Abstract

Climatological annual and seasonal dewpoint, specific humidity, and relative humidity maps for the United States are presented using hourly data from 188 first-order weather stations for the period 1961–90. Separate climatologies were calculated for daytime (three observations per day between 0800 and 1600 LST), nighttime (three observations per day between 2000 and 0400 LST), and the full day (eight observations per day, every 3 h).

With extended datasets for the period 1961–95, trends in these same variables and temperature are calculated for each of 170 stations and for eight regions of the country. The data show increases in specific humidity of several percent per decade, and increases in dewpoint of several tenths of a degree per decade, over most of the country in winter, spring, and summer. Nighttime humidity trends are larger than daytime trends. The specific humidity increases are consistent with upward temperature trends. The upward temperature and humidity trends are also consistent with upward trends in apparent temperature, a measure of human comfort based on temperature and humidity. Relative humidity trends are weaker than the specific humidity trends, but they do show evidence of increases, especially in winter and spring.

The possibility that the detected trends may be artifacts of changes in instrumentation was examined, but several lines of reasoning suggest that they are not. Anthropogenic water vapor produced from fossil fuel consumption, both locally and globally, is too small a source to explain the observed trends.

Different nations use different algorithms or other techniques to convert temperatures and relative humidities from radiosonde observations to dewpoint depressions. Thus, it is possible for identical measured values to result in different reported dewpoints. On the basis of a sample of conversion methods, we calculate the possible differences among the national practices. In general, the discrepancies are not large and would often be lost in the usual round-off procedures associated with transmission over the Global Telecommunications System, but in cold, dry conditions dewpoints different by more than 1°C could be reported for identical conditions. Some of the methods have been changed over time, so there is also the possibility of inhomogeneities in climate records.

This paper considers the use of upper-air data from radiosondes in long-term climate studies. The accuracy and precision of radiosonde humidity measurements, including temperature and pressure measurements used in calculating them, and their effects on the precision of reported and derived variables are estimated. Focusing on the U.S. radiosonde system, we outline the history of changes in instruments and reporting practices and attempt to assess the implications of such changes for studies of temporal variations in lower-tropospheric water vapor. Changes in biases in the data are highlighted, as these can lead to misinterpretation of climate change. We conclude that the upper-air data record for the United States is not homogeneous, especially before 1973. Because of problems with the humidity data in cold, dry conditions, the water vapor climatology in the upper troposphere, nominally above the 500-mb level, is not well known.

Abstract

With radiosonde data from 15 Northern Hemisphere stations, surface-to-400-mb column water vapor is computed from daytime soundings for 1988–1990. On the basis of simultaneous surface visual cloud observations, the data are categorized according to sky-cover amount. Climatological column water vapor content in clear skies is shown to be significantly lower than in cloudy skies. Column water vapor content in tropical regions varies only slightly with cloud cover, but at midlatitude stations, particularly in winter, clear-sky values are much lower. The variation in column water content with cloud cover is not simply due to variations in atmospheric temperature, since the increase in water vapor with cloud cover is generally associated with a decrease in daytime temperature. Biases in radiosonde instruments associated with cloudiness do not explain the station-to-station variations in the magnitude of the increase of column water vapor with cloud cover. Statistics are presented that can be used as guidance in estimating the bias in water vapor climatologies based on clear-sky or partly cloudy-sky measurements. These may be helpful in distinguishing the clear- and cloudy-sky greenhouse effects of water vapor.

Abstract

Climatological surface temperature and humidity variables for China are presented based on 6-hourly data from 196 stations for the period of 1961–90. Seasonal and annual means for daytime, nighttime, and the full day are shown. The seasonal cycle of moisture is primarily controlled by the east Asia monsoon system, with dominant factors of temperature change in northern and western China and of moisture advection associated with monsoon circulations in the southeast.

Trends during 1951–94 are estimated for each station and for four regions of the country, with attention paid to the effects of changes in instrumentation, observing time, and station locations. The data show evidence of increases in both temperature and atmospheric moisture content. Temperature and specific humidity trends are larger at nighttime than daytime and larger in winter than summer. Moisture increases are observed over most of China. The increases are several percent per decade for specific humidity, and several tenths of a degree per decade for temperature and dewpoint. Increasing trends in summertime temperature and humidity contribute to upward trends in apparent temperature, a measure of human comfort.

Abstract

Radiosonde data from a global 118-station network are used to determine the spatial and temporal scales of variability of tropospheric water vapor. Various sources of possible error and bias in the data are analyzed. Changes in instrumentation at U.S. stations are shown to have a considerable influence on the record; information on comparable changes in other countries is not readily available. Mean monthly data are shown to be acceptable at tropical nations but not at high-latitude stations, where the nonlinear dependence of saturation vapor pressure on temperature, coupled with large temperature ranges, leads to biases of up to 10% in mean monthly specific humidity.

A series of three empirical orthogonal function analyses (for the tropics, North America, and the globe) of specific humidity at the surface, 850-mb, 700-mb, and 500-mb levels is presented. All three show evidence of a shift in the specific humidity field in the winter of 1976/77, with generally lower values from the beginning of the record (January 1973) until the shift and higher values through the winter of 1985/86. This shift is shown to be consistent with other evidence for a change in “climate state” in about 1977. The influence of the El Niño-Southern Oscillation is evident in both the tropical and global analyses.

Abstract

Radiosonde data have been used, and will likely continue to be used, for the detection of temporal trends in tropospheric and lower-stratospheric temperature. However, the data are primarily operational observations, and it is not clear that they are of sufficient quality for precise monitoring of climate change. This paper explores the sensitivity of upper-air temperature trend estimates to several data quality issues.

Many radiosonde stations do not have even moderately complete records of monthly mean data for the period 1959–95. In a network of 180 stations (the combined Global Climate Observing System Baseline Upper-Air Network and the network developed by J. K. Angell), only 74 stations meet the data availability requirement of at least 85% of nonmissing months of data for tropospheric levels (850–100 hPa). Extending into the lower stratosphere (up to 30 hPa), only 22 stations have data records meeting this requirement for the same period, and the 30-hPa monthly data are generally based on fewer daily observations than at 50 hPa and below. These networks show evidence of statistically significant tropospheric warming, particularly in the Tropics, and stratospheric cooling for the period 1959–95. However, the selection of different station networks can cause network-mean trend values to differ by up to 0.1 K decade⁻¹.

The choice of radiosonde dataset used to estimate trends influences the results. Trends at individual stations and pressure levels differ in two independently produced monthly mean temperature datasets. The differences are generally less than 0.1 K decade⁻¹, but in a few cases they are larger and statistically significant at the 99% confidence level. These cases are due to periods of record when one dataset has a distinct bias with respect to the other.

The statistical method used to estimate linear trends has a small influence on the result. The nonparametric median of pairwise slopes method and the parametric least squares linear regression method tend to yield very similar, but not identical, results with differences generally less than ±0.03 K decade⁻¹ for the period 1959–95. However, in a few instances the differences in stratospheric trends for the period 1970–95 exceed 0.1 K decade⁻¹.

Instrument changes can lead to abrupt changes in the mean, or change-points, in radiosonde temperature data records, which influence trend estimates. Two approaches to removing change-points by adjusting radiosonde temperature data were attempted. One involves purely statistical examination of time series to objectively identify and remove multiple change-points. Methods of this type tend to yield similar results about the existence and timing of the largest change-points, but the magnitude of detected change-points is very sensitive to the particular scheme employed and its implementation. The overwhelming effect of adjusting time series using the purely statistical schemes is to remove the trends, probably because some of the detected change-points are not spurious signals but represent real atmospheric change.

The second approach incorporates station history information to test specific dates of instrument changes as potential change-points, and to adjust time series only if there is agreement in the test results for multiple stations. This approach involved significantly fewer adjustments to the time series, and their effect was to reduce tropospheric warming trends (or enhance tropospheric cooling) during 1959–95 and (in the case of one type of instrument change) enhance stratospheric cooling during 1970–95. The trends based on the adjusted data were often statistically significantly different from the original trends at the 99% confidence level. The intent here was not to correct or improve the existing time series, but to determine the sensitivity of trend estimates to the adjustments. Adjustment for change-points can yield very different time series depending on the scheme used and the manner in which it is implemented, and trend estimates are extremely sensitive to the adjustments.Overall, trends are more sensitive to the treatment of potential change-points than to any of the other radiosonde data quality issues explored.

Abstract

Simulations of humidity from 28 general circulation models for the period 1979–88 from the Atmospheric Model Intercomparison Project are compared with observations from radiosondes over North America and the globe and with satellite microwave observations over the Pacific basin. The simulations of decadal mean values of precipitable water (W) integrated over each of these regions tend to be less moist than the real atmosphere in all three cases; the median model values are approximately 5% less than the observed values.

The spread among the simulations is larger over regions of high terrain, which suggests that differences in methods of resolving topographic features are important. The mean elevation of the North American continent is substantially higher in the models than is observed, which may contribute to the overall dry bias of the models over that area. The authors do not find a clear association between the mean topography of a model and its mean W simulation, however, which suggests that the bias over land is not purely a matter of orography.

The seasonal cycle of W is reasonably well simulated by the models, although over North America they have a tendency to become moister more quickly in the spring than is observed. The interannual component of the variability of W is not well captured by the models over North America. Globally, the simulated W values show a signal correlated with the Southern Oscillation index but the observations do not. This discrepancy may be related to deficiencies in the radiosonde network, which does not sample the tropical ocean regions well. Overall, the interannual variability of W, as well as its climatology and mean seasonal cycle, are better described by the median of the 28 simulations than by individual members of the ensemble.

Tests to learn whether simulated precipitable water, evaporation, and precipitation values may be related to aspects of model formulation yield few clear signals, although the authors find, for example, a tendency for the few models that predict boundary layer depth to have large values of evaporation and precipitation. Controlled experiments, in which aspects of model architecture are systematically varied within individual models, may be necessary to elucidate whether and how model characteristics influence simulations.