WITH heat waves roasting much of North America and Europe, it's tempting, once again, to think, "Here comes global warming." It's a temptation to be resisted.TV meteorologists are right to advise us that single weather events - or even a succession of several hot summers - don't reveal climatic trends. In fact, James B. Elsner of Florida State University at Tallahassee and Anastasios A. Tsonis of the University of Wisconsin-Milwaukee now show that even the best long-term temperature records are poor guides to Northern Hemisphere climatic trends. As explained in their paper in the July issue of the American Geophysical Union's journal, Geophysical Research Letters, they compared three data sets commonly used by climate analysts looking for hints of global warming. Each set tabulates the annual surface-air-temperature departures from normal. One set was developed at the National Aeronautics and Space Administration (NASA) Goddard Institute for Space Studies in New York, one at the Climate Research Unit of East Anglia in England, and the third was produced at the USSR's Hydrometeorological Scientific Research Center. The data sets were constructed independently, using a different averaging technique and a different observational database in each case. Many climatic-trend studies have used one or another of these data sets. As Elsner and Tsonis note, "The implicit assumption is that the chosen data set is a representative sample from ... the true Northern Hemisphere annual surface air temperature record." The two scientists then show by statistical analysis that this assumption is not true. The data sets all have long-term rising temperature trends. But these are not the same trend. The two analysts conclude that "at least two of the three data sets do not represent the true [temperature record]." Perhaps none of them do. They add that "the confidence of relying on any particular data set for assessing global climate change statistics is thus questioned." Earlier this year, for example, the British Meteorological Office and the Goddard Institute both cited 1990 as the warmest year on record globally. But is that record reliable? Scientists at the NASA Marshall Space Flight Center and the University of Alabama at Huntsville, using different data, ranked the hottest years in the past 10, in descending order, as 1987, 1988, 1983, and 1990. The basic point is that currently available climate data are not reliable enough to show, unambiguously, any long-term temperature trend. They certainly can't highlight any man-made perturbation due to pollution with heat-absorbing gases such as carbon dioxide. This is not the same as saying such pollution has not enhanced the atmosphere's heat-trapping greenhouse effect, as the American Meteorological Society pointed out last year. The society said present observations can't prove man-made warming has occurred. But it explained: "Observations reveal clearly increases in greenhouse gas concentrations that can be linked to human activities. If no other factors counter their influence, these increased concentrations will lead to global warming." This summer's heat doesn't show such warming has begun. But policymakers can't use that as an excuse to dodge the issue of whether and how to curb the greenhouse-gas pollution.