Home > Climate Monitoring > Societal Impacts > Heat Stress Index

Heat Stress Index

Overview

Defining a Heat Stress Index

When both temperature and humidity are high, humans can experience considerable heat stress. In the U.S., extreme heat may have greater impact on human health (Kalkstein and Davis 1989), especially among the elderly (Changnon et al. 1996), than any other type of severe weather. The combined effects of temperature and humidity cannot be directly measured but can be assessed by calculation of an "apparent temperature" (A). Ignoring wind effects, one can estimate apparent temperature as A (°C) = -1.3 + 0.92T + 2.2e, where T is ambient air temperature (°C) and e is water vapor pressure (kPa) (Steadman 1984).

This index of "how hot it feels" should not be confused with the Heat Index used by NOAA's National Weather Service. Because the latter index is not defined for temperature below 80°F (27°C) and relative humidity below 40%, it is not suitable for compilation of a climatology. To include values below these limits would be a misuse of the Heat Index, but rejecting those conditions would introduce bias. Therefore we employ the Steadman (1984) apparent temperature.

Building a U.S. Heat Stress Climatology

In 1998, Gaffen and Ross (1998 and 1999) compiled an apparent temperature climatology for 187 first-order U.S. weather stations. They calculated the 85th percentiles of daily maximum, minimum and average apparent temperatures over the 1961-1990 base period, for each station, which were then used as threshold values for identifying extreme heat stress conditions. The 85th percentile values have been shown to be closely correlated with weather related mortality statistics (Kalkstein and Davis 1989). Analyzing data for 1948-1995, Gaffen and Ross (1998) showed increases in the number of exceedances of these threshold values at most of the U.S. stations studied, with the largest increases in the eastern and western regions.

Addressing Changes in Data Characteristics

NCDC's objective is to extend the Gaffen and Ross dataset and periodically update it. However, as with many climatological datasets, changes in data characteristics posed a challenge to constructing a homogeneous time series. Gaffen and Ross used the National Solar Radiation Data Base (NSRDB), a data base developed by the National Renewable Energy Laboratory (NREL) (NREL 1992) that was derived from surface airways observations. The NSRDB data have not been extended past 1995, so it was necessary to recalculate the base-period climatology using a data set available at NCDC that is continually updated on an operational basis.

We selected the TD3280 dataset of hourly and 3-hourly synoptic observations from first-order National Weather Service stations, which was the basic dataset used in constructing the NSRDB. Currently TD3280 data are routinely quality controlled and updated, and we expect that to continue in the foreseeable future. The highest and lowest daily temperatures were extracted and used to calculate the maximum, minimum and mean daily apparent temperatures. (Note: Because these data were taken from hourly and 3-hourly observations, they are not necessarily the absolute daily maximum and minimum values.) Although relative humidity values were generally available, the calculation of vapor pressure involved an estimation of station level pressure. With the implementation of the Automated Surface Observing System (ASOS), station level air pressure has not been reported in TD3280 since 1998, so mean sea level pressures were extracted and adjusted to station level with the formula P(z) = P(msl)*exp(-z/H), where P(z) is station level pressure, P(msl) is mean sea level pressure, z is station elevation and H is set at 8 km (the assumption of a constant barometer height may be a source of error). To ensure consistency across the entire period of record, this process was used for all years, even though surface pressure data were available prior to 1998.

Results for the 1961 - 1990 base period were consistent with those obtained by Gaffen and Ross, with the 85th percentile values generally within half a degree Celsius. However, because the TD3280 dataset had more missing or erroneous data, complete records could not be constructed for some stations. A total of 187 stations were used in the NCDC reconstruction. Time series of the number of days that exceeded the re-computed 85th percentile values can be found below.

Heat Stress Datasets and Documentation

References

  • Changnon, S.A., K.E. Kunkel and B.C. Reinke, 1996: Impacts and responses to the 1995 heat wave: A call to action. Bull. Am. Meteorol. Soc., 77, 1497-1506.
  • Gaffen, D.J., and R.J. Ross, 1998: Increased Summertime Heat Stress in the U.S. Nature, 396, 529-530.
  • Gaffen, D.J., and R.J. Ross, 1999: Climatology and Trends of U.S. Surface Humidity and Temperature. J. Climate, 12, 811-828.
  • Kalkstein, L.S., and R.E. Davis, 1989: Weather and human mortality: An evaluation of demographic and inter-regional responses in the U.S. ann. Assoc. Am. Geogr., 79, 44-64.
  • NREL, 1992: User's Manual - National Solar Radiation Data Base (1961 - 1990), Version 1.0. NSRDB Vol. 1, 93pp.
  • Steadman, R.G., 1984: A universal scale of apparent temperature. J. Climate Appl. Meteor., 23, 1674-1282.
  • National Weather Service Heatwave information web page.