• Anderson, G. B., and M. L. Bell, 2011: Heat waves in the United States: Mortality risk during heat waves and effect modification by heat wave characteristics in 43 U.S. communities. Environ. Health Perspect., 119, 210218, doi:10.1289/ehp.1002313.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bolton, D., 1980: The computation of equivalent potential temperature. Mon. Wea. Rev., 108, 10461053, doi:10.1175/1520-0493(1980)108<1046:TCOEPT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bosilovich, M. G., 2013: Regional climate and variability of NASA MERRA and recent reanalyses: U.S. summertime precipitation and temperature. J. Appl. Meteor. Climatol., 52, 19391951, doi:10.1175/JAMC-D-12-0291.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bosilovich, M. G., and Coauthors, 2015: MERRA-2: Initial evaluation of the climate. NASA Tech. Memo. NASA/TM-2015-104606/Vol. 43, 145 pp., https://gmao.gsfc.nasa.gov/pubs/docs/Bosilovich803.pdf.

  • Brown, P. J., and A. T. DeGaetano, 2013: Trends in U.S. surface humidity, 1930–2010. J. Appl. Meteor. Climatol., 52, 147163, doi:10.1175/JAMC-D-12-035.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Collins, M., and Coauthors, 2013: Long-term climate change: Projections, commitments, and irreversibility. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 1029–1136.

  • Dai, A., 2006: Recent climatology, variability, and trends in global surface humidity. J. Climate, 19, 35893606, doi:10.1175/JCLI3816.1.

  • Daly, C., 2006: Guidelines for assessing the suitability of spatial climate data sets. Int. J. Climatol., 26, 707721, doi:10.1002/joc.1322.

  • Daly, C., M. Halbleib, J. I. Smith, W. P. Gibson, M. K. Doggett, G. H. Taylor, J. Curtis, and P. P. Pasteris, 2008: Physiographically sensitive mapping of climatological temperature and precipitation across the conterminous United States. Int. J. Climatol., 28, 20312064, doi:10.1002/joc.1688.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Davey, C. A., R. A. Pielke Sr., and K. P. Gallo, 2006: Differences between near-surface equivalent temperature and temperature trends for the eastern United States: Equivalent temperature as an alternative measure of heat content. Global Planet. Change, 54, 1932, doi:10.1016/j.gloplacha.2005.11.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Davis, L. W., and P. J. Gertler, 2015: Contribution of air conditioning adoption to future energy use under global warming. Proc. Natl. Acad. Sci. USA, 112, 59625967, doi:10.1073/pnas.1423558112.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, doi:10.1002/qj.828.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fall, S., N. S. Diffenbaugh, D. Niyogi, R. A. Pielke Sr., and G. Rochon, 2010: Temperature and equivalent temperature over the United States (1979–2005). Int. J. Climatol., 30, 20452054, doi:10.1002/joc.2094.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ford, T. W., and J. T. Schoof, 2016: Oppressive heat events in Illinois related to antecedent wet soils. J. Hydrometeor., 17, 27132726, doi:10.1175/JHM-D-16-0075.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ford, T. W., and J. T. Schoof, 2017: Characterizing extreme and oppressive heat waves in Illinois. J. Geophys. Res. Atmos., 122, 682298, doi:10.1002/2016JD025721.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gaffen, D. J., and R. J. Ross, 1998: Increased summertime heat stress in the U.S. Nature, 396, 529530, doi:10.1038/25030.

  • Gaffen, D. J., and R. J. Ross, 1999: Climatology and trends in U.S. surface humidity and temperature. J. Climate, 12, 811828, doi:10.1175/1520-0442(1999)012<0811:CATOUS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grotjahn, R., and Coauthors, 2016: North American extreme temperature events and related large scale meteorological patterns: A review of statistical methods, dynamics, modeling, and trends. Climate Dyn., 46, 11511184, doi:10.1007/s00382-015-2638-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grundstein, A., and J. Dowd, 2011: Trends in extreme apparent temperatures over the United States, 1949–2010. J. Appl. Meteor. Climatol., 50, 16501653, doi:10.1175/JAMC-D-11-063.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Held, I. M., and B. J. Soden, 2006: Robust response of the hydrologic cycle to global warming. J. Climate, 19, 56865699, doi:10.1175/JCLI3990.1.

  • Hoerling, M., and Coauthors, 2013: Anatomy of an extreme event. J. Climate, 26, 28112832, doi:10.1175/JCLI-D-12-00270.1.

  • Hondula, D. M., R. C. Balling Jr., J. K. Vanos, and M. Georgescu, 2015: Rising temperatures, human health, and the role of adaptation. Curr. Climate Change Rep., 1, 144154, doi:10.1007/s40641-015-0016-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Johnstone, J. A., and N. J. Mantua, 2014: Atmospheric controls on northeast Pacific temperature variability and change, 1900–2012. Proc. Natl. Acad. Sci. USA, 111, 14 36014 365, doi:10.1073/pnas.1318371111.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kobayashi, S., and Coauthors, 2015: The JRA-55 reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 548, doi:10.2151/jmsj.2015-001.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kravchenko, J., A. P. Abernethy, M. Fawzy, and H. K. Lyerly, 2013: Minimization of heatwave morbidity and mortality. Amer. J. Prev. Med., 44, 274282, doi:10.1016/j.amepre.2012.11.015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kunkel, K. E., S. A. Changnon, B. C. Reinke, and R. W. Arritt, 1996: The July 1995 heat wave in the Midwest: A climatic perspective and critical weather factors. Bull. Amer. Meteor. Soc., 77, 15071518, doi:10.1175/1520-0477(1996)077<1507:TJHWIT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lanzante, J. R., 1996: Resistant, robust and non-parametric techniques for the analysis of climate data: Theory and examples, including applications to historical radiosonde station data. Int. J. Climatol., 16, 11971226, doi:10.1002/(SICI)1097-0088(199611)16:11<1197::AID-JOC89>3.0.CO;2-L.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lobell, D. B., M. J. Roberts, W. Schlenker, N. Braun, B. B. Little, R. M. Rejesus, and G. L. Hammer, 2014: Greater sensitivity to drought accompanies maize yield increase in the U.S. Midwest. Science, 344, 516519, doi:10.1126/science.1251423.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • MacDonald, G. M., 2010: Water, climate change, and sustainability in the Southwest. Proc. Natl. Acad. Sci. USA, 107, 21 25621 262, doi:10.1073/pnas.0909651107.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Meehl, G. A., J. M. Arblaster, and C. T. Y. Chung, 2015: Disappearance of the southeast U.S. “warming hole” with the late 1990s transition of the Interdecadal Pacific Oscillation. Geophys. Res. Lett., 42, 55645570, doi:10.1002/2015GL064586.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Melillo, J. M., T. C. Richmond, and G. W. Yohe, Eds., 2014: Climate Change Impacts in the United States: The Third National Climate Assessment. U.S. Global Change Research Program, 841 pp., doi:10.7930/J0Z31WJ2.

    • Crossref
    • Export Citation
  • Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343360, doi:10.1175/BAMS-87-3-343.

  • Metzger, K. B., K. Ito, and T. D. Matte, 2010: Summer heat and mortality in New York City: How hot is too hot? Environ. Health Perspect., 118, 8086, doi:10.1289/ehp.0900906.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Miller, N. L., K. Hayhoe, J. Jin, and M. Auffhammer, 2008: Climate, extreme heat, and electricity demand in California. J. Appl. Meteor. Climatol., 47, 18341844, doi:10.1175/2007JAMC1480.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mishra, V., and K. A. Cherkauer, 2010: Retrospective droughts in the crop growing season: implications to corn and soybean yield in the Midwestern United States. Agric. For. Meteor., 150, 10301045, doi:10.1016/j.agrformet.2010.04.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • NWS, 2016: 76-year list of severe weather fatalities. NOAA, 2 pp., http://www.nws.noaa.gov/om/hazstats/resources/weather_fatalities.pdf.

  • Pan, Z., R. W. Arritt, E. S. Takle, W. J. Gutowski, C. J. Anderson, and M. Segal, 2004: Altered hydrologic feedback in a warming climate introduces a “warming hole.” Geophys. Res. Lett., 31, L17109, doi:10.1029/2004GL020528.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., Sr., C. Davey, and J. Morgan, 2004: Assessing “global warming” with surface heat content. Eos, Trans. Amer. Geophys. Union, 85, 210211, doi:10.1029/2004EO210004.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pryor, S. C., and J. T. Schoof, 2016: Evaluation of near-surface temperature, humidity and equivalent temperature from regional climate models in type-II downscaling. J. Geophys. Res. Atmos., 121, 33263338, doi:10.1002/2015JD024539.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pryor, S. C., R. C. Sullivan, and T. Wright, 2016: Quantifying the roles of changing albedo, emissivity, and energy partitioning in the impact of irrigation on atmospheric heat content. J. Appl. Meteor. Climatol., 55, 16991706, doi:10.1175/JAMC-D-15-0291.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schoof, J. T., Z. A. Heern, M. D. Therrell, and J. W. F. Remo, 2015: Assessing trends in lower tropospheric heat content in the central United States using equivalent temperature. Int. J. Climatol., 35, 28282836, doi:10.1002/joc.4175.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Seidel, D. J., Q. Fu, W. J. Randel, and T. J. Reichler, 2008: Widening of the tropical belt in a changing climate. Nat. Geosci., 1, 2124, doi:10.1038/ngeo.2007.38.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sen, P. K., 1968: Estimates of the regression coefficient based on Kendall’s tau. J. Amer. Stat. Assoc., 63, 13791389, doi:10.1080/01621459.1968.10480934.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Seneviratne, S. I., M. G. Donat, B. Mueller, and L. V. Alexander, 2014: No pause in the increase of hot temperature extremes. Nat. Climate Change, 4, 161163, doi:10.1038/nclimate2145.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Souch, C., and C. S. B. Grimmond, 2004: Applied climatology: ‘Heat waves.’ Prog. Phys. Geogr., 28, 599606, doi:10.1191/0309133304pp428pr.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Steadman, R. G., 1984: A universal scale of apparent temperature. J. Climate Appl. Meteor., 23, 16741687, doi:10.1175/1520-0450(1984)023<1674:AUSOAT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res., 106, 71837192, doi:10.1029/2000JD900719.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., and G. W. Branstator, 1992: Issues in establishing causes of the 1998 drought over North America. J. Climate, 5, 159172, doi:10.1175/1520-0442(1992)005<0159:IIECOT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vose, R. S., S. Applequist, M. J. Menne, C. N. Williams Jr., and P. Thorne, 2012: An intercomparison of temperature trends in the U.S. Historical Climatology Network and recent atmospheric reanalyses. Geophys. Res. Lett., 39, L10703, doi:10.1029/2012GL051387.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, S., and Coauthors, 2014: Attribution of the United States “warming hole”: Aerosol indirect effect and precipitable water vapor. Sci. Rep., 4, 6929, doi:10.1038/srep06929.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • View in gallery

    Map of study area showing regions: Northeast, Southeast, Midwest, northern plains, southern plains, Southwest, and Northwest.

  • View in gallery

    (top) Maps, (middle) Taylor diagrams, and (bottom) regional averages of (a),(e),(i) JJA mean (°C), (b),(f),(j) standard deviation (°C), (c),(g),(k) interannual variability (°C), and (d),(h),(l) 35-yr linear trend (°C decade−1) of 2-m air temperature. The maps show the results from the PRISM dataset. The Taylor diagrams show the correlation between the PRISM field and each reanalysis field (on the radial axis), the spatial standard deviation (on the x and y axes), and the pattern-centered RMSE (on the circular axis). The datasets shown are PRISM (P), ERA-Interim (E), JRA-55 (J), MERRA-2 (M), and NARR (N).

  • View in gallery

    As in Fig. 2, but for 2-m q (g kg−1).

  • View in gallery

    As in Fig. 2, but for 2-m TE (°C).

  • View in gallery

    Mean annual HWD frequency and magnitude by region. Results are shown for the PRISM dataset as well as the ERA-Interim, JRA-55, MERRA-2, and NARR reanalysis products. The frequency of (a) T-only and (b) T-and-TE HWDs are shown. (c) The mean magnitude (see section 3b) of T on T-only HWDs. The mean magnitudes of (d) T and (e) TE on T-and-TE HWDs are shown.

  • View in gallery

    Time series of summer (JJA) HWD frequency (days yr−1) for (a)–(f) U.S. regions (as in Fig. 1) based on the PRISM dataset. For each subplot, time series are presented for T-only HWDs (circles with accompanying solid lines) and T-and-TE HWDs (squares with dotted lines). Time series with significant linear trends (with α = 0.10; see section 3) are indicated with filled symbols and an accompanying trend line. Numerical values of the trends are provided in Table 2.

  • View in gallery

    Scatter diagrams exhibiting the relationships between the Spearman correlation of daily T and q and (a) the frequency of T-and-TE HWDs and (b) the mean TE magnitude (as defined in section 3) on T-and-TE HWDs.

  • View in gallery

    Time series of mean regional proportion impacted by heat wave conditions (proportion per year) for (a)–(g) U.S. regions (as in Fig. 1) based on the PRISM dataset. For each subplot, time series are presented for T-only HWDs (circles with accompanying solid lines) and T-and-TE HWDs (squares with accompanying dotted lines). Time series with significant linear trends (with α = 0.10) are indicated with filled symbols and an added trend line. Numerical values of the trends are provided in Table 3.

  • View in gallery

    Time series of mean heat wave magnitude for U.S. regions (as in Fig. 1) based on the PRISM dataset. For each subplot, time series are presented for T on T-only HWDs (circles with accompanying solid lines) and T-and-TE HWDs (squares with accompanying dotted lines) and TE on T-and-TE HWDs (diamonds accompanying dotted lines). Time series with significant linear trends (with α = 0.10) are indicated with filled symbols and an added trend line. Numerical values of the trends are provided in Table 4.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 393 218 29
PDF Downloads 346 191 47

Recent Changes in U.S. Regional Heat Wave Characteristics in Observations and Reanalyses

View More View Less
  • 1 Department of Geography and Environmental Resources, Southern Illinois University, Carbondale, Illinois
  • | 2 Department of Earth and Atmospheric Sciences, Cornell University, Ithaca, New York
Full access

Abstract

Humidity is a key determinant of heat wave impacts, but studies investigating changes in extreme heat events have not differentiated between events characterized by high temperatures and those characterized by simultaneously elevated temperature and humidity. The authors present a framework, using air temperature (T) and equivalent temperature (TE; a measure combining temperature and specific humidity), to examine changes in local percentile-based extreme heat events characterized by high temperature (T only) and those with high temperature and humidity (T-and-TE events). Application to one observational dataset (PRISM), four reanalysis products (1981–2015), and seven U.S. regions reveals widespread changes in heat wave characteristics over the 35-yr period. Agreement among the datasets employed on several heat wave metrics suggests that many of the findings are robust. With the exception of the northern plains region, all regions experienced increases in both T-only and T-and-TE heat wave day (HWD) frequency in each of the reanalyses. In the northern plains, all datasets have negative trends in T-only HWD frequency and positive trends in T-and-TE HWD frequency. Trends in HWD frequency were generally accompanied by changes in the spatial footprint in heat wave conditions. Temperature has increased significantly during T-only HWDs in the western regions, while increases in TE during T-and-TE HWDs have occurred in the central United States and Northeast region. These findings suggest that equivalent temperature provides an alternative perspective on the evolution of regional heat wave climatology. Studies considering changes in regional heat wave impacts should carefully consider the role of atmospheric moisture.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: J. T. Schoof, jschoof@siu.edu

Abstract

Humidity is a key determinant of heat wave impacts, but studies investigating changes in extreme heat events have not differentiated between events characterized by high temperatures and those characterized by simultaneously elevated temperature and humidity. The authors present a framework, using air temperature (T) and equivalent temperature (TE; a measure combining temperature and specific humidity), to examine changes in local percentile-based extreme heat events characterized by high temperature (T only) and those with high temperature and humidity (T-and-TE events). Application to one observational dataset (PRISM), four reanalysis products (1981–2015), and seven U.S. regions reveals widespread changes in heat wave characteristics over the 35-yr period. Agreement among the datasets employed on several heat wave metrics suggests that many of the findings are robust. With the exception of the northern plains region, all regions experienced increases in both T-only and T-and-TE heat wave day (HWD) frequency in each of the reanalyses. In the northern plains, all datasets have negative trends in T-only HWD frequency and positive trends in T-and-TE HWD frequency. Trends in HWD frequency were generally accompanied by changes in the spatial footprint in heat wave conditions. Temperature has increased significantly during T-only HWDs in the western regions, while increases in TE during T-and-TE HWDs have occurred in the central United States and Northeast region. These findings suggest that equivalent temperature provides an alternative perspective on the evolution of regional heat wave climatology. Studies considering changes in regional heat wave impacts should carefully consider the role of atmospheric moisture.

© 2017 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: J. T. Schoof, jschoof@siu.edu

1. Introduction

Extreme heat is associated with impacts across a range of sectors, including agriculture (Lobell et al. 2014), energy use (Miller et al. 2008), and human health (Anderson and Bell 2011). Despite the presence of air conditioning in more than 90% of U.S. homes (Davis and Gertler 2015), in most years extreme heat remains the largest cause of weather-related mortality in the United States (NWS 2016). Near-surface humidity is a key determinant of impacts associated with extreme temperatures, with high humidity heat waves leading to elevated morbidity and mortality in people (Kravchenko et al. 2013) and prolonged periods of high temperatures and low humidity associated with precipitation deficits (i.e., meteorological droughts) leading to reduced agricultural yields (Mishra and Cherkauer 2010). Given historical increases in temperature extremes (Seneviratne et al. 2014) and specific humidity (Dai 2006) and robust projections of future temperature increases (Collins et al. 2013) and associated changes in column-integrated water vapor (Held and Soden 2006), there is a need to better understand changes in the moisture content of heat waves.

The combined effects of temperature and humidity have been considered in several previous studies through application of apparent temperature (Ta; Steadman 1984), most commonly using the simplified formula that ignores the effects of wind and radiation:
e1
where T is the air temperature (°C) and e is the vapor pressure (kPa). Gaffen and Ross (1998) found a significant increase in Ta heat waves during 1949–95 with the strongest signal in the western United States. More recently, Grundstein and Dowd (2011) reported acceleration of those trends and also noted an increase in the frequency of days with extreme values of Ta. While the utility of Ta for studying heat-related morbidity and mortality has been demonstrated (e.g., Metzger et al. 2010), high values of Ta are possible in the absence of high humidity, given sufficiently high air temperature. For the purpose of differentiating between high and low humidity extreme events, we adopt equivalent temperature (TE; °C), which is given by
e2
where T is the air temperature (°C), Lυ is the latent heat of vaporization (J kg−1), q is the specific humidity (kg kg−1), and Cp is the specific heat of air at constant pressure (1005 J kg °C−1). Relative to Ta, TE is much more responsive to changes in humidity (Pielke et al. 2004). For example, at 35°C, a 1°C increase in the dewpoint temperature from 19° to 20°C raises Ta by 0.31°C, but it raises TE by 2.15°C. The upper tail of the TE distribution is therefore composed only of events characterized by high temperature and high humidity. Previous analyses have identified larger variability and stronger trends in TE, relative to T, particularly during summer as a result of increasing specific humidity (e.g., Davey et al. 2006; Fall et al. 2010; Schoof et al. 2015). Conceptually, variations in TE might result from 1) global or hemispheric changes in specific humidity accompanying large-scale warming, 2) changes in circulation that govern regional moisture transport (Ford and Schoof 2016), and/or 3) changes in local forcing related to local land cover, soil moisture variations (including those deriving from irrigation; Pryor et al. 2016), and land–atmosphere interaction (Ford and Schoof 2017; Pryor and Schoof 2016).

The TE is thus ideal for differentiating between T-only heat wave conditions (T is elevated, but TE is not) and T-and-TE heat wave conditions (both T and TE are elevated) in climatic time series. The former represent hot and dry extremes while the latter represent hot and moist extremes. Here, we use this framework to characterize changes in heat wave day characteristics for U.S. regions (Fig. 1). We adopt the regional delineation used by the recent U.S. National Climate Assessment (Melillo et al. 2014) but follow Bosilovich (2013) and further subdivide the Great Plains region into southern and northern regions to account for differences in hydroclimatology across the central United States. Each of the regions is also characterized by subregional variability resulting from variations in topography, land cover, and other factors. Here, we are primarily interested in broad regional changes, but where appropriate we note instances where changes are subregional in nature. Similar regional definitions have been adopted in previous studies of regional temperature and humidity (e.g., Gaffen and Ross 1999; Brown and DeGaetano 2013). Our investigation is based on results derived from high-resolution quality-controlled observational reference data (PRISM; Daly et al. 2008) and retrospective analysis (reanalysis) data. The data are described in section 2. In section 3, we describe the metrics used in our trend analysis. The results are presented in section 4 and summarized in section 5.

Fig. 1.
Fig. 1.

Map of study area showing regions: Northeast, Southeast, Midwest, northern plains, southern plains, Southwest, and Northwest.

Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0393.1

2. Data

Air temperature and humidity time series from surface stations suffer from inhomogeneities associated with station moves, sensor changes, urbanization, and other factors. These changes sometimes require time series adjustments, such as those applied in Brown and DeGaetano (2013) before temporal trends can be assessed. Given our focus on regional analyses, and the strong effects of local land cover (Fall et al. 2010) and irrigation (Pryor et al. 2016) on temperature and humidity, we opt to use 1) the PRISM gridded temperature and dewpoint temperature products (Daly et al. 2008), which are highly quality controlled and spatially and serially complete, and 2) reanalysis data, which integrate observational data and numerical models to produce temporally and spatially consistent outputs (Bosilovich 2013).

a. Reanalysis data

Four reanalysis products (Table 1) are considered in this study: the European Centre for Medium-Range Weather Forecasts (ECMWF) interim reanalysis (ERA-Interim; Dee et al. 2011), the Japan Meteorological Agency (JMA) 55-year Reanalysis (JRA-55; Kobayashi et al. 2015), the National Aeronautics and Space Administration (NASA) Modern-Era Retrospective Analysis for Research and Applications 2 (MERRA-2; Bosilovich et al. 2015), and the National Centers for Environmental Prediction (NCEP) North American Regional Reanalysis (NARR; Mesinger et al. 2006). These reanalyses were chosen because they span a range of spatial resolutions (32 km for NARR; up to 1.25° for JRA-55), are continuously updated, and vary in terms of assimilated variables. Both ERA-Interim and JRA-55 assimilate near-surface temperature and humidity data from surface stations. MERRA-2 does not assimilate 2-m temperature or humidity over land (M. G. Bosilovich 2016, personal communication), while NARR assimilates near-surface humidity, but not temperature. In an assessment of several reanalyses, including ERA-Interim, NARR, and earlier versions of JMA (JRA-25) and NASA (MERRA) reanalyses, Vose et al. (2012) found that the reanalysis-derived temperature trends for the contiguous United States (1979–2008) bracket those derived from an adjusted station network (USHCN). Bosilovich (2013) compared summer temperature and precipitation variations in several reanalyses with observations and found that variability and trends in temperature were more similar than those for precipitation. In section 3, we assess multiple aspects of reanalysis performance relative to gridded, quality-controlled observations.

Table 1.

General characteristics of datasets used in this study.

Table 1.

b. Observational data

Prior to using reanalysis data to estimate changes in the heat wave characteristics, we first assess the spatiotemporal variability of reanalysis-derived 2-m temperature, specific humidity, and equivalent temperature series with those from PRISM. Here, we use daily temperature and dewpoint temperature from PRISM with a resolution of 2.5 arc min (approximately 4 km). PRISM, described more fully in Daly et al. (2008), combines station observations from multiple networks with a digital elevation model to provide gridded data for the contiguous United States. It has been widely used in applied climatological settings.

3. Methods

The ERA-Interim and JRA-55 output is available at 6-h increments, while NARR data are available at 3-h increments and MERRA-2 data are available at hourly resolution. For consistency, we sampled all products every 6 h at 0000, 0600, 1200, and 1800 UTC and then computed daily averages. The JRA-55, MERRA-2, and NARR products included 2-m temperature and specific humidity in their output. The ERA-Interim output does not include near-surface specific humidity, so it was calculated from 2-m dewpoint temperature and surface pressure using
e3a
e3b
where Td is the dewpoint temperature (°C), e is the vapor pressure (hPa), P is the surface pressure (hPa), and q is the specific humidity (kg kg−1) (Bolton 1980). The observational data (PRISM) also consist of 2-m temperature and 2-m dewpoint temperature. We therefore used the pressure field from the highest-resolution reanalysis product (NARR) to compute q, again using Eqs. (3a) and (3b). Since the observed pressure field is assimilated by all the reanalyses, the resulting q field is not sensitive to the choice of NARR for this purpose.

a. Assessment of reanalyses relative to PRISM and each other

To assess the spatiotemporal variability of reanalysis-derived 2-m temperature (T), specific humidity (q), and equivalent temperature (TE) relative to PRISM, we computed 1) the mean and standard deviation of the summer (JJA) daily values, 2) the interannual variability of the JJA means, defined as the standard deviation of the 1981–2015 (n = 35) values, and 3) the linear trend of the JJA means. Because the reanalyses differ in terms of spatial resolution (section 2a), each metric was computed at the native reanalysis resolution and then interpolated to the PRISM grid covering the contiguous United States using bilinear interpolation. For each metric, we develop a Taylor diagram (Taylor 2001) that displays the pattern correlation between PRISM and each reanalysis product, the standard deviation of the spatial fields from PRISM and each reanalysis product, and the pattern-centered root-mean-square error between PRISM and each reanalysis product. Pattern-centered statistics, such as those employed in Taylor diagrams, are designed to compare spatial patterns but do not aid in identification of regional biases. We therefore supplement our analysis of patterns with regional averages of each metric. This combination allows us to simultaneously assess the reanalyses in terms of their spatial patterns and regional biases with respect to multiple descriptors of T, q, and TE.

b. Heat wave definitions

Although heat waves are commonly understood to be time periods with near-surface air temperatures that are well above average, there is no universally accepted definition of a heat wave (Souch and Grimmond 2004). Given the clear evidence for acclimatization to heat extremes (Hondula et al. 2015), we use locally derived percentile-based thresholds rather than absolute thresholds. Specifically, we use a 15-day moving window centered on each JJA calendar day to define a 90th percentile daily values for T and TE, using data from 1981 to 2015. A day is a heat wave day if both the daily temperature and the three-day average temperature (i.e., the mean of the current day and the previous two days) exceed the 90th percentile daily value. If the current day has a value of TE that falls below the 90th percentile TE value, the current day is considered to be a T-only heat wave day. Conversely, if the current day has a value of TE that exceeds the 90th percentile TE value, the current day is considered to be a T-and-TE heat wave day.

Using the definitions above for T-only and T-and-TE heat wave days, we compute several metrics of heat wave frequency and magnitude for each summer between 1981 and 2015. To characterize heat wave day frequency, we consider the mean (over each region) number of JJA heat wave days (HWDs) for T-only and T-and-TE type heat waves. Assessing changes in heat wave day magnitude (i.e., intensity) requires a different approach since extreme heat events have both a spatial footprint and a magnitude. To address this issue, for each day (n = 3220; 92 days per summer × 35 summers) we compute the spatial proportion [0, 1] of each region impacted by each type of heat wave event as well as the event magnitude, defined as the average departure from the heat wave day threshold (90th percentile as defined above) among grid points experiencing heat wave conditions. We then consider the mean and maximum values of the spatial proportion and mean magnitude for each summer.

To synthesize our results, we compute each heat wave day characteristic at the native data resolution (see section 2) but report values for the regions in Fig. 1. Because of short time series and high interannual variability in heat wave frequency, trend analysis was conducted on each of the heat wave day metrics using the nonparametric Theil–Sen estimator (Sen 1968), also known as median-of-pairwise-slopes regression (Lanzante 1996).

4. Results

a. Assessment of reanalysis products

Each of the reanalysis products considered in this study reproduces the mean and standard deviation of 2-m T, q, and TE (Figs. 24). Comparison of the maps of mean T and mean TE (cf. Figs. 2a and 4a) highlights important differences between these variables. Air temperatures are highest in the desert Southwest (SW) and throughout the southern plains (SP) and Southeast (SE) regions (Fig. 2a), while TE is highest in the warm and humid climate of the SE region as a result of the influence of near-surface humidity, which exhibits a strong contrast between the humid SE region and the dry western regions (Fig. 3a). Variability of daily T is negatively correlated with the mean over space, while the variability of daily q is highest in the monsoon region of the SW (Fig. 3b). Variability of daily TE (Fig. 4b) exhibits a pattern more like that of q than T. Correlations between PRISM- and reanalysis-derived fields of the mean and standard deviation of daily T, q, and TE generally exceed 0.9 (Figs. 2a,b, 3a,b, and 4a,b), indicating strong agreement among observations and reanalyses for these basic climate descriptors. While the regional means derived from the reanalysis generally bracket the estimate derived from PRISM, each of the reanalyses is slightly warmer on average than PRISM in the NP region (Fig. 2i) and NARR exhibits a warm bias across most regions relative to PRISM and the other reanalyses. The reanalyses are also more variable than PRISM for T, q, and TE across most regions (Figs. 2j, 3j, and 4j).

Fig. 2.
Fig. 2.

(top) Maps, (middle) Taylor diagrams, and (bottom) regional averages of (a),(e),(i) JJA mean (°C), (b),(f),(j) standard deviation (°C), (c),(g),(k) interannual variability (°C), and (d),(h),(l) 35-yr linear trend (°C decade−1) of 2-m air temperature. The maps show the results from the PRISM dataset. The Taylor diagrams show the correlation between the PRISM field and each reanalysis field (on the radial axis), the spatial standard deviation (on the x and y axes), and the pattern-centered RMSE (on the circular axis). The datasets shown are PRISM (P), ERA-Interim (E), JRA-55 (J), MERRA-2 (M), and NARR (N).

Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0393.1

Fig. 3.
Fig. 3.

As in Fig. 2, but for 2-m q (g kg−1).

Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0393.1

Fig. 4.
Fig. 4.

As in Fig. 2, but for 2-m TE (°C).

Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0393.1

Since heat waves exhibit substantial variability from one summer to the next, assessment of interannual T, q, and TE variability in observations and reanalysis is an important step toward understanding differences in their heat wave climatologies. Correlations between PRISM and reanalysis-derived fields of interannual variability range from 0.74 to 0.84, 0.65 to 0.75, and 0.85 to 0.95 for T, q, and TE, respectively (Figs. 2g, 3g, 4g). The reanalyses therefore agree with PRISM with respect to the spatial pattern of interannual variability. Examination of regional values of interannual temperature variability indicates that MERRA-2 is biased high relative to PRISM and other reanalyses across the eastern and central United States (Fig. 2k). MERRA-2 and NARR also exhibit some regional biases in interannual variability of q (Fig. 3k).

PRISM and reanalysis-derived trends of T, q, and TE show qualitative similarity, but some notable regional differences (Figs. 24). Apart from the SE region, the T trends vary substantially within U.S. regions. For example, NARR shows much less warming than PRISM or the other reanalyses in the western United States (Fig. 2l) and all of the reanalyses exhibit more warming than PRISM in the SE, Midwest (MW), and SP regions. Despite the small-scale cooling across parts of the northern plains (NP) and MW regions (which is reproduced to varying degrees in the reanalyses), the 35-yr linear trends in temperature are positive across all regions and datasets. Temperature trends derived from PRISM are significant (α = 0.10) in the Northeast (NE), SE, SP, SW, and Northwest (NW) regions. All of the reanalysis-derived temperature trends are significant in the SE, SP, and SW regions and none of the reanalyses indicate significant warming the MW or NP. The warming pattern in NARR (warm faster than other data products in the east and slower than other data products in the west) results in a low pattern correlation between PRISM and NARR-derived temperature trends.

Unlike the temperature trends presented in Fig. 2, the humidity trends (Fig. 3d) differ in sign among regions and in some cases among reanalyses within a region (Fig. 3l). MERRA-2 stands out in this analysis as the only product with positive near-surface specific humidity trends in all U.S. regions, including the SW, where warming and drying have been well documented (e.g., MacDonald 2010). Observed (PRISM) humidity trends are large and positive in the NP and NE regions, but are positive throughout most of the eastern United States. Conversely, the SW and NW regions are characterized by negative humidity trends in PRISM. In all regions, the reanalysis-derived q trends bracket the PRISM-derived regional trends. The trend toward increasing near-surface humidity is significant (α = 0.10) in the NE region in PRISM and in all of the reanalyses. Only ERA-Interim and NARR produce a significant downward trend in q in the SW region. In the JRA-55 and MERRA-2 reanalyses, the upward trend in q is significant.

Since TE is dependent on both T and q [Eq. (2)], temporal trends in TE depend on the underlying temperature and humidity trends. Because NARR (T) and MERRA-2 (q) did not adequately reproduce the spatial patterns of the latter, the pattern correlations between PRISM and reanalysis-derived TE trends for those products are relatively low, although it should be noted that the other reanalyses (ERA-Interim and JRA-55) only correlate spatially with PRISM at around 0.5. Nonetheless, the regional values of the TE trends derived from ERA-Interim and JRA-55 show stronger agreement with PRISM (Fig. 4l). The largest linear trends in TE are found in the NE region, where the positive trends are significant in PRISM as well as three of the four reanalyses. Small increases in temperature coupled with large increases in humidity resulted in a significant trend in the NP region in PRISM. With the exception of the SW, the reanalyses agree with PRISM in terms of the sign of the TE trend, but there are substantial differences among the data products in terms of the trend magnitude.

The analyses presented thus far demonstrate that 1) no single reanalysis product is optimal in terms of all of the metrics considered, and 2) temperature and equivalent temperature differ in terms of both their general statistical characteristics and temporal trends. Consistent with a priori expectations that assimilation of station T and q will improve depiction of weather and climate in reanalyses (Bosilovich 2013), our results indicate generally superior performance of ERA-Interim and JRA-55 with respect to temporal variability and trends, although there are instances where these products perform less well than MERRA-2 and NARR. Because the reanalyses exhibit individual strengths and weaknesses relative to observations, our analysis of heat wave characteristics is conducted with all the products and we use the results presented in this section to contextualize our findings related to changes in heat wave characteristics.

b. Historical changes in heat wave characteristics

1) Heat wave day frequency

The HWD definition adopted in this study is based on the 90th percentile of T (section 3b). However, the requirement that the 3-day average T exceeds the 90th percentile results in substantially fewer than 10% of summer days meeting the criteria. Based on the PRISM dataset, the regions in Fig. 1 experience an average from 4.4 (NP) to 5.5 (SP) HWDs per summer, but the partitioning of days between T-only and T-and-TE HWDs differs among the regions (Figs. 5a,b), and all regions exhibits substantial interannual variability in HWD frequency (Fig. 6). For example, T-only HWDs make up more than 80% of all HWDs in the SP region, but only around 20% of the HWDs in the NE region (Figs. 5a,b).

Fig. 5.
Fig. 5.

Mean annual HWD frequency and magnitude by region. Results are shown for the PRISM dataset as well as the ERA-Interim, JRA-55, MERRA-2, and NARR reanalysis products. The frequency of (a) T-only and (b) T-and-TE HWDs are shown. (c) The mean magnitude (see section 3b) of T on T-only HWDs. The mean magnitudes of (d) T and (e) TE on T-and-TE HWDs are shown.

Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0393.1

Fig. 6.
Fig. 6.

Time series of summer (JJA) HWD frequency (days yr−1) for (a)–(f) U.S. regions (as in Fig. 1) based on the PRISM dataset. For each subplot, time series are presented for T-only HWDs (circles with accompanying solid lines) and T-and-TE HWDs (squares with dotted lines). Time series with significant linear trends (with α = 0.10; see section 3) are indicated with filled symbols and an accompanying trend line. Numerical values of the trends are provided in Table 2.

Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0393.1

Because the heat wave definition adopted here requires multiple days above the 90th-percentile value, persistence of daily temperature is a key characteristic of reanalysis data in the context of T-only HWD frequency. Reanalyses with high persistence in daily T, as characterized by the lag-1 Spearman correlation, generally have higher regional mean HWD counts. Considering each region and data product as a data point (n = 35), the correlation between the lag-1 correlation of T and regional mean HWD frequency is 0.57 (p < 0.01). Similarly, the partitioning of HWDs among T-only and T-and-TE HWDs exhibits dependence on the coupling of T and q, as characterized by the Spearman correlation between their daily means. Figure 7a shows the frequency of T-and-TE HWDs as a function of the correlation between daily mean T and q and demonstrates that tighter coupling between daily T and q is associated with a higher proportion of heat waves characterized by high temperature and high humidity (i.e., T-and-TE HWDs). In general, there is closer accord in both frequency of HWDs and partitioning of HWDs in those reanalyses that assimilate near-surface T and/or q (JRA-55, ERA-Interim, and NARR), while MERRA-2 tends to overestimate T-only HWD frequency and underestimate T-and-TE HWD frequency in the western and central U.S. regions.

Fig. 7.
Fig. 7.

Scatter diagrams exhibiting the relationships between the Spearman correlation of daily T and q and (a) the frequency of T-and-TE HWDs and (b) the mean TE magnitude (as defined in section 3) on T-and-TE HWDs.

Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0393.1

The time series of regional HWD frequencies (Fig. 6) also highlight the utility of our framework for identifying regional variations in the partitioning of HWDs between T-only and T-and-TE varieties as well as important historical events. Note, for example, the high number of T-and-TE HWDs, and relatively few T-only HWDs in the MW region during the summer of 1995 (Fig. 6c), which led to major loss of life in Chicago and surrounding areas (Kunkel et al. 1996). Conversely, the summer of 1988 was characterized by a much larger number of total heat waves, with similar numbers of T-only and T-and-TE HWDs, in the central U.S. regions, consistent with the widespread North American drought of that summer (Trenberth and Branstator 1992). The 2011 heat wave (Hoerling et al. 2013) was also associated with drought conditions and produced record numbers of HWDs, nearly all T-only, in the SP region (Fig. 6e).

Trend analyses conducted on the time series in Fig. 6 (and similar series in the reanalyses products) result in identification of several significant trends and many trends that, while not statistically significant, are of consistent sign among the datasets employed (Table 2). With the exception of the NP region, which is characterized by nonsignificant decreases in all of the data products, all regions are characterized by positive trends in T-only HWDs in PRISM and the four reanalyses tested. The largest increases in T-only HWDs have occurred in the SP and SW regions, where the trends are statistically significant (α = 0.10) in all of the datasets (Table 2). The T-only HWD trends are also significant in the NW region for PRISM, ERA-Interim, and JRA-55 and in the SE for ERA-Interim and JRA-55. While the magnitudes of the trends vary within regions, the PRISM- and reanalysis-derived trends show good qualitative agreement. Similarly, PRISM and all of the reanalyses produce positive trends in T-and-TE HWD frequency in all of the continental U.S. (CONUS) regions. The largest trends (Table 2) are in the NE, NW, and MW regions, and despite large interannual variability in T-and-TE HWD frequency, the trends are statistically significant in some of the datasets (MERRA-2 and NARR for the NE and MW regions, ERA-Interim and JRA-55 for the NW). The SP region is characterized by smaller trends, but given less variability in the time series, the trends are significant in all of the products tested (Table 2).

Table 2.

Linear trends in the number of HWDs (n decade−1) for T-only and T-and-TE heat wave types. Boldface trends indicate significance with α = 0.10.

Table 2.

2) Spatial extent of heat wave conditions

The changes in heat wave day frequency presented in section 4b(1) may be manifest as more or less spatially extensive heat waves. We therefore computed the spatial proportion of each region impacted by each type of heat wave event for each summer day from 1981 to 2015. For each summer, we then computed the mean and maximum proportions over all days for which the proportion was greater than zero. The resulting time series for the mean spatial proportion (from PRISM) are shown in Fig. 8, and linear trends in the mean and maximum spatial proportion from all data products are presented in Table 3.

Fig. 8.
Fig. 8.

Time series of mean regional proportion impacted by heat wave conditions (proportion per year) for (a)–(g) U.S. regions (as in Fig. 1) based on the PRISM dataset. For each subplot, time series are presented for T-only HWDs (circles with accompanying solid lines) and T-and-TE HWDs (squares with accompanying dotted lines). Time series with significant linear trends (with α = 0.10) are indicated with filled symbols and an added trend line. Numerical values of the trends are provided in Table 3.

Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0393.1

Table 3.

Linear trends in the mean and maximum regional areal proportions (% decade−1) of regional area impacted by heat wave conditions for T-only and T-and-TE heat wave types. Boldface trends indicate significance with α = 0.10.

Table 3.

The largest trends in the mean spatial proportion of T-only HWDs correspond to the regions with significant trends in T-only HWD frequency presented in Table 2 and Fig. 6. Specifically, the SP, SW, and NW regions exhibit positive trends in all of the data products with significant trends in PRISM, ERA-Interim, and JRA-55. As with the T-only HWD frequency, the NP region is characterized by (insignificant) decreases in T-only HWD areal proportion in all of the data products. In regions characterized by increases in the mean areal proportion impacted by T-only heat wave conditions, the maximum areal proportion impacted by T-only HWDs has also increased, and has done so at a much faster rate than reported for the mean (Table 3). All of the data products tested produce significant positive trends in the maximum summer proportion of the SP and SW regions impacted by T-only heat wave conditions (Table 3).

Trends in the mean regional proportion impacted by T-and-TE heat wave conditions are positive in all data products for the SE, SP, and NW regions, but lack strong agreement among the reanalyses with respect to the magnitude and statistical significance of the trend. In other regions, the trends in regional proportion derived from different datasets span zero indicating uncertainty in the temporal evolution of the average footprint of T-and-TE heat wave conditions. There is less ambiguity in the maximum spatial proportion of T-and-TE heat wave conditions, which exhibit positive trends across data products in all regions except the NP. In other regions, PRISM and all four reanalysis products are characterized by positive trends in the maximum regional proportion impacted by T-and-TE heat waves, with significant trends in some but not all reanalyses.

3) Heat wave day magnitude

Even though the percentiles used in our heat wave definition are location specific, there are still important regional variations in heat wave magnitude (Fig. 5). For T-only HWDs, the average T magnitude in the PRISM dataset ranges from around 0.9°C in the SE region to around 1.6°C in the NP region. Values derived from the reanalyses are generally slightly higher (Fig. 5c). In most regions, temperatures are lower on T-only HWDs relative to T-and-TE HWDs, which also exhibit greater variability between regions (cf. Figs. 5c and 5d). Overall, the reanalyses show agreement with each other and with PRISM with respect to average temperature characteristics of T-and-TE heat waves. The magnitude of TE during T-and-TE heat waves is likewise characterized by strong regional variations, but good agreement among PRISM and the reanalyses. Variations among regions are intuitive; the SW and SP regions tend to have dry heat waves relative to other regions, resulting in lower TE magnitude across HWD types. Regional differences in TE magnitude on T-and-TE HWDs and, in some cases, variations among reanalysis products within regions are highly correlated (r = 0.84, p < 0.01) with the correlation between daily T and q (Fig. 7b), with greater TE magnitudes being associated with stronger coupling of T and q.

To assess trends in the magnitude of T and TE on HWDs, we computed the mean magnitude (difference between the daily T or TE and the 90th percentile threshold) for each day at locations with positive (i.e., heat wave) values. We then computed the average and maximum value over each summer to produce time series (Fig. 9). In the western regions (SW, NW), where the background warming has been greatest, there has been an increase in the magnitude of T on T-only HWDs. This trend is positive across datasets in both regions and significantly positive in the PRISM and JRA-55 products (Table 4). The SE region is also characterized by trends that are positive across data products but not large enough to be considered statistically significant. In other regions, the data products produce a mixture of positive and negative trends, leading to greater uncertainty in the evolution of the magnitude of temperature during T-only heat wave events. The reanalyses show similarly small trends of mixed sign for T magnitude during T-and-TE heat waves (Table 4).

Fig. 9.
Fig. 9.

Time series of mean heat wave magnitude for U.S. regions (as in Fig. 1) based on the PRISM dataset. For each subplot, time series are presented for T on T-only HWDs (circles with accompanying solid lines) and T-and-TE HWDs (squares with accompanying dotted lines) and TE on T-and-TE HWDs (diamonds accompanying dotted lines). Time series with significant linear trends (with α = 0.10) are indicated with filled symbols and an added trend line. Numerical values of the trends are provided in Table 4.

Citation: Journal of Applied Meteorology and Climatology 56, 9; 10.1175/JAMC-D-16-0393.1

Table 4.

Linear trends in the mean and maximum magnitude (°C decade−1) of regional heat wave conditions for T-only and T-and-TE heat wave types. Boldface trends indicate significance (α = 0.10).

Table 4.

Trends in the magnitude of TE during T-and-TE heat waves show greater similarity among the datasets (Table 4). In the PRISM dataset, the trends are significant for the MW and NP regions. While not all of the reanalyses produce significant trends, they are all positive in agreement with PRISM. Positive trends are also observed in all data products in the NE region, with significant trends in the MERRA-2 and NARR products. The SW region is characterized by declining TE during T-and-TE heat wave events (significantly in the ERA-Interim and JRA-55 reanalyses), consistent with the general drying trend observed in that region.

Trends in maximum T magnitude on T-only HWDs show consistent (among the datasets) and significant (in PRISM and at least one reanalysis) increases in the SE and NW regions over the period 1981–2015 (Table 3). In other regions, there are variations in the sign of the trend among data products. Trends in maximum T magnitude on T-and-TE HWDs are likewise variable among the reanalyses and are only of consistent sign in the NE region, where the PRISM trend is significant (α = 0.1). Trends in maximum TE magnitude on T-and-TE HWDs are more consistent among the data products, with increases in the NE, NP, and SP regions across datasets.

5. Summary and conclusions

We have conducted the first analysis of U.S. regional heat wave characteristics using T and TE and compared time series of PRISM- and reanalysis-derived temperature (T) and equivalent temperature (TE) to investigate recent historical changes in heat wave characteristics. Our results show that heat wave days (HWDs) are generally becoming more frequent and more spatially expansive, consistent with increases in summer air temperature. With the exception of the NP region, where T-only HWD frequency has declined (insignificantly), there is a tendency toward increasing frequency of both dry (T only) and humid (T and TE) heat wave days. In regions characterized by large increases in HWD frequency, there is also a tendency toward larger impacted areas. We also report changes in the magnitude of T and TE when their respective heat wave criteria are met. Our results show significant warming of T-only heat waves in the SW region and significant increases in TE during T-and-TE heat waves in the MW and NP regions.

Our findings are subject to several caveats. Our analysis is based on one observation-driven dataset (PRISM) and four reanalysis products that assimilate surface temperature and humidity observations to varying degrees. However, the observational dataset that is used here as a reference is subject to uncertainties related to observations as well as methods (i.e., interpolation). As noted by Daly (2006), the true errors associated with gridded climate data are largely unknown because the true underlying field is unknown. Nevertheless, PRISM has been shown by Daly et al. (2008) to have smaller errors than other gridded products based on station data, largely because of improved consideration of physiographic features in the interpolation process. Because none of the reanalyses exhibit perfect agreement with PRISM across metrics, our findings are contextualized by considering results from all of the products considered. The general agreement among the results from different reanalysis suggests that our findings are robust. For example, none of the data products considered here have significant trends of opposing sign for any region and any metric in Tables 24. For approximately 60% of the trend tests conducted, PRISM and the four reanalyses agree on the sign of the regional trend, but differ in terms of the magnitude and statistical significance of the reported change. This finding is consistent with the work on CONUS temperature trends by Vose et al. (2012) and regional assessment of reanalysis-derived trends in summer temperature and precipitation by Bosilovich (2013). Nevertheless, development of high-resolution reference datasets for a range of variables should continue to be a key goal for improved validation of reanalysis products.

Here, we have reported regional trends in T and TE and their associated heat wave–related metrics, but largely avoided attribution of significant trends to specific drivers beyond the cross correlation and autocorrelation of temperature and humidity. In the case of heat waves, these drivers are likely to occur across a range of spatial and temporal scales as described in the introduction. Grotjahn et al. (2016) describe methods for identifying and characterizing large-scale meteorological processes associated with extreme temperatures including large-scale modes of climate variability and communication of those variations to the regional scale. For example, the positive tendencies in time series of T-only heat wave frequency and magnitude, coupled with negative trends in specific humidity in the SW region, can be linked to broad regional warming that may be attributable to observed widening of the tropics (Seidel et al. 2008), while changes in other regions have been attributed to combinations of natural variability (e.g., Johnstone and Mantua 2014) and radiative forcing from greenhouse gases.

Recent work indicates that regional forcing can also play an important role in the evolution of the joint distribution of temperature and humidity. A lack of warming in parts of the NP and MW regions is consistent with the observed “warming hole,” which has been attributed to hydrologic feedback (Pan et al. 2004), aerosol forcing (Yu et al. 2014), and tropical Pacific variability (Meehl et al. 2015). Pryor et al. (2016) have demonstrated that the role irrigation plays in equivalent temperature variability depends strongly on the regional climate and land cover. In an application to oppressive heat events in Illinois, Ford and Schoof (2017) identified similar large-scale circulation during T-only and T-and-TE events and an important role for regional soil moisture and land–atmosphere interaction. Along with urban heat island effects, similar drivers were reported in the analysis of the 1995 Midwest heat wave by Kunkel et al. (1996) but have not been widely adopted in studies about future heat wave vulnerability.

Our analysis has demonstrated robust increases in both temperature and equivalent temperature across most U.S. regions in recent decades. In many cases, these changes are manifest as increases in the frequency and intensity of extreme heat events and, in one region, as differential changes in the frequency of heat waves characterized by low (T only) and high (T and TE) humidity. Given the importance of humidity in determining the impacts of extreme temperatures, our findings reflect an important step toward understanding the changing impacts of extreme heat events in U.S. regions. Improved understanding of heat wave characteristics and their dependence on processes occurring across scales should be a key focus for reducing impacts associated with extreme heat.

Acknowledgments

This material is based upon work supported by the U.S. National Science Foundation under Grants 1339629, 1339655, and 1502400, and the U.S. Department of Energy (DE-SC0016438). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the view of the National Science Foundation or Department of Energy. We also acknowledge the constructive comments of three anonymous reviewers.

REFERENCES

  • Anderson, G. B., and M. L. Bell, 2011: Heat waves in the United States: Mortality risk during heat waves and effect modification by heat wave characteristics in 43 U.S. communities. Environ. Health Perspect., 119, 210218, doi:10.1289/ehp.1002313.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bolton, D., 1980: The computation of equivalent potential temperature. Mon. Wea. Rev., 108, 10461053, doi:10.1175/1520-0493(1980)108<1046:TCOEPT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bosilovich, M. G., 2013: Regional climate and variability of NASA MERRA and recent reanalyses: U.S. summertime precipitation and temperature. J. Appl. Meteor. Climatol., 52, 19391951, doi:10.1175/JAMC-D-12-0291.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bosilovich, M. G., and Coauthors, 2015: MERRA-2: Initial evaluation of the climate. NASA Tech. Memo. NASA/TM-2015-104606/Vol. 43, 145 pp., https://gmao.gsfc.nasa.gov/pubs/docs/Bosilovich803.pdf.

  • Brown, P. J., and A. T. DeGaetano, 2013: Trends in U.S. surface humidity, 1930–2010. J. Appl. Meteor. Climatol., 52, 147163, doi:10.1175/JAMC-D-12-035.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Collins, M., and Coauthors, 2013: Long-term climate change: Projections, commitments, and irreversibility. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 1029–1136.

  • Dai, A., 2006: Recent climatology, variability, and trends in global surface humidity. J. Climate, 19, 35893606, doi:10.1175/JCLI3816.1.

  • Daly, C., 2006: Guidelines for assessing the suitability of spatial climate data sets. Int. J. Climatol., 26, 707721, doi:10.1002/joc.1322.

  • Daly, C., M. Halbleib, J. I. Smith, W. P. Gibson, M. K. Doggett, G. H. Taylor, J. Curtis, and P. P. Pasteris, 2008: Physiographically sensitive mapping of climatological temperature and precipitation across the conterminous United States. Int. J. Climatol., 28, 20312064, doi:10.1002/joc.1688.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Davey, C. A., R. A. Pielke Sr., and K. P. Gallo, 2006: Differences between near-surface equivalent temperature and temperature trends for the eastern United States: Equivalent temperature as an alternative measure of heat content. Global Planet. Change, 54, 1932, doi:10.1016/j.gloplacha.2005.11.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Davis, L. W., and P. J. Gertler, 2015: Contribution of air conditioning adoption to future energy use under global warming. Proc. Natl. Acad. Sci. USA, 112, 59625967, doi:10.1073/pnas.1423558112.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, doi:10.1002/qj.828.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fall, S., N. S. Diffenbaugh, D. Niyogi, R. A. Pielke Sr., and G. Rochon, 2010: Temperature and equivalent temperature over the United States (1979–2005). Int. J. Climatol., 30, 20452054, doi:10.1002/joc.2094.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ford, T. W., and J. T. Schoof, 2016: Oppressive heat events in Illinois related to antecedent wet soils. J. Hydrometeor., 17, 27132726, doi:10.1175/JHM-D-16-0075.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Ford, T. W., and J. T. Schoof, 2017: Characterizing extreme and oppressive heat waves in Illinois. J. Geophys. Res. Atmos., 122, 682298, doi:10.1002/2016JD025721.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Gaffen, D. J., and R. J. Ross, 1998: Increased summertime heat stress in the U.S. Nature, 396, 529530, doi:10.1038/25030.

  • Gaffen, D. J., and R. J. Ross, 1999: Climatology and trends in U.S. surface humidity and temperature. J. Climate, 12, 811828, doi:10.1175/1520-0442(1999)012<0811:CATOUS>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grotjahn, R., and Coauthors, 2016: North American extreme temperature events and related large scale meteorological patterns: A review of statistical methods, dynamics, modeling, and trends. Climate Dyn., 46, 11511184, doi:10.1007/s00382-015-2638-6.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Grundstein, A., and J. Dowd, 2011: Trends in extreme apparent temperatures over the United States, 1949–2010. J. Appl. Meteor. Climatol., 50, 16501653, doi:10.1175/JAMC-D-11-063.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Held, I. M., and B. J. Soden, 2006: Robust response of the hydrologic cycle to global warming. J. Climate, 19, 56865699, doi:10.1175/JCLI3990.1.

  • Hoerling, M., and Coauthors, 2013: Anatomy of an extreme event. J. Climate, 26, 28112832, doi:10.1175/JCLI-D-12-00270.1.

  • Hondula, D. M., R. C. Balling Jr., J. K. Vanos, and M. Georgescu, 2015: Rising temperatures, human health, and the role of adaptation. Curr. Climate Change Rep., 1, 144154, doi:10.1007/s40641-015-0016-4.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Johnstone, J. A., and N. J. Mantua, 2014: Atmospheric controls on northeast Pacific temperature variability and change, 1900–2012. Proc. Natl. Acad. Sci. USA, 111, 14 36014 365, doi:10.1073/pnas.1318371111.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kobayashi, S., and Coauthors, 2015: The JRA-55 reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 548, doi:10.2151/jmsj.2015-001.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kravchenko, J., A. P. Abernethy, M. Fawzy, and H. K. Lyerly, 2013: Minimization of heatwave morbidity and mortality. Amer. J. Prev. Med., 44, 274282, doi:10.1016/j.amepre.2012.11.015.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kunkel, K. E., S. A. Changnon, B. C. Reinke, and R. W. Arritt, 1996: The July 1995 heat wave in the Midwest: A climatic perspective and critical weather factors. Bull. Amer. Meteor. Soc., 77, 15071518, doi:10.1175/1520-0477(1996)077<1507:TJHWIT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lanzante, J. R., 1996: Resistant, robust and non-parametric techniques for the analysis of climate data: Theory and examples, including applications to historical radiosonde station data. Int. J. Climatol., 16, 11971226, doi:10.1002/(SICI)1097-0088(199611)16:11<1197::AID-JOC89>3.0.CO;2-L.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Lobell, D. B., M. J. Roberts, W. Schlenker, N. Braun, B. B. Little, R. M. Rejesus, and G. L. Hammer, 2014: Greater sensitivity to drought accompanies maize yield increase in the U.S. Midwest. Science, 344, 516519, doi:10.1126/science.1251423.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • MacDonald, G. M., 2010: Water, climate change, and sustainability in the Southwest. Proc. Natl. Acad. Sci. USA, 107, 21 25621 262, doi:10.1073/pnas.0909651107.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Meehl, G. A., J. M. Arblaster, and C. T. Y. Chung, 2015: Disappearance of the southeast U.S. “warming hole” with the late 1990s transition of the Interdecadal Pacific Oscillation. Geophys. Res. Lett., 42, 55645570, doi:10.1002/2015GL064586.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Melillo, J. M., T. C. Richmond, and G. W. Yohe, Eds., 2014: Climate Change Impacts in the United States: The Third National Climate Assessment. U.S. Global Change Research Program, 841 pp., doi:10.7930/J0Z31WJ2.

    • Crossref
    • Export Citation
  • Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343360, doi:10.1175/BAMS-87-3-343.

  • Metzger, K. B., K. Ito, and T. D. Matte, 2010: Summer heat and mortality in New York City: How hot is too hot? Environ. Health Perspect., 118, 8086, doi:10.1289/ehp.0900906.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Miller, N. L., K. Hayhoe, J. Jin, and M. Auffhammer, 2008: Climate, extreme heat, and electricity demand in California. J. Appl. Meteor. Climatol., 47, 18341844, doi:10.1175/2007JAMC1480.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mishra, V., and K. A. Cherkauer, 2010: Retrospective droughts in the crop growing season: implications to corn and soybean yield in the Midwestern United States. Agric. For. Meteor., 150, 10301045, doi:10.1016/j.agrformet.2010.04.002.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • NWS, 2016: 76-year list of severe weather fatalities. NOAA, 2 pp., http://www.nws.noaa.gov/om/hazstats/resources/weather_fatalities.pdf.

  • Pan, Z., R. W. Arritt, E. S. Takle, W. J. Gutowski, C. J. Anderson, and M. Segal, 2004: Altered hydrologic feedback in a warming climate introduces a “warming hole.” Geophys. Res. Lett., 31, L17109, doi:10.1029/2004GL020528.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., Sr., C. Davey, and J. Morgan, 2004: Assessing “global warming” with surface heat content. Eos, Trans. Amer. Geophys. Union, 85, 210211, doi:10.1029/2004EO210004.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pryor, S. C., and J. T. Schoof, 2016: Evaluation of near-surface temperature, humidity and equivalent temperature from regional climate models in type-II downscaling. J. Geophys. Res. Atmos., 121, 33263338, doi:10.1002/2015JD024539.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Pryor, S. C., R. C. Sullivan, and T. Wright, 2016: Quantifying the roles of changing albedo, emissivity, and energy partitioning in the impact of irrigation on atmospheric heat content. J. Appl. Meteor. Climatol., 55, 16991706, doi:10.1175/JAMC-D-15-0291.1.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Schoof, J. T., Z. A. Heern, M. D. Therrell, and J. W. F. Remo, 2015: Assessing trends in lower tropospheric heat content in the central United States using equivalent temperature. Int. J. Climatol., 35, 28282836, doi:10.1002/joc.4175.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Seidel, D. J., Q. Fu, W. J. Randel, and T. J. Reichler, 2008: Widening of the tropical belt in a changing climate. Nat. Geosci., 1, 2124, doi:10.1038/ngeo.2007.38.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Sen, P. K., 1968: Estimates of the regression coefficient based on Kendall’s tau. J. Amer. Stat. Assoc., 63, 13791389, doi:10.1080/01621459.1968.10480934.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Seneviratne, S. I., M. G. Donat, B. Mueller, and L. V. Alexander, 2014: No pause in the increase of hot temperature extremes. Nat. Climate Change, 4, 161163, doi:10.1038/nclimate2145.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Souch, C., and C. S. B. Grimmond, 2004: Applied climatology: ‘Heat waves.’ Prog. Phys. Geogr., 28, 599606, doi:10.1191/0309133304pp428pr.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Steadman, R. G., 1984: A universal scale of apparent temperature. J. Climate Appl. Meteor., 23, 16741687, doi:10.1175/1520-0450(1984)023<1674:AUSOAT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res., 106, 71837192, doi:10.1029/2000JD900719.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., and G. W. Branstator, 1992: Issues in establishing causes of the 1998 drought over North America. J. Climate, 5, 159172, doi:10.1175/1520-0442(1992)005<0159:IIECOT>2.0.CO;2.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Vose, R. S., S. Applequist, M. J. Menne, C. N. Williams Jr., and P. Thorne, 2012: An intercomparison of temperature trends in the U.S. Historical Climatology Network and recent atmospheric reanalyses. Geophys. Res. Lett., 39, L10703, doi:10.1029/2012GL051387.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Yu, S., and Coauthors, 2014: Attribution of the United States “warming hole”: Aerosol indirect effect and precipitable water vapor. Sci. Rep., 4, 6929, doi:10.1038/srep06929.

    • Crossref
    • Search Google Scholar
    • Export Citation
Save