President Reagan's science adviser says that "a very promising" future concept for defending against missile attack involves using huge mirrors in space to absorb intense laser rays beamed from Earth stations and redirect them to destroy Soviet missiles soon after launching.

In an interview yesterday, George A. Keyworth, the atomic scientist who serves as Reagan's chief science adviser, stressed that such a system might be two decades away and that many technical questions would have to be resolved if it is to be made to work.

But he cited it as the sort of idea that could grow out of the president's call this week for an all-out research effort to determine if an effective defense against enemy missile attack could be developed. Such a development, Reagan said, could shift U.S. strategy from instant atomic retaliation to a more defensive posture.

The laser-mirror combination was the only one cited by Keyworth in the interview around which some potential operational concept had developed. He also pointed out that, of all the new technologies that possibly could be used in such futuristic defenses, lasers were the furthest along.

The mirrors, perhaps 100 feet in diameter, would be stored aboard space boosters and launched only upon warning that an enemy missile attack appeared imminent, Keyworth said.

The idea, he said, is to keep these space boosters ready with an instant "pop-up" launch capability so that the mirrors would not have to remain constantly in space, where they could be knocked out in advance of an attack by Soviet anti-satellite weapons.

To deal with a large Soviet missile attack, possibly hundreds of these mirrors would be needed, and hundreds of ground-based stations in which to generate the intense laser light beams, Keyworth said.

The high-energy laser beams would be aimed at the mirrors. Their beams would be allowed to spread somewhat so that the mirrors could absorb their intense heat and energy without burning up. Then the mirror would essentially refocus the beam, reviving its intensity, and aim it at individual missiles as they rose from their launch silos deep inside the Soviet Union.

Ground-based computer stations would tell the mirrors in space where to aim their beams, using information from data-gathering satellites that would sense the engine heat of the newly launched missiles and track them with radar. The United States already has such satellites, but vastly improved ones would be needed for such a defensive system.

The mirrors could be repositioned quickly to shift their aim from one target to another, in hopes of picking off the Soviet missiles some 6,000 miles away within minutes of their launching. This would be well before the Soviet missiles could release the many individual atomic warheads each carries.

Keyworth says it would also be necessary, using the same techniques, to pick off any Soviet missiles that got through the first attempt to destroy them, before they began diving to the U.S. mainland.

Generally it would take a missile about 30 minutes to fly from the Soviet Union to the United States, and the individual atomic warheads would be dispersed during the last few minutes of that flight. The times are shorter for missiles fired from submarines that are closer to U.S. shores.

The advantages of such a system, if it could ever be developed, Keyworth said, is that the biggest and most complex component--the laser beam generator--would be on the ground where it could be serviced and defended. The system also does not involve putting weapons into space, and the pop-up technique would reduce vulnerability to a Soviet pre-emptive attack on the system.

Keyworth stressed that there are many technical unknowns. He acknowledged that "we don't know how to build lasers today" with as much energy as would be needed for the anti-missile role.

One of the problems that have plagued lasers for years is how to transmit them through rain and atmospheric disturbances without weakening them. Requirements for handling massive amounts of electronic intelligence and rapidly re-aiming the mirrors also go far beyond today's capabilities.

But Keyworth said extraordinary advances in micro-processor technology have been made in recent years that might solve some of these problems. And, he added, "in most of these areas" of potentially promising anti-missile technology "we have a substantial edge" on the Soviets.

The president's proposal has generated considerable controversy in the scientific community.

Prof. Sidney Drell, a leading physicist who is deputy director of the linear accelerator center at Stanford University and a former White House defense consultant, said, "I see no prospect of deploying on the ground or in space an effective defense."