The most sophisticated studies yet of a nuclear war's effect on climate have largely confirmed the controversial theory that such a clash could plunge large regions of the world into a devastating period of freezing temperatures -- the so-called nuclear winter.

The new studies, reported here yesterday at a major scientific symposium on nuclear winter, indicate that although the period of cold might not be as prolonged as earlier studies predicted, a war in July could drop temperatures by 30 to 100 degrees Fahrenheit for a matter of weeks.

The studies also suggest that even a limited nuclear war -- confined, for example, to Europe -- could produce a "quick freeze" in the region as soon as two days after the exchange. Although the freeze might not last more than a few days, it could destroy agriculture for the entire season.

"We sustain the idea that nuclear winter is still a possibility," said Stephen H. Schneider of the National Center for Atmospheric Research in Boulder, Colo. "However, we don't see the kind of picture that has three meters of ice over the continent."

Three years ago the nuclear winter hypothesis emerged from studies that scientists have always warned were crude and subject to great uncertainty. They suggested that a nuclear war would blast so much dust into the atmosphere, along with millions of tons of smoke from burning cities, that the sky would be blackened, cutting off sunlight and allowing the land to cool.

In the ensuing public debate the elements of uncertainty were widely overlooked. Nobody really knew how much material would be blasted into the sky, for example, how high it would go or how long it would stay there.

Last December a panel of the National Academy of Sciences released a study largely confirming the nuclear winter scenario but emphasizing the uncertainties.

Today's reports, at a meeting sponsored by the academy and its sister body, the National Academy of Engineering, removed some of the uncertainties. They reported on the results of computer simulations of atmospheric behavior that include more factors than did previous simulations.

The scientists give the simulation program certain basic facts about the amount of particles in the air at a given altitude -- still strictly a guess. The program then moves the hypothetical smoke around according to prevailing wind patterns for the season and calculates the effect of lost solar heating on the ground below. The computer then displays global maps showing where the smoke is after a certain number of days and how ground temperatures have changed.

Scientists objected to the previous computer models because they omitted or estimated very crudely many factors. A key factor, for example, was the rate at which particles fell out of the air. Critics of the nuclear winter theory have suggested that rain and other processes would cause most of the particles to fall to Earth before they could affect the climate.

The most sophisticated model, in use at the Los Alamos National Laboratory in New Mexico, takes account of such factors and shows that, although much of the smoke rains out very quickly, the balance rises as it absorbs solar heat, gradually escaping the effects of ordinary weather.

Particles in these layers, said Robert C. Malone, who developed the model at Los Alamos, appear able to remain aloft for many months. Within one month the thicker clouds below, accounting for two-thirds of the particles, have fallen back to Earth. It is not known what ground effect the thinner, higher clouds may have.

Schneider's similar but somewhat less sophisticated model also has responded to the criticism that post-attack smoke would be not evenly distributed around the Earth but, rather, patchy, letting the sun through in places. The patches, he found, could cover a large enough area to drop temperatures below freezing in two days.