• Ackerman, F., , and E. A. Stanton, 2011: The last drop: Climate change and the Southwest water crisis. Stockholm Environmental Institute Rep., 36 pp.

  • Apling, D., , G. Higgins, , H. Kiley, , and K. Darmenova, 2010: Analysis and quantification of uncertainty for climate change decision aids for energy consumption in the southwestern US. 2010 Fall Meeting, San Francisco, CA, Amer. Geophys. Union, Abstract NG23D-05.

  • Archer, S. R., , and K. I. Predick, 2008: Climate change and ecosystems of the southwestern United States. Rangelands, 30, 2328.

  • Barnett, T. P., and Coauthors, 2008: Human-induced changes in the hydrology of the western United States. Science, 319, 10801083.

  • Bengtsson, L., , K. Hodges, , and E. Roeckner, 2006: Storm tracks and climate change. J. Climate, 19, 35183543.

  • Box, G., , G. M. Jenkins, , and G. C. Reinel, 2008: Time Series Analysis: Forecasting and Control. 4th ed. Wiley Series in Probability and Statistics, John Wiley and Sons, 784 pp.

  • Brunetti, M., , M. Maugeri, , and T. Nanni, 2001: Changes in total precipitation, rainy days and extreme events in northeastern Italy. Int. J. Climatol., 21, 861871.

    • Search Google Scholar
    • Export Citation
  • Cayan, D., , T. Das, , D. Pierce, , T. Barnett, , M. Tyree, , and A. Gershunov, 2010: Future dryness in the southwest US and the hydrology of the early 21st century drought. Proc. Natl. Acad. Sci. USA, 107, 21 27121 276.

    • Search Google Scholar
    • Export Citation
  • Cohn, S. E., 1997: An introduction to estimation theory. J. Meteor. Soc. Japan, 75, 257288.

  • Daley, R., 1991: Atmospheric Data Analysis. Cambridge Atmospheric and Space Science Series, Vol. 2, Cambridge University Press, 457 pp.

  • Darmenova, K., , G. Higgins, , R. Alliss, , H. Kiley, , and D. Apling, 2010: Development of dynamical downscaling for climate decision aid applications. Extended Abstracts, 11th WRF Users’ Workshop, Boulder, CO, National Center for Atmospheric Research, 6 pp. [Available online at http://www.mmm.ucar.edu/wrf/users/workshops/WS2010/abstracts/P-27.pdf.]

  • Diaz, H. F., , and J. K. Eischeid, 2007: Disappearing “alpine tundra” Köppen climatic type in the western United States. Geophys. Res. Lett., 34, L18707, doi:10.1029/2007GL031253.

    • Search Google Scholar
    • Export Citation
  • DiLuzio, M., , G. L. Johnson, , C. Daly, , J. K. Eischeid, , and J. G. Arnold, 2008: Constructing retrospective gridded daily precipitation and temperature datasets for the conterminous United States. J. Appl. Meteor. Climatol., 47, 475497.

    • Search Google Scholar
    • Export Citation
  • Dominguez, F., , J. Cañon, , and J. Valdes, 2010: IPCC-AR4 climate simulations for the Southwestern US: The importance of future ENSO projections. Climatic Change, 99, 499514.

    • Search Google Scholar
    • Export Citation
  • Fisher, J., , and F. Ackerman, 2011: The water-energy nexus in the western states: Projections to 2100. Stockholm Environment Institute Rep., 39 pp. [Available online at http://sei-us.org/Publications_PDF/SEI-WesternWater-Energy-0211.pdf.]

  • Gu, H., , G. Wang, , Z. Yu, , and R. Mei, 2012: Assessing future climate changes and extreme indicators in east and south Asia using the RegCM4 regional climate model. Climatic Change, 114, 301317.

    • Search Google Scholar
    • Export Citation
  • Guido, Z., cited 2008: Climate and floods in the Southwest. Southwest Climate Change Network. [Available online at http://www.southwestclimatechange.org/impacts/water/floods.]

  • Gutzler, D. S., , and T. O. Robbins, 2010: Climate variability and projected change in the western United States: Regional downscaling and drought statistics. Climate Dyn., 37, 835849.

    • Search Google Scholar
    • Export Citation
  • Haerter, J. O., , S. Hagemann, , C. Moseley, , and C. Piani, 2011: Climate model bias correction and the role of timescales. Hydrol. Earth Syst. Sci., 15, 10651079.

    • Search Google Scholar
    • Export Citation
  • Hagemann, S., , K. Arpe, , and E. Roeckner, 2006: Evaluation of the hydrological cycle in the ECHAM5 model. J. Climate, 19, 38103827.

  • Higgins, G., , K. Darmenova, , D. Apling, , and H. Kiley, 2011: Regional climate modeling and energy demand projections for selected Colorado sites. Proc. Second Conf. on Weather, Climate, and the New Energy Economy, Seattle, WA, Amer. Meteor. Soc., 800. [Available online at https://ams.confex.com/ams/91Annual/webprogram/Paper182988.html.]

  • Jin, J., , S.-Y. Wang, , and R. Gillies, 2011: An improved dynamical downscaling for the western United States. Climate Change—Research and Technology for Adaptation and Mitigation, J. Blanco and H. Kheradmand, Eds., InTech, 23–38. [Available online at http://cdn.intechopen.com/pdfs/18711/InTech-An_improved_dynamical_downscaling_for_the_western_united_states.pdf.]

  • Jones, P. D., , and A. Moberg, 2003: Hemispheric and large-scale surface air temperature variations: An extensive revision and an update to 2001. J. Climate, 16, 206223.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437471.

  • Kinney, P., , M. O’Neill, , M. Bell, , and J. Schwartz, 2008: Approaches for estimating effects of climate change on heat-related deaths: Challenges and opportunities. Environ. Sci. Policy, 11, 8796.

    • Search Google Scholar
    • Export Citation
  • Knowles, N., , M. D. Dettinger, , and D. R. Cayan, 2006: Trends in snowfall versus rainfall in the western United States. J. Climate, 19, 45454559.

    • Search Google Scholar
    • Export Citation
  • Kottek, M., , and F. Rubel, 2007: Global daily precipitation fields from bias-corrected rain-gauge and satellite observations. Part I: Design and development. Meteor. Z., 16, 525539.

    • Search Google Scholar
    • Export Citation
  • Leung, L. R., , and Y. Qian, 2009: Atmospheric rivers induced heavy precipitation and flooding in the western U.S. simulated by the WRF regional climate model. Geophys. Res. Lett., 36, L03820, doi:10.1029/2008GL036445.

    • Search Google Scholar
    • Export Citation
  • Liang, X.-Z., , J. Pan, , J. Zhu, , K. E. Kunkel, , J. X. L. Wang, , and A. Dai, 2006: Regional climate model downscaling of the U.S. summer climate and future change. J. Geophys. Res., 111, D10108, doi:10.1029/2005JD006685.

    • Search Google Scholar
    • Export Citation
  • Liang, X.-Z., , K. Kunkel, , G. Meehl, , R. Jones, , and J. X. L. Wang, 2008: Regional climate models downscaling analysis of general circulation models present climate biases propagation into future change projections. Geophys. Res. Lett., 35, L08709, doi:10.1029/2007GL032849.

    • Search Google Scholar
    • Export Citation
  • MacDonald, G., 2010: Water, climate change, and sustainability in the Southwest. Proc. Natl. Acad. Sci. USA, 107, 21 25621 262.

  • Maurer, E. P., , and H. G. Hidalgo, 2008: Utility of daily vs. monthly large-scale climate data: An intercomparison of two statistical downscaling methods. Hydrol. Earth Syst. Sci., 12, 551563.

    • Search Google Scholar
    • Export Citation
  • Nakicenovic, N., , and R. Swart, Eds., 2000: Special Report on Emissions Scenarios. Cambridge University Press, 599 pp. [Available online at http://www.grida.no/climate/ipcc/emission/index.htm.]

  • Pan, L.-L., , S.-H. Chen, , D. Cayan, , M.-Y. Lin, , Q. Hart, , M.-H. Zhang, , Y. Liu, , and J. Wang, 2011: Influences of climate change on California and Nevada regions revealed by a high-resolution dynamical downscaling study. Climate Dyn., 37, 20052020.

    • Search Google Scholar
    • Export Citation
  • Piani, C., , J. O. Haerter, , and E. Coppola, 2010: Statistical bias correction for daily precipitation in regional climate models over Europe. Theor. Appl. Climatol., 99, 187192.

    • Search Google Scholar
    • Export Citation
  • Qian, Y., , S. J. Ghan, , and L. R. Leung, 2010: Downscaling hydroclimatic changes over the Western US based on CAM subgrid scheme and WRF regional climate simulations. Int. J. Climatol., 30, 675693.

    • Search Google Scholar
    • Export Citation
  • Ray, A. J., , J. J. Barsugli, , K. B. Averyt, , K. Wolter, , M. Hoerling, , N. Doesken, , B. Udall, , and R. S. Webb, 2008: Climate change in Colorado: A synthesis to support water resources management and adaptation. University of Colorado at Boulder Rep., 58 pp. [Available online at http://wwa.colorado.edu/publications/reports/WWA_ClimateChangeColoradoReport_2008.pdf.]

  • Robles, M. D., , and C. Enquist, 2010: Managing changing landscapes in the southwestern United States. Nature Conservancy Rep., Tucson, AZ, 26 pp. [Available online at http://azconservation.org/downloads/managing_changing_landscapes_in_the_southwestern_united_states.]

  • Roeckner, E., and Coauthors, 2003: The atmospheric general circulation model ECHAM5 Part 1. Rep. 349, Max Planck Institute for Meteorology, 140 pp.

  • Salathé, E. P., , R. Steed, , C. F. Mass, , and P. H. Zahn, 2008: A high-resolution climate model for the U.S. Pacific Northwest: Mesoscale feedbacks and local responses to climate change. J. Climate, 21, 57085726.

    • Search Google Scholar
    • Export Citation
  • Salathé, E. P., , L. R. Leung, , Y. Qian, , and Y. Zhang, 2010: Regional climate model projections for the State of Washington. Climatic Change, 102, 5175.

    • Search Google Scholar
    • Export Citation
  • Seager, R., , M. Ting, , and I. Held, 2007: Model projections of an imminent transition to a more arid climate in Southwestern North America. Science, 5828, 11811184.

    • Search Google Scholar
    • Export Citation
  • Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF Version 3. NCAR Tech. Note NCAR/TN-475+STR, 125 pp. [Available online at http://www.mmm.ucar.edu/wrf/users/docs/arw_v3.pdf.]

  • Solomon, S., , D. Qin, , M. Manning, , Z. Chen, , M. Marquis, , K. B. Averyt, , M. Tignor, , and H. L. Miller, Eds., 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.

  • Steadman, R. G., 1979: The assessment of sultriness. Part I: A temperature-humidity index based on human physiology and clothing science. J. Appl. Meteor., 18, 861873.

    • Search Google Scholar
    • Export Citation
  • Steadman, R. G., 1984: A universal scale of apparent temperature. J. Climate Appl. Meteor., 23, 16741282.

  • Stull, R., 2000: Meteorology for Scientists and Engineers. 2nd ed. Brooks/Cole, 60 pp.

  • Swinbank, R., , and W. A. Lahoz, 2003: Data Assimilation for the Earth System. NATO Science Series: IV, Vol. 26, Kluwer, 388 pp.

  • Themeßl, M. J., , A. Gobiet, , and A. Leuprecht, 2011: Empirical-statistical downscaling and error correction of daily precipitation from regional climate models. Int. J. Climatol., 31, 15301544.

    • Search Google Scholar
    • Export Citation
  • Wi, S., , F. Dominguez, , M. Durcik, , J. Valdes, , H. Diaz, , and C. Castro, 2012: Climate change projection of snowfall in the Colorado River basin using dynamical downscaling. Water Resour. Res., 48, W05504, doi:10.1029/2011WR010674.

    • Search Google Scholar
    • Export Citation
  • Winkler, J., and Coauthors, 2011a: Climate scenario development and applications for local/regional climate change impact assessments: An overview for the non-climate scientist. Part I: Scenario development using downscaling methods. Geogr. Compass, 5, 275300.

    • Search Google Scholar
    • Export Citation
  • Winkler, J., , G. Guentchev, , M. Liszewska, , Perdinan, , and P.-N. Tan, 2011b: Climate scenario development and applications for local/regional climate change impact assessments: An overview for the non-climate scientist. Part II: Considerations when using climate change scenarios. Geogr. Compass, 5, 301328.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    WRF domain setup.

  • View in gallery

    Monthly mean temperature bias (2002–09) of NCEP–WRF with respect to the PRISM dataset: (a) spring, (b) summer, (c) autumn, and (d) winter and of ECHAM5–WRF with respect to the PRISM dataset: (e) spring, (f) summer, (g) autumn, and (h) winter.

  • View in gallery

    As in Fig. 2, but for precipitation.

  • View in gallery

    PDFs of daily mean temperature (°C) (2002–09) from GHCN data, NCEP–WRF, and ECHAM5–WRF for Colorado Springs: (a) spring, (b) summer, (c) autumn, and (d) winter and of daily mean precipitation (mm): (e) spring, (f) summer, (g) autumn, and (h) winter.

  • View in gallery

    Diagram of bias-correction approach.

  • View in gallery

    ECHAM5–WRF-simulated differences (2030s − 2000s) in daily mean temperature: (a) spring, (b) summer, (c) autumn, and (d) winter and daily precipitation: (e) spring, (f) summer, (g) autumn, and (h) winter.

  • View in gallery

    Average temperature difference and precipitation change for (a) Colorado, (b) Arizona, and (c) New Mexico, and bivariate PDF of temperature difference and precipitation change for (d) Colorado, (e) Arizona, and (f) New Mexico.

  • View in gallery

    ECHAM5–WRF temperature change with elevation: (a) current (2000s) and (b) future (2030s), and the (c) difference (2030s − 2000s).

  • View in gallery

    ECHAM5–WRF differences (future − current) in (a) rainy days and (b) days with total precipitation exceeding 1 in.

  • View in gallery

    Relative change in ECHAM5–WRF annual-mean accumulated snow.

  • View in gallery

    Mean annual bias-corrected difference (2030s − 2000s) in (a) CDD and (b) HDD.

  • View in gallery

    Difference (2030s − 2000s) in annual-mean days with heat index exceeding 35°C.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 35 35 6
PDF Downloads 12 12 2

Assessment of Regional Climate Change and Development of Climate Adaptation Decision Aids in the Southwestern United States

View More View Less
  • 1 Energy and Environment Department, Northrop Grumman Information Systems, Chantilly, Virginia
© Get Permissions
Full access

Abstract

Over the next 10–50 years, policy makers in the southwestern United States are faced with complex planning and policy issues associated with increasing water and energy demand as a result of warmer temperatures and reduced availability of water, compounded by continued rapid population growth and economic development. This study uses a top-down, end-to-end approach consisting of dynamical downscaling, a novel bias-correction technique, and custom-developed decision-aid tools to assess regional climate changes in the Southwest and to derive decision aids that are based on direct communication with the planners at four military installations in the region. Dynamical downscaling is performed with the Weather Research and Forecasting model driven by the National Centers for Environmental Prediction reanalysis and the Max Planck Institute for Meteorology’s ECHAM5 general circulation model for two time periods: current (2000s) and future (2030s). A unique two-stage bias correction is developed to adjust current and future hourly temperature and precipitation to be consistent with historical reference data. The authors’ assessment of regional climate change, which is based on downscaled bias-corrected fields, points to a dryer and warmer future climate in the Southwest. The energy-usage modeling produced a statistically significant increase in natural gas consumption and a possible decrease in electricity usage in two military installations in Colorado, which is a direct consequence of decrease/increase in heating/cooling degree-days resulting from warmer temperatures in the future. In addition, the results indicate an increasing number of oppressive heat days in the future, which may impact long-term planning practices with respect to heat-stress control and heat-casualty management.

Corresponding author address: Kremena Darmenova, Energy and Environment Dept., Northrop Grumman Information Systems, 15010 Conference Center Dr., Chantilly, VA 20151. E-mail: kremena.darmenova@ngc.com

Abstract

Over the next 10–50 years, policy makers in the southwestern United States are faced with complex planning and policy issues associated with increasing water and energy demand as a result of warmer temperatures and reduced availability of water, compounded by continued rapid population growth and economic development. This study uses a top-down, end-to-end approach consisting of dynamical downscaling, a novel bias-correction technique, and custom-developed decision-aid tools to assess regional climate changes in the Southwest and to derive decision aids that are based on direct communication with the planners at four military installations in the region. Dynamical downscaling is performed with the Weather Research and Forecasting model driven by the National Centers for Environmental Prediction reanalysis and the Max Planck Institute for Meteorology’s ECHAM5 general circulation model for two time periods: current (2000s) and future (2030s). A unique two-stage bias correction is developed to adjust current and future hourly temperature and precipitation to be consistent with historical reference data. The authors’ assessment of regional climate change, which is based on downscaled bias-corrected fields, points to a dryer and warmer future climate in the Southwest. The energy-usage modeling produced a statistically significant increase in natural gas consumption and a possible decrease in electricity usage in two military installations in Colorado, which is a direct consequence of decrease/increase in heating/cooling degree-days resulting from warmer temperatures in the future. In addition, the results indicate an increasing number of oppressive heat days in the future, which may impact long-term planning practices with respect to heat-stress control and heat-casualty management.

Corresponding author address: Kremena Darmenova, Energy and Environment Dept., Northrop Grumman Information Systems, 15010 Conference Center Dr., Chantilly, VA 20151. E-mail: kremena.darmenova@ngc.com

1. Introduction

There is a broad consensus that the U.S. Southwest will become drier and hotter in the twenty-first century (Dominguez et al. 2010; Seager et al. 2007; Solomon et al. 2007). Current projections for regional climate change in the Southwest include fewer frost days, warmer temperatures, and an increased frequency of extreme-weather events (heat waves, droughts, and floods) (Archer and Predick 2008). Projected trends in precipitation in the Southwest have higher levels of uncertainty relative to temperature projections. Nonetheless, the Southwest is one of the few regions globally for which there is consistent agreement among the Intergovernmental Panel on Climate Change (IPCC) projections, which point to a decrease in streamflow and an increase in drought conditions (Dominguez et al. 2010). Over the next 10–50 years, policy makers in the Southwest are facing complex planning and policy issues associated with increasing water and energy demand as a result of warmer temperatures, potentially reduced availability of water, and continued rapid population increase (MacDonald 2010; Fisher and Ackerman 2011).

The number of climate change assessment reports and regional climate studies focusing on local responses to climate change in the Southwest has steadily increased over the last decade (e.g., Ray et al. 2008; Robles and Enquist 2010; Cayan et al. 2010; Gutzler and Robbins 2010; Ackerman and Stanton 2011). The majority of the studies either used the IPCC Assessment Report (AR4) general circulation model (GCM) data directly or used a combination of AR4, statistically downscaled data, and basin-scale hydrological models driven by the future climate scenarios to assess temperature and precipitation changes and potential impacts on water resources planning in the region.

Fewer studies have applied dynamical downscaling to assess regional climate change in the Southwest (e.g., Qian et al. 2010; Jin et al. 2011; Wi et al. 2012). Dynamical downscaling’s advantages relative to other downscaling methods (i.e., spatial disaggregation or statistical downscaling) include resolving atmospheric processes on a finer scale (e.g., orographic effects in mountainous areas) and producing an output that is based on physical modeling of the related processes. Dynamical downscaling is computationally expensive, however, and regional climate model (RCM) simulations are usually performed for relatively short periods from a climatological perspective (i.e., 10–30 yr). The majority of the dynamical downscaling studies in the Southwest were performed on 10-yr time slices. The future 10-yr periods were centered in the middle of the twenty-first century, with the exception of Wi et al. (2011), who performed a continuous 111-yr simulation for the 1969–2079 period. The spatial resolution of the downscaled fields varied between 15 and 50 km, with the exception of the Pan et al. (2011) study, in which a 4-km model grid was used over the states of California and Nevada.

It has to be pointed out that realistic regional dynamical downscaling is achievable only when driven by unbiased GCM forcing or when the RCM results are bias corrected with respect to a reference dataset. This is especially critical when the results from dynamical downscaling are used for deriving decision-aid products and actionable information used for planning and risk assessment. A common approach for bias correcting climate fields is the quantile-mapping technique, which corrects the distributions of selected model output variables by matching empirically determined quantiles of distributions (Maurer and Hidalgo 2008; Haerter et al. 2011; Themeßl et al. 2011). Other commonly used bias-correction techniques are summarized in Winkler et al. (2011b) and Haerter et al. (2011). The majority of statistical bias-correction methods use transfer-function correction. The primary differences between them are the construction of the transfer functions; whether these are single-parameter linear (change-factor approaches), two-parameter linear, multiparameter nonlinear, or cumulative distributional corrections (bias correction and spatial downscaling; quantile–quantile bias correction). Furthermore, Haerter et al. introduced treatment of the roles of time scales within the same transfer-function framework. In this study, we develop a method that introduces a novel treatment of both time scales and temporal autocorrelations by explicitly modeling the time scales as conditional parameterizations of the transfer functions and by introducing an intermediate data representation as an autoregressive model.

In addition, whereas the majority of the downscaling studies in the Southwest focused on assessing the climate change signals in temperature and precipitation, only a few went further into assessing changes in the extreme-events statistics (e.g., Pan et al. 2011) and into describing the implications of climate change for water resources and energy and agricultural management (e.g., Ray et al. 2008). None of the existing studies for the Southwest applied the comprehensive top-down “feed forward” approach for decision-aid development described in Winkler et al. (2011a) that links bias-adjusted downscaled climate scenarios with environmental/impact and decision-making models to develop actionable information on local and regional scales. This study uses a consistent top-down approach, which consists of dynamical downscaling, validation and bias correction, and custom-developed decision-aid tools at fine spatial scale (12 km). The goals of this study are to provide 1) a bias-correction technique, which introduces a novel treatment of both time scales and temporal autocorrelations, 2) assessment of the regional climate change in the Southwest in the 2030s by quantifying the changes in the spatiotemporal characteristics of the bias-adjusted variables, and 3) climate decision aids from the downscaled bias-corrected data on the basis of direct communication with specific end users (i.e., military installations).

In this study, actionable information will be referred to as decision aids. We define a decision aid as an environmental product specifically tailored to meet the needs of a set of decision makers. A decision aid applies the current best-practice scientific modeling and engineering analysis at appropriate spatial and temporal scales to characterize the operational impact of an environmental effect or process on a human or socioeconomic activity. A decision aid also carries a statistical bound on uncertainty: preferably quantitative as a confidence interval.

This study focuses on the 2030s as a planning time horizon that is at the low end of long-term planning strategies. This is far enough into the future to have evidence of a climate change signature that is statistically significant with respect to interannual variability and yet sufficiently near term to cover the targeted user community’s decision horizon [e.g., Fort Carson, Colorado; Peterson Air Force Base (AFB), Colorado; Fort Bliss, Texas; and Fort Irwin, California]. This study also provides a glimpse of the issues and questions that installation planners need addressed and that in a way are very similar to the needs that local and state planners have for actionable engineering information that is based on future climate projections. It is significant that, for the first time in their planning practices, these facility planners considered predictions that are based on downscaled data for future climate scenarios.

The paper is organized as follows: the modeling approach is described in section 2, section 3 discusses model validation and intercomparison with observational data, section 4 focuses on the development of our two-stage bias-correction technique of simulated temperature and precipitation, section 5 presents the regional analysis of the major hydrometeorological variables, section 6 introduces several derived climate-adaptation decision aids and shows translation of climate data into actionable information, and section 7 summarizes our findings.

2. Modeling approach

For dynamical downscaling we used version 3.1.1 of the Weather Research and Forecasting (WRF) model (Skamarock et al. 2008). The WRF model is a fully compressible, nonhydrostatic model with a terrain-following hydrostatic pressure vertical coordinate. A significant advantage of WRF is its nesting capability that allows for multiple (nonoverlapping) domains at the same nest level and multiple nest levels (telescoping). WRF also supports three-dimensional analysis nudging as well as multiple parameterizations of the related physical processes (boundary layer, microphysics, longwave and shortwave radiation, and land–atmosphere interactions). The WRF model has been developed primarily for short-term weather prediction, but it has been successfully used in long-term regional climate studies (Salathé et al. 2008; Leung and Qian 2009). We utilized a number of features implemented in the WRF model that allow realistic representation of the climate system in long-term simulations: for example, variable carbon dioxide concentrations, boundary-condition SST updates, and deep soil temperature update (Salathé et al. 2008). We used the ECHAM5–Max Planck Institute Ocean Model (MPI-OM) coupled-model (Roeckner et al. 2003) runs at 1.9° spatial resolution that were performed with the Special Report on Emissions “A1B” emissions scenario (Nakicenovic and Swart 2000), available from the World Data Center for Climate, Hamburg, Germany (http://cera-www.dkrz.de), for WRF initial and lateral boundary conditions. The atmospheric component of ECHAM5–MPI-OM is the fifth-generation of the general circulation model developed at the European Centre for Medium-Range Weather Forecasts (ECMWF) and the Max Planck Institute for Meteorology in Hamburg. The ocean–sea ice component is the MPI-OM. Previous studies (Bengtsson et al. 2006; Salathé et al. 2010) showed that ECHAM5 performed well both globally and regionally in the western United States in terms of temperature, precipitation, and storm-track statistics when compared with ground-based observations and 40-yr ECMWF Re-Analysis (ERA-40) data. The climate sensitivity in the ECHAM5 model is at the middle range among all IPCC AR4 models (Gu et al. 2012), which ensured that our dynamical downscale is not an outlier. In addition, we used the National Centers for Environmental Prediction (NCEP) reanalysis data (Kalnay et al. 1996) at 2.5° spatial resolution to initialize the WRF run that served as a historical baseline used both for comparison with the ECHAM5 downscale and for bias correction.

Two sets of WRF runs were performed for the current climate conditions (2000–09). WRF was driven with the NCEP reanalysis data and also with the ECHAM5 GCM output. WRF was also run with ECHAM5 output for a future period (2030–39). We used a 1-yr spinup for the three sets of runs to ensure an equilibrium state of the terrestrial hydrology. The ECHAM5 twenty-first-century A1B scenario run started on 1 January 2001 so that our ECHAM5–WRF downscale is shorter than the NCEP–WRF current-period downscale and the ECHAM5–WRF run for the future period. The model integration was conducted in continuous mode without interruption during the entire simulation period. WRF was run in nested mode (see Fig. 1) at spatial resolutions of 108, 36, and 12 km with 28 vertical levels. The coarser 108-km WRF grid was run with nudging terms for horizontal winds, temperature, and water vapor. In our simulations we nudged the winds at all sigma levels; temperature and water vapor mixing ratio were nudged above the 10th model level. We used the nudging configuration used in Salathé et al. (2008, 2010), who nudged the outermost domain, thus allowing the mesoscale model to preserve the large-scale state provided by the global model while generating regional meteorological details on the inner nests. Darmenova et al. (2010) showed that grid nudging has important implications for accurate representation of historical synoptic conditions through preventing development of large discrepancies between the boundary conditions and the regional model.

Fig. 1.
Fig. 1.

WRF domain setup.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

The model output was saved on 1-h intervals. In this study, the microphysics and convective parameterizations used were the WRF single-moment 5-class (WSM5) scheme and the Kain–Fritsch scheme. The land surface model (LSM) and planetary boundary layer (PBL) scheme used were the “Noah” LSM and Yonsei University PBL. Shortwave (SW) and longwave (LW) radiation were computed with the Community Atmosphere Model (CAM) SW and LW schemes.

3. Model comparison and validation

To assess the performance of the WRF model in reproducing the current regional climate of the southwestern United States, we compared the model simulations with the Oregon State University Parameter–Elevation Regressions on Independent Slopes Model (PRISM) gridded observational dataset (DiLuzio et al. 2008) and with the Global Historical Climatology Network (GHCN) data (Jones and Moberg 2003). Figure 2 shows the NCEP–WRF and ECHAM5–WRF monthly mean temperature model bias with respect to the PRISM dataset for the 2002–09 period. NCEP–WRF shows a consistent warm bias (2°–3°C) over the high plains, New Mexico, and central Arizona in all seasons. In addition, the NCEP–WRF downscale shows a cold bias over the Colorado Rocky Mountains, the Wasatch Range, and the Uinta Mountains in Utah. The ECHAM5–WRF has a cold bias throughout the year for most of the southwestern United States except for the high plains and the Sacramento and San Joaquin Valleys in California.

Fig. 2.
Fig. 2.

Monthly mean temperature bias (2002–09) of NCEP–WRF with respect to the PRISM dataset: (a) spring, (b) summer, (c) autumn, and (d) winter and of ECHAM5–WRF with respect to the PRISM dataset: (e) spring, (f) summer, (g) autumn, and (h) winter.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

The ECHAM5–WRF and NCEP–WRF monthly precipitation biases (see Fig. 3) show similar behavior: from spring to autumn the two downscales have a dry bias in the Mojave and Sonoran Deserts and southern California and a wet bias over Colorado and most of New Mexico. In the winter, both downscales show a wet bias throughout the entire domain with the exception of a small portion of the Mojave in the NCEP–WRF downscale. In addition, the ECHAM5–WRF downscale is noticeably wetter relative to the NCEP–WRF downscale. To identify the source of the ECHAM5–WRF wet bias we analyzed the raw 1.9° × 1.9° ECHAM5 precipitation, cloud cover, and temperature data. We found that ECHAM5 tends to produce more cloud coverage, more precipitation, and colder temperatures; this is ultimately translated to the WRF downscale. This is in agreement with previous studies (e.g., Hagemann et al. 2006; Salathé et al. 2008), which found a wet bias in the ECHAM5-simulated precipitation over the oceans and most of the major continental river-system catchments.

Fig. 3.
Fig. 3.

As in Fig. 2, but for precipitation.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

To investigate the performance of the dynamically downscaled data for individual locations, which are of potential interest for deriving local decision-aid products, we compared modeled temperatures with GHCN station data. Figure 4 shows the probability density function (PDF) of daily mean temperatures for spring, summer, autumn, and winter in Colorado Springs, Colorado, from surface observations and for the closest grid cell from the NCEP–WRF and the ECHAM5–WRF downscales. In spring and autumn the ECHAM5–WRF has a cold bias relative to GHCN data; in the summer months the NCEP–WRF has a considerable warm bias. The NCEP–WRF precipitation agrees reasonably well with the surface observations throughout the seasons, whereas the ECHAM5–WRF overestimates the daily precipitation.

Fig. 4.
Fig. 4.

PDFs of daily mean temperature (°C) (2002–09) from GHCN data, NCEP–WRF, and ECHAM5–WRF for Colorado Springs: (a) spring, (b) summer, (c) autumn, and (d) winter and of daily mean precipitation (mm): (e) spring, (f) summer, (g) autumn, and (h) winter.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

Our analysis of the seasonal temperature and precipitation model biases indicated that biases for individual locations/regions can be large. This prompted us to develop the two-stage bias-correction method that is described in detail in section 4.

4. Bias correction

We execute the bias correction in two stages (see Fig. 5). The first stage constrains the reference NCEP–WRF temperature and precipitation time series to closely match the monthly climatological statistics from PRISM. This calibrates the reference dataset as a more accurate historical reconstruction from which to proceed to the ECHAM5–WRF bias corrections. In the second stage, we apply techniques from time series analysis that are methodologically similar to the multivariate statistical techniques commonly used in data assimilation systems (Daley 1991; Cohn 1997; Swinbank and Lahoz 2003). These identify and remove model biases in means, variances, and autocorrelations in the current and future period downscales. These bias-correction procedures are first applied to the current ECHAM5–WRF hourly time series, using the bias-corrected NCEP–WRF time series from stage 1; next, they are applied to the ECHAM5–WRF future time series.

Fig. 5.
Fig. 5.

Diagram of bias-correction approach.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

Our two-stage bias correction is explained in detail below. The first-stage temperature bias correction uses PRISM monthly-averaged minimum and maximum daily temperatures for the current period to derive linear-transform coefficients for each month of the NCEP–WRF simulation, such that monthly-averaged NCEP–WRF minimum and maximum temperatures match the PRISM values at each grid point. The individual hourly temperature values for each model grid cell are then transformed linearly according to Eq. (1):
e1
where TNCEP represents the original NCEP–WRF temperature for a given date and hour at that grid cell, and are the datasets’ average minimum and maximum temperatures for the month at that grid cell, and are the PRISM temperatures for the same month and grid cell, and T′ is the resulting bias-corrected temperature.
Monthly precipitation means from PRISM were used similarly to derive a linear transform to correct the NCEP–WRF downscale on a time series basis. This transformation is shown in Eq. (2):
e2
The variable p represents the original NCEP–WRF downscale precipitation rate for a given date and hour at each grid cell, is the average hourly liquid water precipitation rate determined by units conversion from the PRISM monthly accumulated inches of liquid water equivalent, is the monthly-averaged liquid water hourly precipitation rate from the raw NCEP–WRF downscale, and p′ is the resulting bias-corrected precipitation value. The transformation in Eq. (2) corrects the precipitation output from the NCEP–WRF model on a gridcell and monthly basis without significantly affecting the distribution of relative precipitation rates or extremes.
Second-stage correction involved correcting the ECHAM5–WRF temperature for the current and future periods. This was accomplished by estimating means, standard deviations, and time autocorrelations for all 24 hourly times of day for all 12 months of the year. Each run’s (ECHAM5–WRF current, ECHAM5–WRF future, and NCEP–WRF current) temperature data time series T, at individual grid cells, was transformed into correlated time series of normal deviates through Eq. (3):
e3
where i and j are the hour of the day and month of the year corresponding to time series index t. The means μij and standard deviations σij were computed using their standard definitions and by restricting the input data from the time series to those time steps matching the ith hour of the day and the jth month of the year. This transformation captures the seasonal and diurnal cycles into the parameter sets of μ and σ. The remainder is expected to contain long-wavelength intradecadal trends and oscillations, day-to-day meteorological-scale variability, and shorter-term stochastic variability.
The residual transformed temperature deviates series T′(t) are decomposed using hourly autocorrelation parameters into residual uncorrelated deviates X(t) according to Eq. (4):
e4
Parameter ρij, which is the hourly autocorrelation of the time series T′, is computed by using the standard product–moment calculation method with data restricted for ρij to each hour of the day i and month of the year j of the sequence. This parameterization represents the data in the form of an autoregressive model (Box et al. 2008). The means, standard deviations, and autocorrelations can be used in both a forward mode to produce a correlated, seasonally and diurnally varying temperature series from a set of uncorrelated deviates; or in the inverse mode, as described in Eqs. (3) and (4), to operate on a time series of temperature values and to recover the underlying uncorrelated deviate sequence.

The forward transforms are algebraically trivial inverses of Eqs. (3) and (4). For notational purposes, the transformation of deviate time series X into parameter time series V using dataset D (where D is one of “ref,” “cur,” or “fut” for the PRISM-corrected NCEP–WRF, ECHAM5–WRF current, and ECHAM5–WRF future runs, respectively) to compute the statistical parameters is V = RD(X), and the inverse transform is . For completion, the ECHAM5–WRF current-period Vcur is inverse transformed into Xcur using and then is forward transformed using the historical reference run Vcur,corrected = Rref(Xcur). In turn, the ECHAM/WRF future-period run Vfut is corrected by using the identical paired period transforms as in the current-period run: , and it has seasonal biases in the means, variances, and short-term autocorrelations removed. In addition to correcting the diurnal and seasonal biases in the distributional parameters and autocorrelations, this method also uses analytically smooth transforms so that stochastic irregularities in the NCEP–WRF reference and ECHAM5–WRF current datasets do not become part of the transforms and are not then imposed onto the future run.

Second-stage correction of precipitation required a different approach. Precipitation is not well represented by symmetric Gaussian distributions or by elementary transforms because of the high probability of zero precipitation. Instead, a direct nondistributional functional transform was chosen as the bias-correction mechanism (Kottek and Rubel 2007; Piani et al. 2010). A two-parameter model of the form was used to modify precipitation values to match both the accumulated precipitation and the convectivity C defined in Eq. (5):
e5

This approach ensures that bulk precipitation and convectivity are conserved in comparison with the PRISM-corrected NCEP–WRF downscale. The convectivity can be physically interpreted as the expectation of the rate at which the average mass unit of precipitation fell. This distinguishes short, intense rainfall events from long-duration, low-intensity events even if their sustained rates are equivalent. A numerical solver was used to identify values of aij and bij that allowed the transformed data to match the characteristics of the reference data.

A distinct advantage of this bias-correction method is that the hourly time-series nature of the datasets is preserved. Bias correction is performed conditional on all hours of the day within each calendar month. This is a more precise condition than generalizing on the daily extremes. Furthermore, the correction of the hourly autocorrelations ensures that persistent anomalous high- and low-temperature conditions are represented. Overall, the current-period ECHAM5–WRF downscale is corrected to match the seasonal and diurnal statistics of temperature as defined by the PRISM-corrected NCEP–WRF downscale; the corrected future ECHAM5–WRF downscale, identically corrected, presents an hourly time series that can be intercompared with the current-period ECHAM5–WRF downscale to produce products.

5. Assessment of regional climate change

To quantify regional climate change in the Southwest, we performed a detailed analysis of ECHAM5–WRF bias-corrected differences of temperature and precipitation and of the changes in snow cover with respect to the current period. Furthermore, we investigated the changes in the spatiotemporal patterns of temperature, precipitation, number of rainy days, and days with heavy precipitation, as well as the change of the temperature trends with elevation to identify the response of local climates (high-elevation alpine vs low-elevation semiarid).

a. Temperature and precipitation

Figure 6 shows the daily-average temperature and precipitation differences (2030s − 2000s) over the southwestern United States for each season. The temperature-difference pattern (Figs. 6a–d) changes significantly with the season. In the spring, the highest temperature differences are observed at higher elevations. This is possibly a result of earlier snowmelt and less accumulated snow in the future period. In the summer we see the largest temperature increases in California and Nevada and to a lesser extent in northern Arizona, western Utah, and central New Mexico. In the autumn, the temperature differences are higher in the eastern part of the domain—Colorado, New Mexico, and eastern Arizona. In the winter, the difference pattern changes to show higher values over southern Nevada, northern Arizona, and New Mexico. The precipitation differences (Figs. 6e–h) reveal overall decreases of precipitation in the region with the exception of the Sierra Nevada in the autumn. The precipitation decreases may partially explain the temperature differences. The consistently lower precipitation in spring and summer in northern California and Utah may result in increasing temperature differences in the summer. In the autumn the precipitation amounts are close to their climatological values and there are lower temperature differences in the region. The seasonal distribution of temperature and precipitation differences gives a baseline for assessment of regions that may possibly be affected by summer heat and water shortages.

Fig. 6.
Fig. 6.

ECHAM5–WRF-simulated differences (2030s − 2000s) in daily mean temperature: (a) spring, (b) summer, (c) autumn, and (d) winter and daily precipitation: (e) spring, (f) summer, (g) autumn, and (h) winter.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

Furthermore, we tested whether the observed differences in current versus future temperature changes are statistically significant and can be attributed to regional climate change as opposed to interannual variability. We calculated the standard errors of the differences of seasonal maximum and minimum temperatures between our current and future climate datasets and established that future temperatures were higher at the 95% confidence level over the majority of the experimental domain. The exception to this was the high plains in the extreme eastern portion of our domain where interannual variability was sufficiently large to statistically mask the period temperature differences in the means (Apling et al. 2010).

We also calculated the annual-mean temperature difference (future − current) and annual-mean precipitation change for each individual WRF grid point in New Mexico, Arizona, and Colorado (see Fig. 7). The precipitation change is defined in Eq. (6):
e6
where and are the average annual precipitation for the current and future periods, respectively.
Fig. 7.
Fig. 7.

Average temperature difference and precipitation change for (a) Colorado, (b) Arizona, and (c) New Mexico, and bivariate PDF of temperature difference and precipitation change for (d) Colorado, (e) Arizona, and (f) New Mexico.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

Depending on the sign of the temperature/precipitation differences, we defined four future climate regimes: 1) wetter/colder, 2) wetter/warmer, 3) dryer/colder, and 4) dryer/warmer. We found that the majority of the model grid points in Arizona and New Mexico are clustered in the lower-right quadrant, indicating a dryer and warmer future climate (Figs. 7b,c). While the majority of the model grid points in Colorado experience 0.8°–1.2°C increase in the average temperature, the precipitation trend is not as obvious (Fig. 7a). A larger number of points are clustered around the line with zero precipitation change. Our results agree with the overall trends identified in previous studies (i.e., Liang et al. 2006, 2008; Seager et al. 2007; Dominguez et al. 2010), which show widespread warming and decreasing precipitation in the American West.

We also calculated the bivariate PDF of temperature difference and precipitation change. The PDF provides the distribution of gridpoint values of temperature and precipitation for each state. Figures 7d–f show that the bivariate PDFs are multimodal for New Mexico, Arizona, and Colorado. To investigate whether the multimodal behavior is caused by the different response of climate regimes (high-elevation alpine vs low-elevation semiarid) we performed the same analysis for high-elevation points (elevation higher than 2000 m) only. We found that higher elevations experience larger changes in temperature and precipitation when compared with grid points with lower terrain elevation (not shown).

Temperature is a significant factor in defining the ecosystems and habitats at different elevations, and therefore any changes in the temperature trend with elevation may have an important implication for ecosystem dynamics. We investigated the temperature-change elevation trend by producing box-and-whisker plots for successive 250-m elevation bands. Figure 8 shows the median, minimum, and maximum values and the 25th and 75th percentiles of the annual-average temperature for the current period and future period, and their difference. Our results indicate that the largest changes in temperatures are observed for elevations between 1500 and 2750 m. This is in agreement with Diaz and Eischeid (2007), who found larger warming trends at higher elevations on the basis of PRISM temperature data for the 1979–2006 time period. Higher rates of temperature increase at higher elevations will likely result in reduced snow accumulation and decreased water availability during the warm season. Reservoir operations and planning are highly dependent on the timing and magnitude of snowpack melting, and these may change substantially in the future.

Fig. 8.
Fig. 8.

ECHAM5–WRF temperature change with elevation: (a) current (2000s) and (b) future (2030s), and the (c) difference (2030s − 2000s).

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

Variations in total precipitation can be caused by a change in the frequency of precipitation events, the intensity of precipitation per event, or a combination of both (Brunetti et al. 2001). To understand the precipitation behavior in the Southwest, we calculated the annual-average number of rainy days (accumulated rainfall in the grid cell in a 24-h period of >1 mm) and days with precipitation accumulations of 25.4 mm (1 in.) in a 24-h period for the current and future conditions (see Fig. 9). Our analysis shows that rainy days are decreasing in the future for the majority of the domain (Fig. 9a), except for eastern Colorado and southern Wyoming, where a slight increase is observed. The maximum decrease in rainy days is observed in the mountains of Arizona. The number of heavy-precipitation days is unchanged for most of the model domain. The Cascade Range and Sierra Nevada in California experience the largest decrease in heavy-precipitation days (Fig. 9b) followed by the California coastal region and the high plains in the eastern part of the model domain. The decreasing number of wet days will have a negative impact on the fragile semiarid ecosystems in the region as well as on water availability for irrigation and residential and commercial consumption.

Fig. 9.
Fig. 9.

ECHAM5–WRF differences (future − current) in (a) rainy days and (b) days with total precipitation exceeding 1 in.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

b. Snow accumulation

Snowpack is crucial to the water resources of the southwestern United States since their seasonal dynamics depend heavily on snowpack to store the wintertime precipitation for use in the drier summer months. An increasing body of literature (i.e., Knowles et al. 2006; Barnett et al. 2008) points to decreasing snow accumulation in the western states that will possibly result in increasing water shortages and lack of storage capability to meet seasonally changing river flow. While some studies suggest that snowpack trends may be partially attributable to interdecadal climate variability associated with the Pacific decadal oscillation and the El Niño–Southern Oscillation, they also appear to result from longer-term climate shifts (Guido 2008).

To assess the future changes in snowpack we calculated the ECHAM5–WRF annual-mean accumulated snow relative change (see Fig. 10) using
e7
where and are the average accumulated snow for the current and future periods.
Fig. 10.
Fig. 10.

Relative change in ECHAM5–WRF annual-mean accumulated snow.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

We computed the relative changes only for grid cells with terrain elevation higher than 1800 m. Low-elevation areas rarely experience snowfall, and calculating relative changes in such locations may lead to misleading results because of the low count of snowfall events. Figure 10 shows that eastern Colorado is the only region in the Southwest that shows even a slight increase in the annual-mean accumulated snow. The rest of the mountain ranges (i.e., the Sierra Nevada, the Sacramento Mountains, and the San Andres Mountains) experience a decrease in annual accumulated snow of between 5% and 30%, which may have important implications on seasonal dynamics of freshwater availability, hydropower generation, and snow-related recreational activities. Water planners can use these estimated changes in downscaled accumulated snow directly or in combination with simulated streamflows from hydrological models to assess the seasonal snowpack dynamics and timing of the snowmelt runoff.

6. Development of decision-aid products

There is a fundamental and pressing need for a systematic approach to providing the latest accepted results from research and analysis in climate change science to community leaders and the populations that will be impacted by climate change. Our objective is to translate the climate information provided by the WRF downscales into actionable information for policy and decision-making by developing various climate-adaptation decision aids. Here we show several examples of decision aids derived from our downscaled bias-corrected ECHAM5–WRF. Products are expressed in statistical terms to convey significance levels and/or standard errors for facilitating sound planning practices and effective risk management. Results from downscaled runs were communicated to planners in four military installations in the Southwest (Fort Carson, Peterson AFB, Fort Bliss, and Fort Irwin). They had general interest in the temperature, precipitation, and snowpack trends described in the previous section, since water and energy sustainability and environmental protection are key features of current military planning. We worked extensively with the public-works departments of Fort Carson and Peterson AFB to develop the energy-consumption products described in section 6a. We also developed and presented the heat-stress products described in section 6b, since all four installations have outdoor training programs that are carefully regulated relative to extreme temperatures.

a. Degree-days and energy consumption

Heating degree-days (HDD) and cooling degree-days (CDD) are quantitative indices often related to residential/commercial energy use for heating/cooling. Climatological values of HDD/CDD are also useful for strategic-planning purposes. HDD/CDD are calculated by taking the daily-average temperature excess of deficit relative to a base temperature; if it is colder/warmer than the base temperature, the difference is calculated. HDD and CDD are often included on residential and commercial energy bills and are commonly used by energy planners and oil/gas suppliers to assess daily and seasonal energy usage.

We used the downscaled bias-corrected ECHAM5–WRF temperatures to derive HDD/CDD and to estimate energy consumption scenarios in the 2030s for two military installations located near Colorado Springs—Fort Carson and Peterson AFB. Current energy-planning horizons of the two facilities are near term (1–5 years into the future). Their interest in long-term planning horizons was dictated by their anticipated infrastructure upgrades (heating, ventilation, and air-conditioning; lighting; and building upgrades) within the next 10–20-yr period and possible changes in building codes to accommodate climate change. The facilities already have spreadsheet methods to estimate future energy usage; the climatic inputs (HDD/CDD) are based on past climate conditions, however. In fact, planners are inclined to use past climate data as a surrogate for the future in their strategic planning rather than relying on future climate projections. Making decisions on the basis of historical climate data or by assuming a range of hydrological conditions similar to the past climate may result in inappropriate implementation of adaptation measures. The energy planners at the two military installations were interested in understanding the benefits of using dynamically downscaled data for future climate conditions and in including a climate change signal in their calculations.

First we calculated HDD/CDD in the Southwest to assess the changes in degree-days in the future. Figure 11 shows the mean annual difference (2030s − 2000s) in Fahrenheit-based cooling degree-days for a base temperature of 18.3°C. The 18.3°C value is a very commonly used base temperature in the U.S. energy industry (65°F). Our analysis indicates an increase/decrease of CDD/HDD over the Southwest in the future, which is a direct consequence of the increase of average temperature in the region. The highest increase in CDD is observed in the San Joaquin and Sacramento Valleys and parts of the Mojave and Sonoran Deserts. The highest decrease in HDD is observed in the mountain areas.

Fig. 11.
Fig. 11.

Mean annual bias-corrected difference (2030s − 2000s) in (a) CDD and (b) HDD.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

To estimate future energy consumption in the two military installations and to perform what-if energy-scenario analyses, we developed an advanced regression-analysis system and tools (Higgins et al. 2011). The form of the regression system is shown in Eq. (8):
e8
where E is the modeled energy consumption of either natural gas or electricity per square foot of facility space, t is the time coordinate expressed in years from beginning of record, and CTT and HTT are CDD and HDD, respectively, for given outdoor temperature thresholds. The coefficients a1, b1, b2, b3, c1, c2, and c3 are solved to fit end-user-supplied historically recorded energy-consumption levels. Large distributed facilities have buildings with a variety of heating and cooling responses to exterior temperatures; therefore, we included a number of degree-days indices for different temperature thresholds. The seasonal distribution of our ECHAM5–WRF-simulated bias-corrected HDD correlates well with the natural gas consumption data in both military installations, indicating that natural gas is used primarily for heating purposes (not shown). There is a less pronounced correlation between electrical consumption and CDD. The reason is that electricity is used for other activities besides cooling of the buildings during the summer months. In addition, air-conditioner use is not widespread at these installations.

Table 1 shows the annual-average electricity usage (in gigawatt hours) and natural gas usage (in million cubic feet of gas) for Fort Carson and Peterson AFB for the current period and the projected usage in the future. In addition, differences in usage (future − current) and standard error of the differences are also included in the table. Our results indicate a statistically significant increase in natural gas consumption and a possible decrease in electricity usage (assuming that all other engineering and infrastructure parameters remain unchanged), which is a direct consequence of warmer temperatures in the summer months (increasing cooling demand) and warmer winter temperatures (decreasing heating demand) in the future.

Table 1.

Current and projected annual energy usage for Peterson AFB and Fort Carson [annual mean, difference, and standard error (SE) of the difference]. Values are given for electricity in gigawatt hours and for natural gas in millions of cubic feet.

Table 1.

Our results were briefed to the facility managers of both installations. Our energy-usage modeling suggests a small increase in electricity use and a small decrease in natural gas use in the future. These figures are consistent with the installations’ current infrastructure and technology, however, and changes to these things, such as more widespread use of air-conditioning than at present, could have major implications for energy use in the future. It is significant that, for the first time in their planning practices, these facility planners considered predictions of energy usage that were based on downscaled data for future climate scenarios in addition to considering the historical climatological HDD/CDD information.

b. Heat-stress index

Exposure to extreme heat is a significant public-health problem and a major cause of weather-related mortality in the United States (Kinney et al. 2008). A variety of heat-stress indices are introduced to quantify the combined effect of high temperature and humidity on human physiology (e.g., Steadman 1984; Stull 2000). Heat-stress index between 32° and 42°C is related to sunstroke, heat cramps, or heat exhaustion, and heat stroke is possible with prolonged exposure and/or physical activity. Heat-stress indices are used operationally for day-to-day health and safety in military installations. Some military installations, located in hot regions, display a flag to indicate the heat category on the basis of the wet-bulb globe temperature, which takes into account surface temperature, humidity, wind, and radiant heat. Strict guidelines are then followed for water intake and physical-activity level for acclimated and unacclimated individuals on the basis of the heat index.

In this study we focused on the long-term change in the heat stress by estimating the change in days with heat stress above a certain threshold. We calculated the heat-stress index (HSI) using the Stull (2000) formulation
e9
where T and R are the hourly temperature and relative humidity and the 16 coefficients are a result of a multivariate fit to a model of the human body (Steadman 1979). The heat stress was calculated only for temperatures that are equal to or greater than 26.7°C (80°F) and for relative humidity that is equal to or greater than 40%.

Figure 12 shows the difference (future − current) in average number of days with heat-stress index exceeding 35°C in the Southwest. Our results indicate widespread increase in the number of heat-stress days with the exception of high-elevation mountain regions. The increase is particularly evident in the Sacramento and San Joaquin Valleys in California.

Fig. 12.
Fig. 12.

Difference (2030s − 2000s) in annual-mean days with heat index exceeding 35°C.

Citation: Journal of Applied Meteorology and Climatology 52, 2; 10.1175/JAMC-D-11-0192.1

Table 2 shows the annual-mean heat-stress days and their standard errors in the Colorado Springs/Fort Carson/Peterson AFB and El Paso (Texas)/Fort Bliss areas. Other major cities in the Southwest are also included in the table for comparison. Some areas show moderate change in oppressive-heat days (El Paso/Fort Bliss; Colorado Springs/Fort Carson/Peterson AFB) while others (Albuquerque, New Mexico; Reno, Nevada) show significant increase.

Table 2.

Annual-mean days with heat index above 95°F for major cities in the Southwest.

Table 2.

Our results were communicated to the planners at Fort Carson, Peterson AFB, Fort Bliss, and Fort Irwin. The calculated heat-stress index for current and future periods allowed the planners to grasp the full range of operational decisions they will need to address in the future. The range of indices expected in the future may have significant impact on long-term planning with respect to heat-stress control and heat-casualty management, allocation of training, and sustainment mission roles. These include which bases may best support various training missions, which bases will need to reserve lands currently assigned to mission uses for ecosystem preservation, and the shifting relative climatic advantages and disadvantages of various bases that will need be accounted for in the base realignment and closure process.

7. Conclusions

The Southwest has been experiencing steadily increasing temperatures over the last decade and is facing complex planning and policy issues associated with increasing water and energy demand as a result of warmer temperatures and continued rapid population growth and economic development. This study used a top-down, end-to-end approach to assess regional climate change in the Southwest in the 2030s and to derive decision aids in the areas of energy and health for four military installations in the region. Our workflow consisted of several components: dynamical downscaling, model validation and bias correction, assessment of regional climate change, development of custom impact models with specific infrastructure data provided by the end user, and user education and communication.

As Winkler et al. (2011a) pointed out, the development of local/regional climate change scenarios can be the “bottleneck” in the assessment process. Producing validated bias-corrected downscaled datasets on regional and local scales is critical for any assessment study of climate impacts, and such data are not readily available to planners in the Southwest. In this work, we showed that each component of the climate decision-aid development process is equally important in providing a practical answer in the engineering and statistical terms that are usually sought by the nonscientist end users. Another challenge that we met along the process was the multidisciplinary aspect of the decision-aid development. While downscaling/bias correction sits well within the realm of climate science, the impact-assessment component bridges expertise from civil and environmental engineering and social sciences. In this pilot study, we perceived ourselves as a mediator between the basic science research and the unmet needs of installation planners for practical engineering decision aids that are based on future climate projections. Major results and conclusions for each stage of our workflow are outlined below.

The dynamical downscaling component of the work was performed with the WRF model driven by the ECHAM5 GCM and the NCEP reanalysis data. The ECHAM5 model was selected for generating the future climate projection in the Southwest because ECHAM5 performed well both globally and regionally in the western United States in terms of temperature, precipitation, and storm-track statistics. In addition, the climate sensitivity in the ECHAM5 model is at the middle range among all IPCC AR4 models, which ensured that our dynamical downscale is not an outlier.

A rigorous validation of our regional climate modeling with gridded datasets and point observations was performed to identify biases in the downscaled data. The NCEP–WRF downscale showed a consistent warm bias over the high plains, New Mexico, and central Arizona and a cold bias over the Colorado Rocky Mountains, Wasatch Range, and Uinta Mountains in Utah throughout all seasons. ECHAM5–WRF had a cold bias throughout the year for the most of the southwestern United States except for the high plains and Sacramento and San Joaquin Valleys in California.

To correct the biases identified in the validation stage, we developed a novel two-stage bias correction. The reason for spending a significant effort on the bias-correction technique was dictated by the fact that end users are skeptical if modeled climate differs from current climate in their region of interest. The new method differs from the traditional bias-correction techniques used in climate studies by introducing a novel treatment of both time-scale and temporal autocorrelations that explicitly models the time scales as conditional parameterizations of the transfer functions and by introducing an intermediate data representation as an autoregressive model.

Our regional climate change assessment that is based on bias-corrected datasets points to a dryer and warmer future climate in the Southwest. The temperature differences between the current and future periods show distinct seasonal patterns. In the spring, the highest temperature differences are observed at higher elevations. This is possibly a result from earlier snowmelt and lower amount of accumulated snow in the future period. We also found that different climate regimes (alpine vs low-elevation semiarid) expose different behavior in terms of temperature and precipitation changes. Our analysis of the difference in the number of rainy days and days with heavy precipitation showed a decrease in wet days in the region and no significant change in the heavy-precipitation episodes, except for a slight decrease in the California mountains.

We used bias-corrected temperatures for current and future periods to derive quantities such as heating and cooling degree-days and oppressive-heat days that are commonly used for day-to-day operations in the military installations but that are currently based on ground measurements and/or past climatological data. Our energy-usage modeling produced a statistically significant increase in natural gas consumption and a possible decrease in electric usage in two military installations in Colorado, which is a direct consequence of decrease/increase in heating/cooling degree-days as a result of warmer temperatures in the future. Furthermore, we found an increase in the number of days with oppressive heat in the future, which may have important implications for human health and for training practices for the installations considered in this study.

Our results were communicated to the planners at Fort Carson and Peterson AFB in Colorado, Fort Bliss in Texas, and Fort Irwin in California. Possible implications to their long-term energy planning, various training missions, and lands currently assigned to mission uses were discussed. It is significant that, for the first time in their planning practices, these facility planners considered predictions that are based on downscaled data for future climate scenarios in addition to the historical climatological information that they currently use.

While our analysis confirms the broad scientific consensus of a dryer and warmer Southwest in the future, we also acknowledge its limitations: it is based solely on a single ECHAM5–WRF downscale and the simulated periods are relatively short from a climatological perspective. Despite the limitations, our 10-yr runs are the first step toward ensemble climate simulations performed with the WRF model that will enable us to bracket the uncertainties associated with different climate projections and to develop a variety of climate-adaptation decision aids. Our future plan is to perform ensemble WRF downscales driven with the upcoming Fifth IPCC Assessment Report ECHAM5 and the Community Climate System Model, version 4, GCM fields and with varying physical parameterizations. Our results suggest that bias-corrected decision aids derived from dynamical downscaling can provide valuable insights to decision makers to support and develop their strategic long-term planning.

Acknowledgments

This work was funded by the Northrop Grumman Information Systems Energy and Environment Initiative. The authors thank Dr. Robert Brammer for the valuable comments on the original manuscript. We also thank the facility managers at the four military installations described in this study for the valuable feedback on the decision-aid products and thank the three anonymous reviewers who provided helpful and constructive comments that substantially improved the quality of the manuscript.

REFERENCES

  • Ackerman, F., , and E. A. Stanton, 2011: The last drop: Climate change and the Southwest water crisis. Stockholm Environmental Institute Rep., 36 pp.

  • Apling, D., , G. Higgins, , H. Kiley, , and K. Darmenova, 2010: Analysis and quantification of uncertainty for climate change decision aids for energy consumption in the southwestern US. 2010 Fall Meeting, San Francisco, CA, Amer. Geophys. Union, Abstract NG23D-05.

  • Archer, S. R., , and K. I. Predick, 2008: Climate change and ecosystems of the southwestern United States. Rangelands, 30, 2328.

  • Barnett, T. P., and Coauthors, 2008: Human-induced changes in the hydrology of the western United States. Science, 319, 10801083.

  • Bengtsson, L., , K. Hodges, , and E. Roeckner, 2006: Storm tracks and climate change. J. Climate, 19, 35183543.

  • Box, G., , G. M. Jenkins, , and G. C. Reinel, 2008: Time Series Analysis: Forecasting and Control. 4th ed. Wiley Series in Probability and Statistics, John Wiley and Sons, 784 pp.

  • Brunetti, M., , M. Maugeri, , and T. Nanni, 2001: Changes in total precipitation, rainy days and extreme events in northeastern Italy. Int. J. Climatol., 21, 861871.

    • Search Google Scholar
    • Export Citation
  • Cayan, D., , T. Das, , D. Pierce, , T. Barnett, , M. Tyree, , and A. Gershunov, 2010: Future dryness in the southwest US and the hydrology of the early 21st century drought. Proc. Natl. Acad. Sci. USA, 107, 21 27121 276.

    • Search Google Scholar
    • Export Citation
  • Cohn, S. E., 1997: An introduction to estimation theory. J. Meteor. Soc. Japan, 75, 257288.

  • Daley, R., 1991: Atmospheric Data Analysis. Cambridge Atmospheric and Space Science Series, Vol. 2, Cambridge University Press, 457 pp.

  • Darmenova, K., , G. Higgins, , R. Alliss, , H. Kiley, , and D. Apling, 2010: Development of dynamical downscaling for climate decision aid applications. Extended Abstracts, 11th WRF Users’ Workshop, Boulder, CO, National Center for Atmospheric Research, 6 pp. [Available online at http://www.mmm.ucar.edu/wrf/users/workshops/WS2010/abstracts/P-27.pdf.]

  • Diaz, H. F., , and J. K. Eischeid, 2007: Disappearing “alpine tundra” Köppen climatic type in the western United States. Geophys. Res. Lett., 34, L18707, doi:10.1029/2007GL031253.

    • Search Google Scholar
    • Export Citation
  • DiLuzio, M., , G. L. Johnson, , C. Daly, , J. K. Eischeid, , and J. G. Arnold, 2008: Constructing retrospective gridded daily precipitation and temperature datasets for the conterminous United States. J. Appl. Meteor. Climatol., 47, 475497.

    • Search Google Scholar
    • Export Citation
  • Dominguez, F., , J. Cañon, , and J. Valdes, 2010: IPCC-AR4 climate simulations for the Southwestern US: The importance of future ENSO projections. Climatic Change, 99, 499514.

    • Search Google Scholar
    • Export Citation
  • Fisher, J., , and F. Ackerman, 2011: The water-energy nexus in the western states: Projections to 2100. Stockholm Environment Institute Rep., 39 pp. [Available online at http://sei-us.org/Publications_PDF/SEI-WesternWater-Energy-0211.pdf.]

  • Gu, H., , G. Wang, , Z. Yu, , and R. Mei, 2012: Assessing future climate changes and extreme indicators in east and south Asia using the RegCM4 regional climate model. Climatic Change, 114, 301317.

    • Search Google Scholar
    • Export Citation
  • Guido, Z., cited 2008: Climate and floods in the Southwest. Southwest Climate Change Network. [Available online at http://www.southwestclimatechange.org/impacts/water/floods.]

  • Gutzler, D. S., , and T. O. Robbins, 2010: Climate variability and projected change in the western United States: Regional downscaling and drought statistics. Climate Dyn., 37, 835849.

    • Search Google Scholar
    • Export Citation
  • Haerter, J. O., , S. Hagemann, , C. Moseley, , and C. Piani, 2011: Climate model bias correction and the role of timescales. Hydrol. Earth Syst. Sci., 15, 10651079.

    • Search Google Scholar
    • Export Citation
  • Hagemann, S., , K. Arpe, , and E. Roeckner, 2006: Evaluation of the hydrological cycle in the ECHAM5 model. J. Climate, 19, 38103827.

  • Higgins, G., , K. Darmenova, , D. Apling, , and H. Kiley, 2011: Regional climate modeling and energy demand projections for selected Colorado sites. Proc. Second Conf. on Weather, Climate, and the New Energy Economy, Seattle, WA, Amer. Meteor. Soc., 800. [Available online at https://ams.confex.com/ams/91Annual/webprogram/Paper182988.html.]

  • Jin, J., , S.-Y. Wang, , and R. Gillies, 2011: An improved dynamical downscaling for the western United States. Climate Change—Research and Technology for Adaptation and Mitigation, J. Blanco and H. Kheradmand, Eds., InTech, 23–38. [Available online at http://cdn.intechopen.com/pdfs/18711/InTech-An_improved_dynamical_downscaling_for_the_western_united_states.pdf.]

  • Jones, P. D., , and A. Moberg, 2003: Hemispheric and large-scale surface air temperature variations: An extensive revision and an update to 2001. J. Climate, 16, 206223.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437471.

  • Kinney, P., , M. O’Neill, , M. Bell, , and J. Schwartz, 2008: Approaches for estimating effects of climate change on heat-related deaths: Challenges and opportunities. Environ. Sci. Policy, 11, 8796.

    • Search Google Scholar
    • Export Citation
  • Knowles, N., , M. D. Dettinger, , and D. R. Cayan, 2006: Trends in snowfall versus rainfall in the western United States. J. Climate, 19, 45454559.

    • Search Google Scholar
    • Export Citation
  • Kottek, M., , and F. Rubel, 2007: Global daily precipitation fields from bias-corrected rain-gauge and satellite observations. Part I: Design and development. Meteor. Z., 16, 525539.

    • Search Google Scholar
    • Export Citation
  • Leung, L. R., , and Y. Qian, 2009: Atmospheric rivers induced heavy precipitation and flooding in the western U.S. simulated by the WRF regional climate model. Geophys. Res. Lett., 36, L03820, doi:10.1029/2008GL036445.

    • Search Google Scholar
    • Export Citation
  • Liang, X.-Z., , J. Pan, , J. Zhu, , K. E. Kunkel, , J. X. L. Wang, , and A. Dai, 2006: Regional climate model downscaling of the U.S. summer climate and future change. J. Geophys. Res., 111, D10108, doi:10.1029/2005JD006685.

    • Search Google Scholar
    • Export Citation
  • Liang, X.-Z., , K. Kunkel, , G. Meehl, , R. Jones, , and J. X. L. Wang, 2008: Regional climate models downscaling analysis of general circulation models present climate biases propagation into future change projections. Geophys. Res. Lett., 35, L08709, doi:10.1029/2007GL032849.

    • Search Google Scholar
    • Export Citation
  • MacDonald, G., 2010: Water, climate change, and sustainability in the Southwest. Proc. Natl. Acad. Sci. USA, 107, 21 25621 262.

  • Maurer, E. P., , and H. G. Hidalgo, 2008: Utility of daily vs. monthly large-scale climate data: An intercomparison of two statistical downscaling methods. Hydrol. Earth Syst. Sci., 12, 551563.

    • Search Google Scholar
    • Export Citation
  • Nakicenovic, N., , and R. Swart, Eds., 2000: Special Report on Emissions Scenarios. Cambridge University Press, 599 pp. [Available online at http://www.grida.no/climate/ipcc/emission/index.htm.]

  • Pan, L.-L., , S.-H. Chen, , D. Cayan, , M.-Y. Lin, , Q. Hart, , M.-H. Zhang, , Y. Liu, , and J. Wang, 2011: Influences of climate change on California and Nevada regions revealed by a high-resolution dynamical downscaling study. Climate Dyn., 37, 20052020.

    • Search Google Scholar
    • Export Citation
  • Piani, C., , J. O. Haerter, , and E. Coppola, 2010: Statistical bias correction for daily precipitation in regional climate models over Europe. Theor. Appl. Climatol., 99, 187192.

    • Search Google Scholar
    • Export Citation
  • Qian, Y., , S. J. Ghan, , and L. R. Leung, 2010: Downscaling hydroclimatic changes over the Western US based on CAM subgrid scheme and WRF regional climate simulations. Int. J. Climatol., 30, 675693.

    • Search Google Scholar
    • Export Citation
  • Ray, A. J., , J. J. Barsugli, , K. B. Averyt, , K. Wolter, , M. Hoerling, , N. Doesken, , B. Udall, , and R. S. Webb, 2008: Climate change in Colorado: A synthesis to support water resources management and adaptation. University of Colorado at Boulder Rep., 58 pp. [Available online at http://wwa.colorado.edu/publications/reports/WWA_ClimateChangeColoradoReport_2008.pdf.]

  • Robles, M. D., , and C. Enquist, 2010: Managing changing landscapes in the southwestern United States. Nature Conservancy Rep., Tucson, AZ, 26 pp. [Available online at http://azconservation.org/downloads/managing_changing_landscapes_in_the_southwestern_united_states.]

  • Roeckner, E., and Coauthors, 2003: The atmospheric general circulation model ECHAM5 Part 1. Rep. 349, Max Planck Institute for Meteorology, 140 pp.

  • Salathé, E. P., , R. Steed, , C. F. Mass, , and P. H. Zahn, 2008: A high-resolution climate model for the U.S. Pacific Northwest: Mesoscale feedbacks and local responses to climate change. J. Climate, 21, 57085726.

    • Search Google Scholar
    • Export Citation
  • Salathé, E. P., , L. R. Leung, , Y. Qian, , and Y. Zhang, 2010: Regional climate model projections for the State of Washington. Climatic Change, 102, 5175.

    • Search Google Scholar
    • Export Citation
  • Seager, R., , M. Ting, , and I. Held, 2007: Model projections of an imminent transition to a more arid climate in Southwestern North America. Science, 5828, 11811184.

    • Search Google Scholar
    • Export Citation
  • Skamarock, W. C., and Coauthors, 2008: A description of the Advanced Research WRF Version 3. NCAR Tech. Note NCAR/TN-475+STR, 125 pp. [Available online at http://www.mmm.ucar.edu/wrf/users/docs/arw_v3.pdf.]

  • Solomon, S., , D. Qin, , M. Manning, , Z. Chen, , M. Marquis, , K. B. Averyt, , M. Tignor, , and H. L. Miller, Eds., 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.

  • Steadman, R. G., 1979: The assessment of sultriness. Part I: A temperature-humidity index based on human physiology and clothing science. J. Appl. Meteor., 18, 861873.

    • Search Google Scholar
    • Export Citation
  • Steadman, R. G., 1984: A universal scale of apparent temperature. J. Climate Appl. Meteor., 23, 16741282.

  • Stull, R., 2000: Meteorology for Scientists and Engineers. 2nd ed. Brooks/Cole, 60 pp.

  • Swinbank, R., , and W. A. Lahoz, 2003: Data Assimilation for the Earth System. NATO Science Series: IV, Vol. 26, Kluwer, 388 pp.

  • Themeßl, M. J., , A. Gobiet, , and A. Leuprecht, 2011: Empirical-statistical downscaling and error correction of daily precipitation from regional climate models. Int. J. Climatol., 31, 15301544.

    • Search Google Scholar
    • Export Citation
  • Wi, S., , F. Dominguez, , M. Durcik, , J. Valdes, , H. Diaz, , and C. Castro, 2012: Climate change projection of snowfall in the Colorado River basin using dynamical downscaling. Water Resour. Res., 48, W05504, doi:10.1029/2011WR010674.

    • Search Google Scholar
    • Export Citation
  • Winkler, J., and Coauthors, 2011a: Climate scenario development and applications for local/regional climate change impact assessments: An overview for the non-climate scientist. Part I: Scenario development using downscaling methods. Geogr. Compass, 5, 275300.

    • Search Google Scholar
    • Export Citation
  • Winkler, J., , G. Guentchev, , M. Liszewska, , Perdinan, , and P.-N. Tan, 2011b: Climate scenario development and applications for local/regional climate change impact assessments: An overview for the non-climate scientist. Part II: Considerations when using climate change scenarios. Geogr. Compass, 5, 301328.

    • Search Google Scholar
    • Export Citation
Save