Water Research Foundation Scope of Work

Project Title:

Joint Front Range Climate Change Vulnerability Study

Project Number:

4205

Contractor:

Riverside Technology, inc.

Project Manager: Susan Turnquist Year Funded:

2008

This Scope of Work is made available to Foundation subscribers for informational purposes only. The scope of work was developed by the principal investigator to describe the approach to be taken in the research project. Scopes of work frequently are modified throughout the course of the project as new information is uncovered and preliminary hypotheses are investigated. For updates on the status of the project, review the project update or contact the Foundation project manager. The Foundation assumes no responsibility for the content of this scope of work or for the opinions or statements expressed. The mention of trade names or commercial products does not represent or imply the approval or endorsement of the Foundation. Do not distribute outside your organization.

©2009 Water Research Foundation. ALL RIGHTS RESERVED

Joint Front Range Climate Change Vulnerability Study Project Funding Agreement 04205 Scope of Work January 9, 2009 Background The following scope of work summarizes the work to be performed under a Joint Project Funding Agreement (PFA) between the Water Research Foundation, the project Cofunders, and Riverside Technology, inc. (RTi), under the direction of the Principal Investigator (PI), Mark Woodbury. RTi will subcontract with the National Center for Atmospheric Research (NCAR) for key components of the project, which will be performed under the direction of the Co-PI, Dr. David Yates. The scope of work incorporates changes described in a response to technical review comments on the proposal dated May 13, 2008. Project/Research Objectives Metropolitan water providers along Colorado’s Front Range are concerned about the impact that climate change may have on their future water supply availability. To better understand the possible impacts of climate change, several Front Range providers are working together to provide the education, tools, and methodology needed to study these effects. This project is designed to enable entities that obtain their water supplies from the upper Colorado, South Platte, Arkansas, Cache la Poudre, St. Vrain, Boulder

Figure 1. General of study area JFRCCVS Scope ofmap Work

Page 1

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

Creek, and Big Thompson River Basins to examine the potential effects that climate change may have on these supplies. Figure 1 shows a general map of the study area. The participating group, called the Project Steering Committee (PSC), consists of representatives from Aurora Water (Aurora), City of Boulder (Boulder), City of Fort Collins (Fort Collins), Colorado Spring Utilities (Colorado Springs), Denver Water (Denver), and Northern Colorado Water Conservancy District (Northern). Additional expert resources are assisting with the project through the help of the Western Water Assessment (WWA) and the Colorado Water Conservation Board (CWCB). The WWA aids the PSC in interpreting and selecting downscaled General Circulation Model (GCM) temperature and precipitation output data. The selected datasets will be used to drive the hydrologic models that will be applied to assess the sensitivity of streamflow to changes in climate. The CWCB will work with the PSC to establish historic naturalized streamflow datasets for use in calibrating the hydrologic models and as the baseline dataset of historic water availability. This unified approach is intended to help Front Range water providers communicate with their customers and the media cohesively, using consistent hydrometeorological data, methods, and climate change scenarios. Lessons learned from this unified approach could also be used to encourage and establish further regional efforts in Colorado and throughout the country. The goal of this study is a streamflow sensitivity analysis of the seven watersheds noted above with a focus on the development of baseline naturalized streamflow datasets representing potential future conditions based on selected climate change scenarios. Each provider can use these future streamflow scenarios in conjunction with its water rights allocation assumptions to estimate the impacts of various climate change scenarios on its water supply. Work Plan Task 1 – Historic Naturalized Streamflow Data Development

Many water providers evaluate vulnerability using water allocation models that simulate system operations based on historic sequences of natural streamflow. An important goal of the study is to be able to compare computed historical natural flows at specific nodes with historic natural flow sequences that have been adjusted to reflect the impact of possible future climate scenarios. The majority of the gauging stations in the study area are influenced by streamflow regulation at upstream locations, including reservoir storage and release, trans-basin imports and exports, and diversions for municipal, industrial, and agricultural uses. The degree of regulation varies across the study area, but generally, there is some level of streamflow regulation in almost every basin of interest. This human influence on streamflow clearly affects the amount of flow reaching downstream points. The study needs to begin with an assessment of natural, unregulated streamflow at different points in order to derive the impact of changes in precipitation, temperature, or other factors on the natural flow available in a basin. The development of historic natural streamflow for this study has three important purposes. First is to have a common set of natural flows to use as a basis for developing climate adjusted flows or re-sequenced flows. Second is to have appropriate streamflow sequences to use in calibrating hydrologic models that will be used to simulate the change in hydrologic response due to potential changes in climate. Finally, the naturalized flow sequences form the basis against which climate adjusted

JFRCCVS Scope of Work

Page 2

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

naturalized flow sequences can be compared. For this study, historical streamflow for the period from 1950 – 2005 will be evaluated. The first proposed task of the study includes the collection or development of natural streamflow time series that have the impact of the upstream regulation removed from the observed flow time series. For instance, upstream diversions or the change in storage of an upstream reservoir will be added to the downstream flow to derive a time series of the unregulated, natural streamflow that would have occurred at a given point in the absence of the upstream diversions, reservoirs, or other streamflow regulation. In most locations, the state of Colorado or the different municipalities have already developed natural streamflow at basin outlets during previous work. In locations where a consensus exists concerning the legitimacy and quality of previously modeled natural flow time series, these time series will be utilized. A list of the gauges in the study area of primary interest to members of the PSC is presented below in Table 1. Historic naturalized streamflow sequences have been developed for many of the gauge locations listed for significant portions of the study period. Developing natural streamflow datasets for this study will involve the following principal activities: 1. Compile and review existing historical natural flow datasets from project participants. Identify any gaps that may exist in the datasets in relation to the study period. 2. Develop natural flow datasets as necessary to fill gaps and complement existing datasets for the complete study period. 3. Evaluate the quality of natural flow records and coordinate with project participants to select and agree on the final natural flow datasets. 4. Document the natural streamflow dataset development process, including the general procedures used in developing the individual natural flow datasets from each source, as well as criteria used in selecting the data sources. Table 1. Streamflow locations and available data between 1950-2005 Basin Upper Colorado

Station Fraser River at Granby (09034000) Williams Fork near Leal (09035700) Blue River below Green Mountain Reservoir (09057500) Blue River below Dillon, CO (09050700) Colorado River near Granby, CO (09019500) Colorado River near Dotsero (09070500) Colorado River near Cameo (09095500) Homestake Creek at Gold Park (09064000) Roaring Fork River near Aspen (09073400)

JFRCCVS Scope of Work

Period of Record Daily natural flow 1947-1991 (DW); Monthly natural flow 1991-present (NCWCD) Daily natural flow 1947-1991 (DW); Monthly natural flow 1991-pres (DW); Yates has data. Daily natural flow 1947-1991 (DW) and 1975-2005 (CDSS); Monthly natural flow 1909-2005 (CDSS) Daily natural flow 1947-1991 (DW) and 1975-2005 (CDSS); Monthly natural flow 1915-present (DW) and 1909-2005 (CDSS) Daily natural flow 1947-1991 (DW) and 1975-2005 (CDSS); Monthly natural flow 1909-2005 (CDSS) Daily natural flow 1947-1991 (DW) and 1975-2005 (CDSS); Monthly natural flow 1909-2005 (CDSS) Daily natural flow 1947-1991 (DW) and 1975-2005 (CDSS); Monthly natural flow 1909-2005 (CDSS) Daily natural flow 1947-1991 (DW); Measured daily flow from 1947 to present (USGS); Yates has data. Daily measured flow 1964-present. Daily natural flow 1975-2005 (CDSS). Monthly natural flow 1909-2005 (CDSS)

Page 3

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

Upper Arkansas Upper South Platte

Cache la Poudre St. Vrain Big Thompson Boulder Creek

Arkansas River at Salida (07091500)

Measured daily flow from 1910 to present.

S.Platte River above Spinney Mountain Reservoir (06694920) South Platte River below Cheesman Reservoir South Platte River at South Platte

Daily natural flows 1947-1991 (DW); Measured daily 1991-present. RTi has data. Daily natural flows 1947-1991 and monthly thereafter (DW). RTi has data. Daily natural flows 1947-1991 and monthly thereafter (DW). RTi has data. DW has daily natural flows from 1947-1991; Measured daily 1991-2005 (CDSS). RTi has data. Monthly natural flow available from 1950 - 2006 (NCWCD). Measured daily flow data goes back to 1881. Monthly natural flow available from 1965 - 2005 (NCWCD). Monthly natural flow available from 1947 - 2006 (NCWCD). Monthly natural flow available from 1963 - 2005 (Boulder).

South Platte River at Henderson (06720500) Cache la Poudre River at Mouth of Canyon (06752000) St. Vrain Creek at Canyon Mouth near Lyons Big Thompson River at Mouth of Canyon near Drake (06738000) Boulder Creek at Orodell

Both the RTi and NCAR research teams will participate in the development of natural flows based on researcher experience with data in the study basins. At locations where there is overlap between data from multiple sources, the data will be evaluated in an attempt to determine the most appropriate data sources. As a part of the Colorado Decision Support System (CDSS), the State of Colorado has documented information concerning surface water projects including active diversions, water rights, and transfers of water rights among entities in a series of reports and memorandums for the Colorado and South Platte River Basins. Straight-line diagrams have also been prepared for many Water Districts. These diagrams detail the location of diversions and other important information affecting surface water. The State’s Hydrobase database contains time series information for diversions throughout the state. Limited information also is available through the CDSS for the Arkansas River Basin, although some time series data are available in Hydrobase. Researchers will use information from the CDSS and Hydrobase to determine where significant diversions and surface water operations occur that affect the natural flow in reaches where naturalized flows must be developed. Major diversions, trans-basin imports, and changes in upstream reservoir storage will be accounted for in the flow naturalization. The PIs will consult with PSC members familiar with the basins of interest and the CWCB to assure that information used for the natural flow calculations is appropriate and inclusive of the important streamflow regulation. The Hydrobase data records will provide the basis for time series data, unless data of higher quality are provided by members of the PSC for specific areas. For storage projects, the Hydrobase data are typically of a lower quality, and therefore data may be obtained from the PSC members or alternate sources. Task 2 - Model Development and Calibration

Two hydrologic models will be applied independently to assess the change in hydrologic response resulting from changes to the climate inputs. Model parameters first will be calibrated to assure that the predicted change in streamflow due to climate variation is consistent with the response of the study area under the current climatic conditions. Because the hydrologic models will be used to simulate flows that will be compared with

JFRCCVS Scope of Work

Page 4

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

historic natural flow sequences, researchers will refine the calibrations for the basins of interest to be consistent with the natural flows developed in Task 1. The models that have been identified for use in the study are the National Weather Service (NWS) models that are currently used by NWS River Forecast Centers (RFCs), and the WEAP (Water Evaluation and Planning) model (developed by the Stockholm Environment Institute). The NWS models include the Sacramento Soil Moisture Accounting model coupled with the Snow-17 snow accumulation and ablation model (also known as the Anderson snow model). These models, hereafter referred to as the Sacramento Model, have been calibrated previously, and are actively used by the NWS in developing river forecasts for the Colorado, Missouri, and Arkansas River Basins, which collectively cover the study area. RTi will lead the development of updated model calibrations of the Sacramento model and the simulation of climate change impacts using this model. Dr. David Yates of NCAR is one of the developers of the WEAP model, and he, along with Colorado Springs Utilities and the Foundation, is in the process of calibrating the model for the Upper Colorado River Basin. NCAR will calibrate WEAP for the other River Basins of interest including the Upper Arkansas, Upper South Platte, Big Thompson, Cache la Poudre, St. Vrain, and Boulder Creek, and will apply it in simulating climate change impacts. WEAP Model Development Background

Version 21 of the WEAP model attempts to address the gap between water management and watershed hydrology by integrating physical hydrologic processes (surface water, groundwater, evapotranspiration, snow accumulation and melt) with the management of demands and installed infrastructure in a seamless and coherent manner (Yates et al. 2005a,b). Embedded within the water resource systems logic is a watershed hydrology module that allows for the direct assessment of hydrologic changes on managed water systems. Dr. Yates and his research team will calibrate and validate simulated streamflow against historic observations using a classical split-sample procedure and the independent, non-linear parameter estimation model software PEST (Doherty 2002). PEST will be used to automatically adjust a set of scaling parameters such as soil water capacity, hydraulic conductivity, and snowmelt thresholds until the simulated and observed inflows more closely match, in a weighted, least squares analysis. The calibration and validation of natural streamflow simulations with WEAP compared against computed natural streamflow will be for the years between 1980 and 2003, which correspond to the period of record for the daymet.org dataset. A full simulation of natural flows will be conducted for all seven Basins for the period 1950 through 2003 using a hybrid climate dataset described later. For each of the basins, the research team will first build models of the largely unregulated catchments where there are good estimates of naturalized flows. The research team will calibrate these tributary streams and move incrementally down the watershed to simulate more of each basin until the complete contributing area is considered. For the Colorado River, this would be all areas upstream of the gauge at Cameo (09095500).

JFRCCVS Scope of Work

Page 5

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

Climate Series Length

The research team recognizes the need to study a long climate record, 1950-2005. Of course, if one wants an observed streamflow record that extends back that far, and a simulated estimate of streamflow for the same period that is climate driven, one needs a corresponding climate record for the same period. For the Upper Colorado Basin, the NCAR team is making use of a high quality, high-resolution climate dataset that has been converted to 1-km resolution grids across the continental US. Problematically, this dataset only covers the period 1980 through 2003. Fortunately, a companion dataset (Mauer et al. 2002) contains a nearly complete 50-year climate record for the continental US, but on a coarser, 1/8th degree grid. The WEAP team will use the Mauer et al. dataset and the daymet.org dataset together to create a continuous time series from 1950 through 1980. After 1980, the daymet data set can be used exclusively. WEAP is not a “gridded” hydrology model, so the team is not making use of every daymet grid point, but is extracting climate data for a unique latitude/longitude location. The team will use the 1980 through 2003 daymet record for calibration and validation of each basin. The longer climate record will only be used for simulation and comparison against the historic record. Data Acquisition, Processing, Calibration and Streamflow Validation

The WEAP Modeling Team will acquire and process a Digital Elevation Model (DEM), Land Use/Land Cover, and Soils Datasets for the Arkansas, South Platte, Cache la Poudre, St. Vrain, Boulder Creek, and Big Thompson river basins. This work has already been done for the Upper Colorado Basin. Based on the analysis above, the catchment boundaries will be defined for each of the six river systems. After this step, data developed in the previous steps will be entered into the WEAP modeling framework along with the geophysical data for each stream and catchment object. The elevation band of each catchment will be used to determine its centroid (latitude and longitude) and this information will be used to acquire the 23year, daily climate record for the six basins. To calibrate and validate the hydrologic simulations of naturalized streamflow the data will be compared against computed natural streamflow for select points. The data will also be compared against SNOTEL data to evaluate how well the model calculated snowpack accumulation and melt-off. The team will make use of PEST to help in the calibration procedure for tuning model parameters. Each of the seven basins will be individually calibrated and validated. This procedure is nearly complete for the Upper Colorado. Sacramento Model Development

RTi will employ the Sacramento model to simulate the hydrologic response of the basins. Together, RTi and the NWS RFCs have calibrated the Sacramento Model for the South Platte, Colorado, and Arkansas River Basins, based on naturalized streamflow time series that may be different from those developed in Task 1. The hydrologic models and historic climate data associated with each of the three river basins involved in this study are managed by three different NWS RFCs. RTi will obtain current model parameters from each RFC and will review and refine the calibrated parameters. The hydrologic model configurations used by the RFCs sub-divide the River Basins into multiple sub-basins and elevation zones to account for spatial variability in precipitation, JFRCCVS Scope of Work

Page 6

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

temperature, and basin snowmelt and runoff characteristics. The historic temperature and precipitation time series that have been developed for use in calibrating the models are defined for each sub-basin to best represent the mean areal characteristics of that sub-basin based on a combination of gauges in the climate network. The resulting time series are defined at 6-hour time steps. This effort will focus on identifying differences between the natural flow time series used during previous calibration efforts and those natural flow time series produced in Task 1, and attempting to adjust model parameters to better simulate the updated natural streamflow. The calibration effort is not intended to be a re-calibration of existing models, but rather a refinement of model parameters based on information from the new natural flow time series. The calibration effort will attempt to balance capturing the monthly water balance with capturing daily hydrograph shapes and magnitudes of flows. Prior to calibrating downstream locations in a given basin, the hydrologic models will be calibrated to upstream locations in that same basin. Certain downstream locations, such as the Colorado River at Dotsero and the Colorado River at Cameo, will require modeling of multiple sub-basins between upstream stream gauge locations and the location of interest. The hydrologic model calibration adjustments will be made based on natural flow at locations identified as part of this study, and will not focus on tributary streams or intermediate stream gauge locations except as necessary to improve the simulation at the points of interest. The Arkansas Basin has historic climate data available from the Arkansas Basin RFC for the period from 1950 to 2007. Data for the South Platte Basin are available from the Missouri Basin RFC for the period from 1948 to 2004. Data for the Colorado River Basin are available from the Colorado Basin RFC for the period from 1950 to 2005, although the RFC considers the data prior to 1975 to be of inferior quality because of the lack of SNOTEL observations prior to that time. RTi will compare the temperature and precipitation datasets, including comparisons of pre and post 1975 periods, and make adjustments if necessary to assure consistency through the historic period. ET Demand Adjustments

An important input into the Sacramento Model is the evapotranspiration demand (ET Demand). The ET Demand is strongly dependent on prevailing temperatures and land cover in the area. As such, potential evapotranspiration (PET) is typically estimated using temperature driven ET estimation methods, such as Penman-Monteith. These PET estimates serve as initial input into the Sacramento Model and are adjusted during calibration to account for the influence of local vegetation cover. The adjusted PET estimates then represent ET Demand for the modeled area, when the ET is not limited by moisture availability. Because the procedure for estimating ET demand parameters incorporates prevailing temperature, a procedure for adjusting the ET demand parameters for each hypothetical climate scenario will be applied so that the adjusted temperature inputs are properly reflected. For this study, RTi proposes to estimate changes in ET Demand caused by changes in projected temperature as follows: 1. For each sub-basin, estimate average monthly PET using the Penman-Monteith method and available Mean Areal Temperature (MAT) datasets for the calibration period.

JFRCCVS Scope of Work

Page 7

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

2. Adjust these estimates during calibration to create average monthly ET Demand values. 3. After calibration, establish monthly relationships between the original PET estimated and the final ET Demand. 4. For each climate change scenario, re-estimate PET using the associated scenario MAT data. 5. Using the relationships established in step 3, convert the scenario PET estimates into scenario ET Demand. 6. Use these scenario ET Demand parameters as input to the Sacramento models. The ET Demand adjustment procedure will be automated to be applied individually to each sub-basin in the study area and will be repeated for each climate change scenario that is performed. Task 3 - Evaluation and Determination of Climate Variables

The overall approach or concept of the study is to use calibrated hydrologic models to simulate natural streamflow, and to compare model output based on historical climate inputs against model output based on historical climate inputs that have been adjusted to represent possible changes to climate in the future. A key aspect of the study, therefore, is the methodology for selecting future climate scenarios and applying them to the historical climate inputs to create the adjusted historical climate inputs. The procedure that will be used begins with the selection of a baseline climate. For this study, the ‘baseline’ climate is that which has been observed in the historical period from 1950 to 1999, and for which natural streamflow records have been or will be developed. Although the later part of this baseline includes initial warming, it is still a valid baseline for comparison with a presumed future condition because it generally corresponds to the period currently used by utilities for planning purposes. The changes or perturbations to be applied to the historical period are computed relative to it, based on GCM model results for that period compared with model results for the future periods to be evaluated. The future periods to be evaluated are 2025-2054 (representing potential conditions in 2040), and 2055-2084 (representing potential conditions in 2070). The methodology for developing the adjusted historical climate inputs includes the following important elements: 1. Selecting CO2 emissions scenarios, which characterize possible future economic and technological development and its resultant impact on atmospheric CO2; 2. Obtaining GCM output for the selected emissions scenarios; 3. Downscaling GCM output to a higher resolution grid for the area of interest in Colorado; and 4. Computing average monthly temperature shifts and precipitation adjustment factors (offsets) between the baseline climate period and each of the two future evaluation periods for each GCM projection. 5. Selecting individual scenarios of temperature and precipitation adjustment for each future period that represent a reasonable range of possibilities for which the study participants wish to evaluate impacts on streamflow. 6. Computing monthly gridded fields of temperature and precipitation adjustments (Delta TP) for each climate scenario.

JFRCCVS Scope of Work

Page 8

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

Emission Scenarios

The IPCC created a Special Report on Emissions Scenarios (SRES) to make a clear distinction between the uncertainty in climate model projections and emissions curves. The scenarios differ in demographic, socioeconomic, and technical development (IPCC SRES, 2000). The study will examine the three most intensively studied emissions scenarios; A2, A1b, and B1. The scenarios represent three possible changes to future greenhouse gas emissions including; no reduction of CO2 emissions (A2), a leveling-off of CO2 emissions by mid-21st Century followed by reductions (A1b), and immediate reductions of CO2 emissions (B1). The three selected scenarios have been used to drive 16 of the 22 developed GCMs. Variation exists in temperature and precipitation changes between GCM simulations, due to differences in model formulation and assumed emissions. Climate Models

Many research institutions worldwide have developed Global Climate Models (GCM). A model projection results from the simulation of a given GCM using a specific CO2 emissions scenario as input and using a specific set of initial conditions. While the models are in general agreement about trends in future temperature, there is much less agreement about future precipitation, and there are definite, significant variations in the projections, including between multiple projections using the same models with different initial conditions. Selecting a single GCM projection for evaluation, therefore, would not represent the variability nor the uncertainty in current understanding of future climate trends. A better approach for investigating climate change impacts and adaptation strategies for water utilities is to evaluate results from an ensemble of GCM simulations. A list of the GCMs selected for this study, including the associated emissions scenarios and the specific projections developed for that scenario, is shown below in Table 2. Columns labeled A1b, A2, and B1 refer to future greenhouse gas emissions scenarios where A1b has CO2 emissions leveling off by mid-21st Century and then decreasing thereafter; A2 has CO2 emissions increasing throughout the 21st Century; and B1 has CO2 emissions decreasing immediately. The numbers in the columns indicate the number of ensemble members for the corresponding GCM, meaning the number of projections by the corresponding GCM and greenhouse gas future emissions scenario. The climate models are from modeling centers in the United States (GFDL-CM2.0, GFDL-CM2.1, GISS-ER, CCSM3, and PCM), Canada (CGCM3.1), France (CNRMCM3, IPSL-CM4), Russia (INM-CM3.0), Australia (CSIRO-MK3.0), Japan (MIROC3.2, MRI-CGCM2.3.2), Germany (ECHO-G, ECHAM5/MPI-OM), Norway (BCCR-BCM2.0), and the United Kingdom (UKMO-HadCM3). There are 36 A1b, 39 A2, and 37 B1 climate scenario simulations. A total of 112 simulations were completed. Output from the GCMs provides simulated future climate information on a large scale. Since the purpose of this study is to estimate regional effects of climate change on streamflow yields, downscaled GCM outputs are needed.

JFRCCVS Scope of Work

Page 9

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

Table 2.Climate models used to generate the data set #

WCRP CMIP3 Model I.D.

# A1b

# A2

# B1

1

BCCR-BCM2.0

1

1

1

2

CGCM3.1 (T47)

1…5

1…5

1…5

3

CNRM-CM3

1

1

1

4

CSIRO-MK3.0

1

1

1

5

GFDL-CM2.0

1

1

1

6

GFDL-CM2.1

1

1

1

7

GISS-ER

1

2, 4

1

8

INM-CM3.0

1

1

1

9

IPSL-CM4

1

1

1

10

MIROC3.2(medres)

1…3

1…3

1…3

11

ECHO-G

1…3

1…3

1…3

12

ECHAM5/MPI-OM

1…3

1…3

1…3

13

MIR-CGCM2.3.2

1…5

1…5

1…5

14

CCSM3

1…4

1…3, 5…7

1…7

15

PCM

1…4

1…4

2…3

16

UKMO-HadCM3

1

1

1

Downscaling

Downscaling is a generic term used to describe the translation of low-resolution climate model output to higher resolution output. The 112 climate scenario simulations are too coarse to represent the variable climate across Colorado, but are capable of identifying patterns of broad-scale climate change. This study will not undertake its own downscaling procedure, but is making use of datasets already generated through a ‘spatial interpolation’ technique that has been made available through the Lawrence Livermore National Laboratory. The Bureau of Reclamation Technical Service Center, Santa Clara University Civil Engineering Department, and The Institute for Research on Climate Change and its Societal Impacts at Lawrence Livermore National Laboratory have statistically downscaled the results of the GCMs using a percentile mapping technique that substitutes real-world data for climate data while retaining the broadscale climate change signals (Wood, et al, 2004 and Mauer, 2007). The climate change signals were interpolated from a 2º latitude-longitude grid to a 1/8º latitude-longitude grid. The data frequency is monthly. The downscaled data provides important guidance and input for the sensitivity studies. It has two components: a bias correction (more accurately described as a correction of the entire climatological distribution) and a mapping onto local climatology that implicitly includes an adjustment for terrain. Therefore, the downscale procedure first improves the accuracy of the climate data and then applies the broad changes in climatological patterns within the local climatology. Computation of Offsets

Based on downscaled data for each GCM projection of interest, the WWA has generated temperature and precipitation offsets for the study region for the two future JFRCCVS Scope of Work

Page 10

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

periods, representing the 2040 and 2070 conditions, with guidance from the PSC and PIs. These offsets, averaged over the study region, will provide guidance in selecting individual scenarios of Delta TP adjustment for each future period. The scenarios will represent a reasonable range of possibilities for which the study participants wish to evaluate impacts on streamflow. The following procedure, developed in cooperation with a team working on a complimentary study with the CWCB, is proposed: Five qualitative scenarios have been identified as follows: • Hot and Dry • Hot and Wet • Warm and Dry • Warm and Wet • Median For each of the two future periods to be evaluated, a projection will be selected, such that up to ten projections may be used in the analysis (it also is possible that the same projection will be selected for both evaluation periods for a given qualitative scenario). For each scenario, a characteristic value will be determined for the projected change in temperature and precipitation. Change in temperature will be expressed as an absolute projected increase or decrease while change in precipitation will be expressed as a percent projected increase or decrease. The characteristic values will be determined as shown in Table 3. Table 3. Characteristics of selected climate scenarios

Scenario Characteristic T Characteristic P Hot and Dry 90th Percentile 10th Percentile Hot and Wet 70th Percentile 70th Percentile Warm and Dry 30th Percentile 30th Percentile th Warm and Wet 10 Percentile 90th Percentile Median 50th Percentile 50th Percentile Projections will be selected based on their proximity (in terms of Euclidean distance in the T and P dimension space) to the characteristic values for the five scenario points. Five neighbors will be selected as candidate projections at each scenario point. One of these candidate projections will be selected based on the following criteria: • Proximity to the characteristic point • Having a monthly precipitation pattern representative of the mean pattern of the five neighbors Once the specific GCM projections to be used in the analysis have been selected, average monthly precipitation and temperature offsets will be computed for the complete grid over the study area for use in the hydrologic simulation. It is noted that this procedure does not address how streamflow might respond to changes in precipitation frequency or intensity or the duration of droughts and wet periods. Task 4 – Preliminary Investigation

A preliminary analysis will be performed to demonstrate the hydrologic simulation approach, to determine the sensitivity of each basin to constant temperature increase (with no change to precipitation) and to constant precipitation adjustments (with no JFRCCVS Scope of Work

Page 11

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

change to temperature), and to evaluate the impact of hydrologic model bias. This part of the study uses minimal resources and should be completed quickly. The results will be presented to the project participants for discussion before further climate analyses take place. Sensitivity analysis

The same procedure the PIs will follow to simulate climate adjusted natural streamflow sequences will be used for this initial analysis and is described below in Task 5. The chosen constant temperature increases to be applied to each model are: • an increase of 1º C, and • an increase of 4º C. The chosen constant precipitation factors to be applied to each model are: • an increase of 7.5% • a decrease of 3% These changes reflect much of the variability in the projected changes expected through 2099 (IPCC AR4 Global “Best Estimates”). This analysis will provide insight as to how streamflow are impacted by independent changes in either temperature or precipitation. Additionally, this step will examine the model sensitivity to these simple deviations to be sure that model results are reasonable. Model Bias

The output of any hydrologic model can be biased relative to observed flows, and will also include random errors that are a function of uncertainties in parameter estimation during calibration, as well as inaccuracies in the input data. Such biases may be insignificant in the context of a hydrologic modeling task, yet in the context of water planning under climate change, these biases can have significant consequences. Two aspects of the study will limit the impact of model bias. The first is to apply wellcalibrated models as described previously. The second is to compare the climate adjusted hydrologic simulations with the unadjusted simulations to obtain the climate change signal, and then to apply only the change signal to the historical natural flows to obtain climate-adjusted natural flows. This is conveyed by the following equations: ΔQCC = QCCSim − QSim QCC = Q Nat + ΔQCC Where

QSim = unadjusted hydrologic simulation QCCSim = climate adjusted hydrologic simulation ΔQCC = climate change signal Q Nat = historical natural flow

QCC = climate adjusted natural flow To assess the effectiveness of this approach in limiting model bias, a bias elimination technique will be employed for a single sub-basin as part of the preliminary investigation. The envisioned procedure is as follows: • For the selected sub-basin, a monthly frequency analysis of mean daily flows will be prepared from the historic natural flows.

JFRCCVS Scope of Work

Page 12

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

•

A corresponding frequency analysis will be prepared from the simulated flows from the hydrologic models. • The bias model will be derived from the relationship between the observed and simulated monthly frequency analyses and will be used to develop bias adjusted time series values. The bias corrected climate change signal will be computed as the difference between the bias-corrected base simulation and the bias-corrected climate adjusted simulation. It will then be compared against the change signal computed without bias correction and monthly and annual volume differences between the two will be computed and evaluated. Task 5 – Streamflow Sensitivity to Changes in Climate

Climate impact modeling will be performed using the Delta TP Scenarios developed in Task 3. In this method natural hydrologic runoff under various climate change scenarios is estimated by making adjustments to historic temperature and precipitation inputs to calibrated hydrologic models. Each Delta TP scenario developed in Task 4 will consist of a set monthly 1/8 degree grids of temperature increases and a set of monthly 1/8 degree grids of precipitation adjustment factors covering the study area. These grids will be used to compute monthly adjustments to the temperature and precipitation inputs for each sub-basin or modeling unit defined in the respective hydrologic models. Model simulations will be executed for each scenario using the associated inputs. In each case, the climate change signal will be computed as the difference between the climate adjusted hydrologic simulation and the unadjusted simulation. This change signal will be applied to the natural flows for each node in the study. Simulations using the Sacramento model will result in daily time-series of climate adjusted natural flows at each node. Simulations using the WEAP model will result in weekly time series of climate adjusted natural flows. From these climate sensitivity experiments the research team will report on hydrologic changes that include changes in precipitation including phase (rain vs. snow), temporal distribution (seasonality), and spatial distribution (location). In addition, changes in snowpack, snow-pack melt-off, and change in evapotranspiration, soil moisture, and streamflow will be examined. Task 6 - Reports

Three periodic (quarterly) reports will be prepared in addition to the draft and final reports for the study. The periodic progress reports will provide detailed information about completed, on-going work efforts, and upcoming tasks expected to begin during the next 3-month period. The second periodic report will include a technical summary of work completed to date. Sections of the final report such as the literature review, methodology, bibliography, glossary, and other sections will be included as part of the technical summary report. Table 4 outlines the anticipated schedule of interim and final work products.

JFRCCVS Scope of Work

Page 13

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

Table 4. Quarterly report schedule

Report # Periodic Report #1 Periodic Report #2

Due Date January 26, 2009 April 27, 2009

Periodic Report #3 Draft Report

July 27, 2009

Final Report

August 24, 2009 January 25, 2010

Description and Products Included Work completed during first 3 months of project – data collection, hydrologic modeling update

Work completed during first 6 months of project – natural flows, hydrologic modeling, climatology adjustment methodology Work completed during first 9 months of project – climate impact modeling and analysis Draft final report summarizing findings and methodology. A three-month review and revision period is provided.

Deliverables The following deliverables have been identified as outcomes of the above scope of work: 1. Natural flows – A complete set of computed natural flow time series will be developed as part of the study for the specified nodes in the study area. 2. Historical precipitation and temperature datasets – The historical datasets used to drive the hydrologic models and that form the basis for adjustments to represent varying climate conditions will be developed as part of the study and will be compiled and made available to project participants. 3. Calibrated hydrologic models – Hydrologic models designed and calibrated to compute natural streamflow based on given climate inputs will be developed. The resulting model parameters will be made available for use by project participants. With additional funding, automated procedures could be developed and implemented to permit participants to easily reproduce study results and evaluate additional scenarios. 4. Gridded Delta TP datasets – The gridded datasets of temperature increase and precipitation adjustment factors for all scenarios for both of the two future time periods evaluated will be compiled and made available as part of the study. 5. Climate adjusted streamflow – Climate adjusted streamflow time series for specified nodes for all scenarios for both of the two future time periods evaluated will be generated and made available as part of the study. 6. Periodic Reports – Quarterly periodic reports will be prepared and submitted to the Foundation and the study participants. 7. Draft Report – A draft report will be submitted documenting all of the activities of the study and presenting the results and evaluation of the potential impacts of climate change on water availability. 8. Final Report – A final report will be submitted incorporating comments from the Foundation project team.

JFRCCVS Scope of Work

Page 14

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED

Outreach A key aspect of this study is the use of regular educational sessions to develop the scope of the study, to educate study participants regarding climate change and hydrological science, and to provide regular updates on the study progress. Examples of educational topics include: • Climate variables and modeling techniques (including emissions scenarios, GCMs, and downscaling approaches) through educational sessions with the WWA, • The process of developing standardized historic naturalized streamflow sequences, with oversight from the CWCB, • Hydrologic models calibrated for several river basins in Colorado, • The procedures required for this type of sensitivity analysis, and • Correctly interpreting study results with guidance from WWA. Project participants, including the members of the PSC, have noted significant interest in this study, both within the State of Colorado as well as nationally. Additional communities along Colorado’s Front Range and into Wyoming have begun participating in the study’s educational sessions. Presentations discussing the proposed project have been given in several local and national forums. The project team anticipates submitting papers to conferences and technical periodicals and journals as the study proceeds and concludes, and will seek assistance and collaboration with the Foundation in sharing the insights derived from this study. References Doherty, J. 2002, Model Independent Parameter Estimation, (PEST), User’s Manual, 5th Addition, Watermark Numerical Computing, 7944 Wisconsin Ave, Bethesda MD 20814. IPCC (2000). “IPCC Special Report on Emissions Scenarios.” Report can be found at http://www.ipcc.ch/ipccreports/sres/emission/index.htm. Maurer E.P., A.W. Wood, J.D. Adam, D.P. Lettenmaier, and B. Nijssen (2002) A longterm hydrologically-based data set of land surface fluxes and states for the conterminous United States. J Climate 15(22):3237-3251. Maurer, E.P. (2007): Uncertainty in hydrologic impacts of climate change in the Sierra Nevada, California under two emissions scenarios. Climatic Change 82:309-325. Wood, A. W., Leung, L. R., Sridhar, V., and Lettenmaier, D. P. (2004): Hydrologic implications of dynamical and statistical approaches to downscaling climate model outputs, Climatic Change, 62, 189-216. Yates, D., D. Purkey, H. Galbraith, A. Huber-Lee, and J. Sieber, 2005b, WEAP21 a demand, priority, and preference driven water planning model: Part 2, Aiding freshwater ecosystem service evaluation, Water International, 30,4, pp. 501-512.

JFRCCVS Scope of Work

Page 15

1/9/2009 .

©2009 Water Research Foundation. ALL RIGHTS RESERVED