Pentagon strategists are recommending to Secretary of Defense Casper W. Weinberger that part of the nation's abandoned anti-ballistic-missile system be reactivated, reversing the policy of the last two presidents.

The proposal is to reopen the ABM site at Grand Forks, N.D., without breaking the ABM treaty the United States and Soviet Union signed in 1972 limiting the defensive systems of both countries.

If Weinberger and President Reagan accept the recommendation, the present American policy of relying solely on offensive rather than defensive missiles to deter nuclear war would be changed significantly.

Reopening Grand Forks, even within the terms of the treaty, would also signal that the Reagan administration was distancing itself from the agreement signed by Richard Nixon in the heyday of detente. The treaty comes up for formal review next April.

Finally, deploying missiles around the abandoned ABM site as recommended would fortify the president's doomsday option of launching intercontinental ballistic missiles (ICBMs) if he were told enemy warheads were on the way. This could prevent U.S. missiles from being destroyed on the ground.

Asked recently about the reopening of Grand Forks, Weinberger said he was "not leaning any way" on the proposal "but certainly considering it. But this shouldn't be taken as any new policy."

Seymour L. Zeiberg, deputy undersecretary of defense for space and strategic systems, confirmed that he and fellow strategists at the Pentagon have drafted a proposal to reopen Grand Forks. He has briefed some members of Congress in secret session on the idea.

Zeiberg's group would deploy 70 long-range and 30 short-range defensive missiles at the North Dakota site -- actually in the tiny prairie town of Nekoma more than 100 miles northwest of Grand Forks -- to protect the giant Air Force warning radar at Concrete, another dot on the rolling farmland.

The perimeter acquisition radar (PAR) at Concrete is the last electronic sentry between Soviet missile launch sites and the 1,000 Minuteman ICBMs spread around the northwestern United States.

PAR scans the skies for incoming Soviet warheads. If it detected any, the radar is powerful and precise enough to tell how many warheads were on the way and where they were headed, even down to the specific Minuteman sites.

By giving this radar defensive missiles to destroy the first enemy warheads headed for it, goes the argument for reopening Grand Forks, the radar could provide the president reliable, precise and up-to-the minute information on the size of the attack.

With this extra assurance that he was not responding to a false alarm, the president might feel more comfortable about exercising his option to launch land missiles before they were caught in their underground silos.

This is called "launch on warning" or "launch on attack" strategy. It horrifies some arms control specialists. They warn it would put a hair trigger on nuclear war. A nervous president, they contend, might bring doomsday to the world by responding to radar images that turned out to be a flock of ducks rather than a flock of Soviet warheads.

Zeiberg, in making the case for reopening Grand Forks, said the radar at Concrete is so vital that "it's worth protecting with a threshold defense" of improved ABMs and radars developed since the ABM treaty was signed in 1972.

Besides protecting the radar, Zeiberg said in a telephone interview, reopening Grand Forks would help focus the Pentagon's currently diverse ABM research efforts. The Army is taking the lead in ABM research and development.

The ABM treaty of 1972 allowed the United States and Soviet Union to deploy no more than 100 missiles at two places, their capitals and one ICBM site in each country. In 1974, the two countries agreed to limit themselves to one ABM site, either the capital or an ICBM area. The Soviets opted for ABM protection for Moscow, the United States for Grand Forks.

The Grand Forks ABM site at Nekoma went into operation on April 1, 1975, and was closed on Feb. 10, 1976, a life of less than a year. It was the only U.S. site to go operational. The Army said it had no separate cost figures for the Nekoma facility, adding they were lumped into the $5.7 billion for the whole Safeguard ABM program.

Nekoma, a crossroads town of 102 persons living mostly in white frame houses on the black, rolling farmland, boomed while the ABM facility there was under construction. The population shot up to 3,000 in 1971.

Today, the huge cement pyramid holding the radar on site, not the PAR at Concrete, still dominates the landscape. The pyramid, missile holes and other buildings at the ABM site at Nekoma are bearded with prarie grass and filling up with water. The Air Force is preparing to haul away some of the duplex homes built for Army people formerly stationed at the site. s

If Zeiberg and his Pentagon allies prevail, the Nekoma site will rise Phoenix-like from its ashes under a $3 billion to $3.5 billion rebuilding program. New missiles would be installed at and around the site, perhaps in the old holes for Safeguard's long-range Spartan and short-range Sprint missiles.

One problem with the old Safeguard ABM was that its radars might have been blinded by the nuclear explosions of the first missiles it sent up to stop incoming warheads, giving a second wave of warheads unimpeded entry.

The long-range ABM envisioned for Grand Forks would not create this blinding condition because it would rely on a planned midair collision with the warhead, not a nuclear blast, for its stopping power.

The same booster rocket used on the existing Trident I submarine missile would carry a folded umbrella into space where it would open and zoom toward the incoming warhead. The umbrella would have iron weights hanging from its ribs for more coverage and killing power. A heat seeker on the umbrella would home in on the enemy warhead.

On the basis of highly secret tests to date, the Pentagon is claiming pinpoint accuracy for the umbrella, thanks largely to advances in long-wave infra-red radar guidance.

The short-range, last-ditch defensive missile would be the nuclear-tipped Sprints left over from Safeguard days.

Another problem with Safeguard was the softness of its eyes -- the guiding radars. Pentagon ABM enthusiasts said the new radars are smaller and harder to destroy.

Given the urgency Reagan and Weinberger have put on righting what they consider the adverse strategic balance with the Soviet Union, the Grand Forks scheme would seem to have a fairly good chance of approval.

In the longer term, it could result in more emphasis on building defensive missiles rather than continuing to leave the nation naked to Soviet warheads on the theory that there was no way to stop them.