The Congressional Budget Office said yesterday that it could cost an additional $5 billion over the next five years to diversify the U.S. space program by replacing the shuttle Challenger and building more unmanned booster rockets to carry payloads into space.

A White House senior interagency group for space (SIG-Space) has tentatively decided to follow such a course but is reconsidering how many unmanned rockets are needed and how they are to be financed, informed sources said.

The price of the group's first proposal, which included about 20 more unmanned launch vehicles for the Pentagon and a new orbiter for NASA, was set at $5.6 billion, sources said. That was termed too high and the group is reconsidering its proposal, one source said.

William R. Graham, acting administrator of the National Aeronautics and Space Administration, told Congress Tuesday that it would cost $350 million to correct shuttle problems including those in the solid rocket boosters -- thought to be responsible for the explosion that destroyed Challenger Jan. 28 -- and $2.8 billion for a replacement orbiter.

A Navy salvage ship yesterday located a part of Challenger's solid rocket boosters in 650 feet of water about 32 miles off the Florida coast. The 4-by-5 foot piece of debris was described as part of a ring-like fitting for a strut that connects the booster to the shuttle's external fuel tank. It could be a crucial clue, because flames emerged near a similar fitting on the right booster about 14 seconds before the explosion.

In another development, the presidential commission investigating the accident said it is exploring independent testing of seals used to join segments of the booster to see how they react to cold. Last week, commission members questioned the reliability of tests by the booster manufacturer, Morton Thiokol.

The CBO study, while recognizing that cost estimates "may change significantly" depending on the findings of the Challenger investigation, said that "NASA cost estimates may be low."

Underlying the uncertainties in these administration and congressional studies is the realization, voiced in the CBO study, that the "Challenger accident probably will prompt a reconsideration of many aspects of U.S. space policy."

The CBO said all current planning involves the knowledge that if NASA were to lose one of its three remaining shuttles, "a two-orbiter fleet would be incapable of fulfilling even minimum national security needs, let alone civilian research or commercial demand."

Graham said Tuesday that NASA has developed a planning schedule that would halt shuttle flights until at least February 1987. Thereafter, the agency is looking at slowly accelerating the three orbiters' launch rates. A fourth orbiter would not be available before 1990, he said.

The White House plan -- to build new unmanned boosters or modify old ones to share payload-carrying duty with the shuttles -- is also time-consuming and costly. The first of 10 new large Air Force rockets, approved by Congress last year, will not be ready until 1989. That fleet will cost $2 billion, almost the same as a new shuttle.

Reopening production lines for old rocket boosters would be a long, expensive undertaking, sources said.

The CBO study said 21 shuttle flight opportunities could be lost this year and next under its estimate of NASA operations after the accident. When flights resume, it said, delayed national-security flights would take precedence in the first two years of limited operations and prevent NASA from carrying more than five full nondefense shuttle payloads in that period.

To remedy that situation, the White House group is considering shifting 18 Pentagon Navstar satellites, now scheduled to be launched from shuttles beginning early next year, to unmanned launch vehicles. Air Force officials are studying whether they can turn Titan II missiles, retired as nuclear-weapons carriers, into space-launch boosters by 1987 rather than 1988 as planned.

In any event, the Global Positioning System (GPS), as the Pentagon navigation satellite system is called, will not meet its planned 1988 operational date, sources said.

CBO based its $5 billion estimate on "returning the shuttle system to safe operation, procuring a fourth orbiter and establishing a backup [unmanned rocket booster] program." It also projects "a more conservative and probably more costly operating mode for the shuttle system."

The CBO study concludes that Congress may have to slow the pace of the space-station program if it is to contain the added costs resulting from the Challenger accident. Congressional sources said yesterday that the space-station program, already cut sharply by the administration, may be reduced further in line with CBO findings.