100-years of simulated data for the McGregor basin.

Stochastic Weather Generation Report Page 13 of 24 11/3/2021 Applied Weather Associates Figure 4: Example of McGregor basin 100-year simulated data from stochastic simulation number 6.

4.0 LONGTERM CALIBRATION AWA utilized the Storm Precipitation Analysis System (SPAS) to analyze rainfall over the Quad Cities basin (Hultstrand and Kappel, 2017). Three water years were selected for calibration: a wet, a dry, and a normal precipitation year. Based on streamflow records, the top three wet years were 1986, 1993, and 2001 and the top three dry years were 1987, 1988, and 2009. AWA used the daily PRISM gridded dataset (Daly et al., 1997) to assess the precipitation spatial component and basin magnitude component for each water year (Figure 5 and Table 2). Based on sub-basin frequency counts, the 1993 water year was selected to represent the wet calibration period, and the 1988 water year was selected to represent the dry calibration period. The 1991 period was selected to represent the normal precipitation period because of its streamflow record and because it bracketed the dry year (1988) and wet year (1993) calibration periods (Figure 6). Daily values of precipitation, temperature, and SWE were generated for each of the Quad Cities sub-basins utilizing the 49 meteorological stations.

Stochastic Weather Generation Report Page 14 of 24 11/3/2021 Applied Weather Associates Figure 5: Water years investigated for model calibration, i) top row represents the three wet years investigated 1986, 1993, and 2001; ii) bottom row represents the three dry years investigated 1987, 1988, and 2009. Image shows precipitation in inches.

Table 2. Water year average sub-basin total precipitation for each calibration year investigated Basin Name Area (mi2)

Wet Years Dry Years 1986 1993 2001 1991 1987 1988 2009 Quad Cities 329.8 46.5 52.8 33.7 29.7 32.5 20.4 44.2 Minnesota 17064.9 34.6 37.9 30.6 34.0 19.0 18.7 25.3 Chippewa 9078.9 39.1 34.0 35.4 36.1 24.4 24.5 23.3 St. Croix 7726.7 39.3 30.3 35.6 37.3 19.9 24.1 24.3 Anoka 19904.6 34.8 30.8 30.9 31.4 19.9 21.0 26.2 Wapsipinicon 2333.4 41.9 53.1 36.7 34.8 31.2 21.9 38.9 McGregor 7881.6 38.1 43.7 34.6 33.5 29.4 24.5 28.6 Winona 5969.1 41.7 42.6 35.1 34.1 27.4 23.1 26.1 Clinton 7891.5 41.4 50.9 36.5 34.5 31.1 22.4 36.3 Wisconsin 10419.4 37.4 39.3 33.1 33.0 25.1 26.2 27.2 Wet/Dry Count 3

7 0

3 6

1

Stochastic Weather Generation Report Page 15 of 24 11/3/2021 Applied Weather Associates Figure 6: Normal water year (1991) investigated for model calibration, precipitation in inches.

5.0 SNOW WATER EQUIVALENT MODELLING AWA utilized NASAs Oak Ridge National Laboratory Daymet gridded dataset. Daymet is a collection of gridded estimates of daily weather parameters generated by interpolation and extrapolation from daily meteorological observations (Thornton et al., 2016). The model has high spatial (1-km2) and temporal (daily) resolutions and is run for all of North America. Daymet products include snow water equivalent (SWE) and are available starting in 1980 and though present, providing 40 years of data to help create snowpack climatologies.

In addition to Daymet, the National Operational Hydrologic Remote Sensing Center (NOHRSC)

SNOw Data Assimilation System (SNODAS) gridded dataset was investigated (Carroll et al.,

2001). SNODAS is a physically-based, near real-time energy and mass balance, spatially-uncoupled, vertically-distributed, multi-layer snow model (Carroll et al., 2001; NOHRSC, 2004).

The SNODAS data are available starting in 2003 though present, providing 17 years of data to create snowpack Climatologies, since the three calibration events were 1988, 1991, and 1993 the Daymet data were used for SWE calibration. Gridded daily Daymet SWE were extracted for the three calibration water years (1988, 1991, and 1993), a timeseries containing daily basin average SWE was extracted along with the basin average maximum temperature, minimum temperature, and precipitation.

For each water year, the daily time series data were used as input into the R SnowMelt library (Fuka et al., 2018) based on the SNOW-17 snow accumulation and ablation model described by Anderson (1973) and Anderson (1976) as a component of the National Weather Service River

Stochastic Weather Generation Report Page 16 of 24 11/3/2021 Applied Weather Associates Forecast System (NWSRFS). SNOW-17 is a conceptual model. SNOW-17 is an index model using air temperature as an index to determine the energy exchange across the snow-air interface (Anderson, 2006). In addition to temperature, the only other required input variables needed to run the model are precipitation and latitude. Air temperature is used as an indicator to estimate snowmelt and snowpack water equivalent. Snowmelt calculations are based on snow energy balance that account for snowpack accumulation, heat exchange between the snow and ambient air temperature, snow cover extent, heat storage and exchange, and water retention or transmission (Walter et al., 2005).

The daily time series data were used input into the R SnowMelt model to simulate daily SWE, to calibrate the amount of forest cover (f) and snow covered area (SCA). Goodness-of-fit measures used to assess the calibration were the Mean Error (ME), Mean Absolute Error (MAE), Root-mean Squared Error (RMSE), Nash-Sutcliffe Efficiency (NSE), Pearson correlation (r), and the coefficient of determination (r2). Each years basin calibration had great goodness-of-fit measures, ME was typically less than 1 mm, the MAE was typically within 2 mm, the NSE, r, and r2 values were all greater than 0.90. An example of the McGregor subbasin SWE calibration results are shown in Figure 7. The average f and SCA values, by basin, were used to simulated SWE for the stochastically generated temperature and precipitation time series discussed in section 3.0.

Stochastic Weather Generation Report Page 17 of 24 11/3/2021 Applied Weather Associates Figure 7: Example of simulated SWE calibration for the McGregor subbasin for a) water year 1988, b) water year 1991, and c) water year 1993.

Stochastic Weather Generation Report Page 18 of 24 11/3/2021 Applied Weather Associates 6.0 CLIMATE CHANGE In addition to the observational data, two climate change projection scenarios representing daily weather sequences of precipitation, minimum and maximum temperatures, and SWE were evaluated. These climate change projections utilized regional downscaled model output for the region driven by representative concentration pathway (RCP) 4.5 and 8.5 from CMIP5 global climate model output. These model projections were used to generate twelve (12) 1,000-year iterations representing possible future climate scenarios. The Intergovernmental Panel on Climate Change (IPCC) fifth assessment report (AR5) (IPCC, 2017) contains representative concentration pathways (RCPs), a greenhouse gas concentration trajectory; often referred to as emission scenarios. The pathways describe different climate futures, all of which are considered possible depending on the volume of greenhouse gases (GHG) emitted in the years to come (Figure 8). The RCPs investigated; RCP45 and RCP85; represent a mid-level mitigation and no mitigation to limit radiative forcing (IPCC, 2017).

Figure 8: Representative Concentration Pathway (RCP) Trajectories from IPCC AR5 AWA utilized the ClimateNA software interface to provide locally downscaled spatial climate data for historical and future climate scenarios (Wang et al, 2016). ClimateNA utilizes historical weather station data and global circulation model regional predictions to project future seasonal and annual climate variables in North America. ClimateNA extracts and downscales Parameter Regression of Independent Slopes Model (PRISM) (Daly et al. 2002) and ANUSPLIN (Hutchinson 1991) generated monthly data for numerous reference periods to calculate seasonal and annual climate variables for specific locations based on latitude, longitude, and elevation.

ClimateNA also downscales and integrates historical (1901-2020) (Mitchell and Jones 2005) and

Stochastic Weather Generation Report Page 19 of 24 11/3/2021 Applied Weather Associates future climate projections (2020s, 2050s and 2080s) generated by fifteen global circulation models from IPCC AR5 and two RCP scenarios (RCP45 and RCP85).

For the Quad Cities study, the historic reference period and climate projection periods were estimated for each of the ten Quad Cities subbasins. The reference period is based on the 1980-2010 PRISM monthly climate grids. To quantify change in temperature (Tmax and Tmin) and precipitation by month, a delta methodology was used to compare the historical period to the 2040-2070 future projections (Wang et al., 2016). The delta/change from the historical period to the future period 2040-2070, were used to update the calibrated RMAWGEN, VAR precipitation, VAR maximum temperature, and VAR minimum temperature models future climatology (Cordano and Eccel, 2016; Cordano and Eccel, 2017). The results of the Quad Cities basin monthly climate normal data compared to future projection climate normal are shown in Figure 9. The monthly change in temperature (C) and precipitation (%) from historical period (observation period) to climate projections based on RCP45 and RCP85 scenarios are shown in Figure 10.

Figure 9: Quad Cities basin monthly climate normal data compared to future projection climate normal data based on RCP45 and RCP85 climate projections.

Stochastic Weather Generation Report Page 20 of 24 11/3/2021 Applied Weather Associates Figure 10: The monthly change in temperature and precipitation from historical period (observation period) to climate projections based on RCP45 and RCP85 scenarios.

6.1 1000-year Simulation The calibrated VAR models based on 1949-2019 data for Ppt, Tmax, and Tmin were used along with the updated climate projection climatologies to simulate daily precipitation, maximum temperature, and minimum temperature for twelve 1000-year future scenarios. The 1000-year period was randomly selected to start on January 1, 2000 and simulate daily data through December 31, 3000. The simulated daily time series included leap years. To save space, the 1000-year daily generated time series at the 49 stations were packaged together (i.e., the Ppt data for all 49 stations were in one file, the Tmax data for all 49 stations were in one file, and the Tmin data for all 49 stations were in one file) for each of the twelve simulations. The average f and SCA values, by basin, were used to simulated SWE for the stochastically generated future temperature and precipitation time series.

7.0 RESULTS AND CONCLUSIONS The calibrated precipitation and temperature VAR models based on daily data from 1949-2019 were used to simulated daily data for twelve 1000-year periods at 49 stations within the Quad Cities. The RMAWGEN model was selected because it is able to maintain the temporal and spatial correlations among the several sites, account for seasonally changing weather variables, and account for future climate change scenarios (Cordano and Eccel, 2016; Cordano and Eccel, 2017).

The significant amount of meteorological data available both spatially and for a long period of

Stochastic Weather Generation Report Page 21 of 24 11/3/2021 Applied Weather Associates record resulted in high confidence in the results and simulations. The results of this study provide stochastically simulated daily sequences that provide a set of alternate realizations, based on both observational data and climate change projections, that can be used for hydrologic assessments throughout both basins, as well as risk and reliability assessment in the design and operation of various Quad Cities systems.

Stochastic Weather Generation Report Page 22 of 24 11/3/2021 Applied Weather Associates

8.0 REFERENCES

Ailliot P, D. Allard, V. Monbet, P. Naveau, 2015. Stochastic weather generators: an overview of weather type models. Le Journal de la Société Française de Statistique 156(1):101-113 English version at: http://perso.univ-rennes1.fr/valerie.monbet/doc/papiers_pdf/SWGEN_review.pdf. Accessed 15 June 2020.

Anderson, E.A., 1973. National Weather Service River Forecast System-Snow Accumulation and Ablation Model, NOAA Technical Memorandum: NWS Hydro-17, US National Weather Service Anderson, E., 1976. A point energy and mass balance model of a snow cover, NOAA Tech.

Rep. NWS 19, Silver Spring, MD.

Anderson, E. 2006. Snow Accumulation and Ablation Model -- SNOW-17 Carroll, T., D. Cline, G. Fall, A. Nilsson, L. Li, and A. Rost, 2001. NOHRSC Operations and the Simulation of Snow Cover Properties for the Conterminous U.S. Proceedings of the 69th Annual Meeting of the Western Snow Conference, pp. 1-14.

Chambers J. M., and T.J. Hastie, 1992. Statistical models in S. Wadsworth & Brooks/Cole, Pacific Grove, California.

Cordano E. and E. Eccel E., 2016. Tools for stochastic weather series generation in R environment, Italian Journal of Agrometeorology, doi:10.19199/2016.3.2038-5625.031 Cordano E. and E. Eccel, 2017. RMAWGEN: Multi-Site Auto-Regressive Weather GENerator.

R Package, https://CRAN.R-project.org/package=RMAWGEN Cordano, E. 2017. RGENERATEPREC: Tools to Generate Daily-Precipitation Time Series. R package, https://CRAN.R-project.org/package=RGENERATEPREC Cordano, E., 2017. RGENERATE: Tools To Generate Vector Time Series. R package, https://CRAN.R-project.org/package=RGENERATE Daly, C., Taylor, G., and W. Gibson, 1997: The PRISM Approach to Mapping Precipitation and Temperature, 10th Conf. on Applied Climatology, Reno, NV, Amer. Meteor. Soc., 10-12.

Daly C., W.P. Gibson, G.H. Taylor, G.L. Johnson, and P. Pasteris, 2002. A knowledge-based approach to the statistical mapping of climate. Climate Research, 22:99-113.

Fuka D.R., M.T. Walter, J.A. Archibald, T.S. Steenhuis, and Z.M. Easton, 2018. A Community Modeling Foundation for Eco-Hydrology. R package, https://cran.r-project.org/web/packages/EcoHydRology/

Stochastic Weather Generation Report Page 23 of 24 11/3/2021 Applied Weather Associates Hutchinson M.F. and J.D. Jasper, 1991. The application of thin plate smoothing splines to continent-wide data assimilation, BMRC Research Report No. 27, pp. 104-113, 1991.

Hultstrand, D.M., and Kappel, W.D., 2017. The Storm Precipitation Analysis System (SPAS)

Report. Nuclear Regulatory Commission (NRC) Inspection Report No 99901474/2016-201, Enercon Services, Inc., 95pp.

IPCC, 2017. Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5)

Observed Climate Change Impacts Database, Version 2.01. Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC).

https://doi.org/10.7927/H4FT8J0X.

Kappel, W.D., Hultstrand, D.M., Tomlinson, E.M., Muhlestein, G.A., and T.W. Parzybok, September 2012: Site-Specific Probable Maximum Precipitation (PMP) Study for the Quad Cities Nuclear Generating Station, Quad Cities, IA.

Kappel, W.D., Hultstrand, D.M., Muhlestein, G.A., Steinhilber, K. M., and D. McGlone, September 2014: Site-Specific Local Intense Precipitation (LIP) Study for the Quad Cities Nuclear Station Lawrence, D., E. Paquet, J. Gailhard, and A.K. Fleig., 2014. Stochastic semi-continuous simulation for extreme flood estimation in catchments with combined rainfall-snowmelt flood regimes, Nat. Hazards Earth Syst. Sci 14, 1283-1298. https://doi.org/10.5194/nhess-14-1283-2014 Mehrotra R, R. Srikanthan, A. Sharma, 2006. A comparison of three stochastic multi-site precipitation occurrence generators, Journal of Hydrology 331:280-292.

https://doi.org/10.1016/j.jhydrol.2006.05.016 Mitchell, T.D. and P.D. Jones, 2005. An improved method of constructing a database of monthly climate observations and associated high-resolution grids. International Journal of Climatology, 25, 693-712.

National Operational Hydrologic Remote Sensing Center (NOHRSC), 2004. Snow Data Assimilation System (SNODAS) Data Products at NSIDC, Version 1. Boulder, Colorado USA. NSIDC: National Snow and Ice Data Center. doi:

https://doi.org/10.7265/N5TB14TC.

Paquet, E., F. Garavaglia, R. Garçon, and J Gailhard, 2013. The SCHADEX method: A semi-continuous rainfall-runoff simulation for extreme flood estimation, Journal of Hydrology, 495:23-37. https://doi.org/10.1016/j.jhydrol.2013.04.045.

Thornton, P.E., M.M. Thornton, B.W. Mayer, Y. Wei, R. Devarakonda, R.S. Vose, and R.B.

Cook. 2016. Daymet: Daily Surface Weather Data on a 1-km Grid for North America, Version 3. ORNL DAAC, Oak Ridge, Tennessee, USA.

https://doi.org/10.3334/ORNLDAAC/1328

Stochastic Weather Generation Report Page 24 of 24 11/3/2021 Applied Weather Associates Walter, M.T., E.S. Brooks, D.K. McCool, L.G. King, M. Molnau, J. Boll, 2005. Process-based snowmelt modeling: Does it require more input data than temperature-index modeling?

Journal of Hydrology 300(1-4): 65-75 Wang, T, A. Hamann, D. Spittlehouse, and C. Carroll, 2016. Locally Downscaled and Spatially Customizable Climate Data for Historical and Future Periods for North America. PLoS ONE 11(6): e0156720. doi:10.1371/journal.pone.0156720 Wilks D and R. Wilby, 1999. The weather generation game: a review of stochastic weather models. Progress in Physical Geography 23(3):329-357.

https://doi.org/10.1177/030913339902300302

v t e Letter Sequence Response to RAI
EPID:L-2023-LLA-0084, TSTF-505 (Approved, Closed)
Initiation Request Acceptance Administration Questions Answers 19 March 2024 5 April 2024 10 May 2024 6 June 2024 Results Approval Other: ML23319A334, ML24110A049
MONTHYEAR2024-04-05 [Table View]

RS-24-042, Response to Request for Additional Information Related to License Amendment Request to Revise Technical Specifications to Adopt Risk Informed Completion Times TSTF-505, Revision 2, Provide Risk-Informed Extended Completion

Text

Subject:

References:

8.0 CONCLUSION

10.0 REFERENCES

8.0 CONCLUSION

10.0 REFERENCES

1.0 INTRODUCTION

8.0 REFERENCES

1.0 INTRODUCTION

8.0 REFERENCES

Navigation menu

RS-24-042, Response to Request for Additional Information Related to License Amendment Request to Revise Technical Specifications to Adopt Risk Informed Completion Times TSTF-505, Revision 2, Provide Risk-Informed Extended Completion

Text

Subject:

References:

8.0 CONCLUSION

10.0 REFERENCES

8.0 CONCLUSION

10.0 REFERENCES

1.0 INTRODUCTION

8.0 REFERENCES

1.0 INTRODUCTION

8.0 REFERENCES

Navigation menu

Search