Learn how plants respond to higher atmospheric CO2 concentrations

How does rising atmospheric CO2 affect marine organisms?

Click to locate material archived on our website by topic

Conterminous United States Temperature Trends -- Summary
Has the global warming of the past century, and especially that of the past few decades, been as dramatic as climate alarmists claim it has been, leading to unprecedented high temperatures and unsurpassed temperature variability?  We here explore this question as it applies to the conterminous United States by summarizing the results of a number of scientific papers that have broached this subject.

Defining an extreme temperature day as a day that exceeds by one standard deviation the long-term average temperature for a given day, Henderson and Muller (1997) analyzed trends in both warm and cold extreme temperature days for the four seasons of the year - winter, spring, summer, fall - for six states (Texas, Oklahoma, Arkansas, Louisiana, Mississippi and Tennessee) in the south-central United States over the period 1901-1987.  This effort revealed, in their words, that "since 1901 the overall temperature trend appears to be downward with more frequent cold days and less frequent warm days at most stations in all seasons."  In the winter, in fact, all stations studied showed an increase in the frequency of cold days through the study period, while over half of them displayed a decrease in the frequency of warm days.  In the spring, the majority of stations also showed an increasing number of cold days and a decreasing number of warm days; and summer results were similar, with even higher significance levels.  For this region of the United States, therefore, the bulk of the 20th century was a time of decreasing frequency of extreme warm days and increasing frequency of extreme cold days, both of which trends are just the opposite of what climate alarmists claim to be the case worldwide.

Suckling and Mitchell (2000) studied the spatial and temporal variation of the C/D Koppen climate boundary in the central United States over the 100-year period 1900 to 1999.  In this climate classification system, the C and D climates are both considered mid-latitude rainy climates, but with mild and cold winters, respectively.  The data used in the analysis were mean January temperatures obtained from the U.S. Historical Climatology Network for 67 sites located between 37 and 41.5N latitude and 90 and 100W longitude, comprising much of Missouri, eastern and central Kansas, south-central and southeastern Nebraska, southern Iowa and west-central Illinois.

Breaking the hundred-year time period into four equal parts, Suckling and Mitchell found that the C/D climate boundary was located slightly farther south during the last two periods (1950-1974 and 1975-1999) than it was during the first two periods (1900-1924 and 1925-1949).  As they describe it, "this implies that winters were colder or more severe during these latter periods compared to the two earlier quarter-century periods."  They also note that "cooler conditions for the latter half of the 20th century are further illustrated by the network-wide mean January temperature values," which were -3.34C for 1950-1974 and -3.24C for 1975-1999, as compared to -2.67C for 1900-1924 and -2.62C for 1925-1949.  What is more, we note in our Editorial of 14 Mar 2001 that the mean November-December temperature of 2000 was the lowest ever measured throughout the conterminous United States.

The results of the study of Suckling and Mitchell thus show, contrary to climate-alarmist predictions, that a northward migration of climatic zones in central North America does not appear to be occurring.  The authors say that this observation "suggests a lack of evidence for any systematic wintertime warming in the central United States that might be anticipated under a global-warming scenario."  They also note that the same holds true for the summer-sensitive Dfa/Dfb climate boundary (where Dfa climates have distinctly warmer summers than Dfb climates), as demonstrated by Mitchell and Kienholz (1997) in a similar study based on July mean temperatures in the north-central and northeastern United States.

Out on the east coast of the country, Cronin et al. (2003) reconstructed a 2200-year record of spring sea surface temperature from four sediment cores extracted from Chesapeake Bay (CB) between 1996 and 2000, using the magnesium/calcium (Mg/Ca) proxy method as a paleothermometer (Chivas et al., 1986).  They report that mean 20th century CB temperature was "not warmer than mean temperatures during MWP-I," the first stage of the Medieval Warm Period, which they delineate as occurring between 450 and 900 AD.  In addition, we note there were similar periods of equivalent warmth during the Roman Warm Period, between approximately 100 BC and 200 AD.  The warmest quarter-century of all, however, occurred between 1850 and 1875, after which temperatures plummeted, only to rise again about 1910.  This latter warming, which peaked in the 1940s and 50s, was more erratic than its immediate predecessor; and, therefore, its highest temperatures were not sustained for as long a period of time.  Then came another cooling, which was only briefly mitigated in the early 1980s (as close as we can tell from Cronin et al.'s graphs), after which temperatures dropped to some of their lowest levels of both the 19th and 20th centuries, during the final years of the 20th century, when climate alarmists claim the earth experienced a warming that was unprecedented over the past two millennia.

On the other side of the country, Schwing and Moore (2000) reviewed a number of data sets pertaining to meteorological and oceanographic phenomena just off the west coast of California.  They report that October 1999 marked a 13-month run of below-normal temperatures at Monterey, California, with a March-July mean that was 1.4C below the long-term average.  This remarkable run of unusually cool weather was linked to extreme oceanic conditions that prevailed over the same time period.  As the two researchers recount it, "in less than 2 years, sea-surface temperatures (SSTs) in this region went from being the warmest on record, during the height of El Nio in late 1997, to the lowest in decades," with a rapid ocean cooling of nearly 10C in some areas that they describe as "unprecedented."  This cooling was "especially impressive," they said, "given that the annual range of SST in coastal waters off central California is about 3C, and its interannual variance is only 1C."  These dramatic climate changes, in turn, produced a number of changes in the local ecology of the ocean, as zooplankton communities off Oregon and British Columbia shifted from predominantly warm-water to exclusively cold-water species, and as marine bird populations shifted from a prevalence of subtropical to subarctic species.

Moving inland across the northern part of the country, Klasner and Fagre (2002) studied summer temperatures and spring snowpack over the period 1927-1991 in the McDonald Creek drainage basin of Glacier National Park in Montana, as well as altitudinal and areal changes in subalpine fir forests between 1945 and 1991 at six 40-hectare sites located between elevations of 1900 and 2200 meters.  Interestingly, and in spite of recurring climate-alarmist warnings of highly-negative global-warming-induced impacts on Glacier National Park, their data revealed no net change in spring snowpack from the beginning to the end of the record.  Likewise, there was no net change in summer minimum temperature; but in the case of summer maximum temperature, there was a change: a net drop of about 0.7C from the beginning to the end of the study period.  Hence, it should come as no surprise, in their words, that "altitudinal changes in the location of the alpine treeline ecotone were not observed."  However, and in spite of the drop in summer maximum temperature, there was a 3.4% increase in the area of tree coverage from 1945 to 1991, as well as an increase in the density of trees.

In discussing their findings, Klasner and Fagre note that ecotones - or ecosystem gradients (such as from tundra to krummholz to patch-forest to continuous-canopy forest) resulting from environmental gradients - "are hypothesized to be sensitive indicators of climate change."  Hence, in light of the highly-hyped melting of glaciers in Glacier National Park, one would have expected to have seen trees moving upward in altitude over the half-century study period.  Such, however, was not the case, nor was there any indication of warming in the meteorological record.  What was observed was an increasing aerial coverage and density of trees, as would be expected to result from the ongoing rise in the air's CO2 content over the study period.  It would appear, therefore, that dreaded global warming is not only not affecting the ecosystems of Glacier National Park, it is not even occurring there.

With respect to the entire conterminous United States, Vega et al. (1998) combined 344 separate climate divisions encompassing the country into five different regions for purposes of temperature analysis for trends and variability shifts over the period 1895 to 1991.  This exercise revealed that temperature increased from the beginning to the end of the record in only one of the five temperature regions (the southwest), and that a decrease in interannual variability was reported for three of the five regions, which results are definitely not what climate alarmists claim has occurred throughout the world as a whole.

Iskenderian and Rosen (2000) analyzed two different mid-tropospheric temperature data sets spanning the last forty years, calculating day-to-day variability within each month, season and year.  Averaged over the conterminous United States, the two data sets both showed "mostly small positive trends in most seasons," but none of the trends was statistically significant.  Therefore, they concluded they "cannot state with confidence that there has been a change in synoptic-scale temperature variance in the midtroposphere over the United States since 1958."

DeGaetano and Allen (2002a,b) worked with data from the U.S. Historical Climatology Network to determine how both hot and cold temperature extremes - defined in terms of the number of exceedences of the 90th, 95th and 99th percentiles of their respective databases - have varied across the conterminous United States over a number of different time scales.  For the period 1960-96, they found that "a large majority of stations show increases in warm extreme temperature exceedences," which would seem to corroborate the claims of the world's climate alarmists.  They also report that "about 20% of the stations experience significant increases in warm maximum temperature occurrence," again in seeming vindication of climate-alarmist claims.  Also, they note that "similar increases in the number of >2 and >3 runs of extreme temperatures occur across the country," apparently substantiating the climate-alarmist contention of an increasing frequency of deadly heat waves.  However, when the two scientists extended their analyses further back in time, some quite different results were obtained.

Adding another 30 years of data onto the front ends of their databases, DeGaetano and Allen discovered there were "predominantly decreasing warm exceedence trends across the country during the 1930-96 period."  In fact, they found that "in the 1930-96 period 70% of the stations exhibit decreasing high extreme maximum temperature trends."  It is thus abundantly clear there has been no net increase in extreme hot weather events in the conterminous United States since 1930.  In fact, the preponderance of the evidence suggests just the opposite; and the case for this conclusion is even stronger than what has been presented to this point in time, as outlined in the following paragraphs.

In the opening sentence of their summary, DeGaetano and Allen state that "trends in the occurrence of maximum and minimum temperatures greater than the 90th, 95th, and 99th percentile across the United States are strongly influenced by urbanization."  With respect to daily warm minimum temperatures, for example, the slope of the regression line fit to the data of a plot of the annual number of 95th percentile exceedences vs. year over the period 1960-96 was found to be +0.09 exceedences per year for rural stations, +0.16 for suburban stations, and +0.26 for urban stations, making the rate of increase in extreme warm minimum temperature 95th percentile exceedences at urban stations nearly three times greater than the rate of increase at rural stations less affected by growing urban heat islands.  Likewise, the rate of increase in the annual number of daily maximum temperature 95th percentile exceedences per year over the same time period was found to be 50% greater at urban stations than it was at rural stations.  Yet in spite of this vast uncorrected-for-bias, when computed over the much longer 1930-96 period, 70% of all stations still exhibited "decreasing high extreme maximum temperature trends."

Other strong evidence for a spurious urban-heat-island-induced component of warming in even the best of surface air temperature records in the United States has been provided by Changnon (1999), who used a series of measurements of soil temperatures obtained in a totally rural setting in central Illinois between 1889 and 1952 and a contemporary series of air temperature measurements made in an adjacent growing community, as well as similar data obtained from other nearby small towns, to evaluate the magnitude of unsuspected heat island effects that may be present in small towns and cities that are typically assumed to be free of urban-induced warming.  The data obtained in the totally rural setting revealed the existence of a 0.4C temperature increase from the decade of 1901-1910 to that of 1941-1950.  This warming is 0.2C less than the 0.6C warming determined for the same period from data of the U.S. Historical Climatology Network, which is supposedly corrected for urban heating effects.  It is also 0.2C less than the 0.6C warming determined for this time period by eleven benchmark stations in Illinois with the highest quality long-term temperature data, all of which are located in communities with populations of less than 6,000 people as of 1990.  And it is 0.17C less than the 0.57C warming derived from data obtained from the three benchmark stations closest to the site of the soil temperature measurements and with populations of less than 2,000 people.

Changnon says his findings suggest that "both sets of surface air temperature data for Illinois believed to have the best data quality with little or no urban effects may contain urban influences causing increases of 0.2C from 1901 to 1950."  He further notes - in a grand understatement - that "this could be significant because the IPCC (1995) indicated that the global mean temperature increased 0.3C from 1890 to 1950."  Therefore, until the challenge of very-small-town heat island effects is resolved, the climate alarmists' "unprecedented" warming of the past century cannot be accepted at face value.  In all likelihood, it is artificially inflated, and perhaps severely so.

Another approach to the problem of growing urbanization and associated land use changes was recently developed by Kalnay and Cai (2003), who used differences between trends in observed surface air temperatures in the conterminous United States and corresponding trends in a reconstruction of surface temperatures derived from a reanalysis of global weather over the past 50 years -- the NCEP-NCAR 50-year Reanalysis (NNR) project -- which is insensitive to surface observations, to estimate the impact of land-use changes on surface warming.  This reconstruction of surface air temperature, it must be emphasized, does not employ surface observations.  Rather, it uses atmospheric vertical soundings derived from both satellites and balloons, so that surface temperatures are estimated from atmospheric values.

What did the two scientists discover when they applied this new approach to evaluating surface air temperature trends?  In contrast to what is suggested by surface observations, they say they could find "no statistically significant difference in the NNR estimation of urban and rural station trends," which suggested to them that "we could attribute the differences between monthly or annually averaged surface-temperature trends derived from observations and from the NNR primarily to urbanization and other changes in land use."

Kalnay and Cai went on to evaluate those differences for the city of Baltimore, Maryland, where they found that over the past 50 years there was "a growing trend in the difference between the surface observations and NNR," which increased to 1.4C during the 1990s.  As described above, the two scientists attributed this difference "to urbanization and other surface changes that do not affect the NNR," which is assumed to be a pure and unadulterated manifestation of the natural climatic state of earth's atmosphere in each particular part of the world where it is derived.

Applying this technique to the entire coterminous Unites States, where surface air temperature data imply a warming of 0.088C per decade over the past 50 years, Kalnay and Cai calculated a mean NNR-derived warming of only 0.061C per decade, demonstrating thereby that the surface air temperature data yield a century-long warming trend that is 0.27C too high.  This real-world observation, if applicable to the rest of the planet, suggests that the "unprecedented" global warming of the past century, which has been derived from the surface air temperature record, is significantly inflated.  From the Global Historical Climatology Network data base and the data base of Jones et al., for example (both of which can be accessed via our World Temperatures link), we calculate the mean global warming of the 20th century to have been 0.69 and 0.66C, respectively, for a mean global warming of 0.675C.  Now, however, it appears that that figure should be reduced to something on the order of 0.4C, or 40% less than what climate alarmists have long claimed it to be.

On the basis of these observations, we can only conclude that the warming of the past century or so was nothing more nor less than the natural recovery of the earth from the global chill of the Little Ice Age, and that as suggested by the comprehensive literature review of Soon et al. (2003), the planet is not nearly as warm currently as it was about a thousand years ago during the peak warmth of the Medieval Warm Period, when, of course, the atmosphere's CO2 concentration was about 100 ppm less than it is today.

In yet another approach to detecting spurious signals in the U.S. Historical Climatology Network (USHCN) database, Balling and Roy (2004) employed the concept of spatial entropy to estimate the degree of disorder in the pattern of temperature change across the 1221 USHCN stations over the period 1951-2000, examining the relationship between a station's entropy and the magnitude of the linear temperature change at that station.  They found that "spatial entropy levels are significantly and positively related to the observed temperature trends," which suggests, in their words, that "stations most unlike their neighbors in terms of temperature change tend to have a higher temperature trend than their neighbors."  In fact, they found that the warming trend of the highest of the seven spatial entropy classes they encountered was more than double that of the lowest spatial entropy class.  Overall, they report a +0.06C temperature increase over the period of their study for each unit increase in spatial entropy, where the average temperature increase for all stations over the period 1951-2000 registered +0.26C.  These results, in their words, "suggest that the USHCN contains some questionable warming signals at some stations, despite the many attempts to quantitatively control for these contaminants."

Equally unique and enlightening was the study of Maul and Davis (2001), who analyzed air and seawater temperature data obtained over the past century at sites of primary tide gauges of the U.S. Coast and Geodetic Survey, calculating trends for the 14 longest records.  This work yielded a mean century-long seawater warming of 0.74C for the 14 sites.  The result for Boston (a 100-year warming of 3.6C), however, seemed anomalous in the extreme, being almost seven times greater than the mean of the other 13 stations (0.52C); and, hence, it should probably be rejected, especially in light of the additional fact that Boston was the only site where seawater temperature increased faster than air temperature.

Now a warming of 0.52C does not seem extraordinary for the 20th century, when the planet experienced a significant portion of its recovery from the global chill of the Little Ice Age.  Nevertheless, Maul and Davis note that "each site has experienced significant population growth in the last 100 years, and ... with the increase in maritime traffic and discharge of wastewater one would expect water temperatures to rise," which implies that some unspecified portion of the 0.52C warming is due to a maritime analogue of the urban heat island, making the true century-scale non-urban-influenced increase in coastal seawater temperature probably significantly less than 0.5C.  They also note that "on the time scales investigated herein, one would expect the water temperatures to equilibrate to the air," which suggests that the true 100-year warming of USA coastal near-surface air temperature is something significantly less than 0.5C.

In light of the findings of the many studies that reveal the existence of significant urban warming biases in even the most pristine of the USHCN data series, it is clear that in the conterminous United States there has been very little warming over the past century (even neglecting to account for these urban warming biases) and likely no net increase in true background near-surface air temperature since the 1930s or even earlier.  In light of these facts, one could logically expect that data from other parts of the world are similarly biased, and that the true warming of the planet over the past century or so may be much less than what is typically claimed by climate alarmists, who in their zeal to demonize anthropogenic CO2 emissions claim that the likely natural warming that followed on the heels of the Little Ice Age was unprecedented over the past two millennia.

Balling Jr., R.C. and Roy, S.S.  2004.  A spatial entropy analysis of temperature trends in the United States.  Geophysical Research Letters 31: 10./1029/2004GL019630.

Changnon, S.A.  1999.  A rare long record of deep soil temperatures defines temporal temperature changes and an urban heat island.  Climatic Change 42: 531-538.

Chivas, A.R., DeDeckker, P. and Shelley, J.M.G.  1986.  Magnesium content of non-marine ostracod shells: a new palaeosalinometer and palaeothermometer.  Palaeogeography, Palaeoclimatology, Palaeoecology 54: 43-61.

Cronin, T.M., Dwyer, G.S., Kamiya, T., Schwede and Willard, D.A.  2003.  Medieval Warm Period, Little Ice Age and 20th century temperature variability from Chesapeake Bay.  Global and Planetary Change 36: 17-29.

DeGaetano, A.T. and Allen, R.J.  2002a.  A homogenized historical temperature extreme dataset for the United States.  Journal of Atmospheric and Oceanic Technology 19: 1267-1284.

DeGaetano, A.T. and Allen, R.J.  2002b.  Trends in twentieth-century temperature extremes across the United States.  Journal of Climate 15: 3188-3205.

Henderson, K.G. and Muller, R.A.  1997.  Extreme temperature days in the south-central United States.  Climate Research 8: 151-162.

Intergovernmental Panel on Climate Change.  1995.  Climate Change 1995, The Science of Climate Change.  Cambridge University Press, Cambridge, U.K.

Iskenderian, H. and Rosen, R.D.  2000.  Low-frequency signals in midtropospheric submonthly temperature variance.  Journal of Climate 13: 2323-2333.

Kalnay, E. and Cai, M.  2003.  Impact of urbanization and land-use change on climate.  Nature 423: 528-531.

Klasner, F.L. and Fagre, D.B.  2002.  A half century of change in alpine treeline patterns at Glacier National Park, Montana, U.S.A.  Arctic, Antarctic, and Alpine Research 34: 49-56.

Maul, G.A. and Davis, A.M.  2001.  Seawater temperature trends at USA tide gauge sites.  Geophysical Research Letters 28: 3935-3937.

Mitchell, M. and Kienholz, J.  1997.  A climatological analysis of the Koppen Dfa/Dfb boundary in eastern North America, 1901-1990.  Ohio Journal of Science 97: 53-58.

Schwing, F. and Moore, C.  2000.  A year without summer for California, or a harbinger of a climate shift.  EOS, Transactions, American Geophysical Union 81: 301,304-305.

Soon, W, Baliunas, S., Idso, C., Idso, S. and Legates, D.R.  2003.  Reconstructing climatic and environmental changes of the past 1000 years: A reappraisal.  Energy & Environment 14: 233-296.

Suckling, P.W. and Mitchell, M.D.  2000.  Variation of the Koppen C/D climate boundary in the central United States during the 20th century.  Physical Geography 21: 38-45.

Vega, A.J., Sui, C.-H. and Lau, K.-M.  1998.  Interannual to interdecadal variations of the regionalized surface climate of the United States and relationships to generalized flow parameters.  Physical Geography 19: 271-291.

Last updated 31 August 2005