HERE'S AN IDEA. Let's pave part of coastal Libya with asphalt so that sea air will be heated as it passes over and will rise high enough to dump its moisture on the parched deserts of the Sudan. Every acre of asphalt will provide 30 inches of rain a year for two acres inland.

Here's another idea. Let's drill a tunnel from New York to Los Angeles and run a train through it at 6,000 miles per hour so that we can make the coast-to-coast trip in about 45 minutes. The train would run in a vacuum and be driven by electromagnetic pulses.

How about this one? Most of the Soviet Union's Siberian rivers flow into the Artic Ocean, leaving little water for irrigation. Let's reverse the flow of one of those rivers and refill the dying Caspian Sea.

These projects and many others that sound just as hare-brained were proposed by serious, respected scientists. They have thoughtful defenders and are the subject of careful study by governments and in learned journals.

The project to reverse part of the flow of the Irtysh River is Siberia where it meets the Ob, for example, is under intensive study in the current Soviet five-year plan. Massive pumps, dams and locks would push the water uphill through a new canal, 35 feet deep and a football field wide, that would stretch 1,020 miles south to Kazakhstan. The water would irrigate an area larger than Europe which is now as arid as New Mexico.

"Many people say this is not possible, but they don't truly understand the idea," said V. M. Krakovets, irrigation counselor at the Soviet embassy in Washington.

No one has made any serious attempt to pave part of Libya since J. F. Black and B. L. Tarmy outlined the notion in the Journal of Applied Meteorology in 1963. The idea was that strips of asphalt 35 miles long and 7 to 10 miles wide along the coast would levitate enough moisture-laden air to make the southern deserts bloom. Not only has this plan never been written off, but the theory is the basis for some current study of the way shopping center asphalt islands affect U.S. weather patterns. They Laughed at Marconi

NO PROJECT proposal, no matter how wild, is rejected out of hand nowadays. Too many scientists wince to recall the ideas that were ridiculed when first offered but turned into earth satellites, space travel, Mars landings and color television, and nobody wants to be caught with a rash condemnation showing.

"I remember that jet planes, transistors, helicopters and a radio small enough to wear on your wrist were very far-out ideas in the '20s and '30s," said Frank Drake, director of the National Astronautics and Ionosphere Center at the Cornell University Research Institute. "If a thing is physically possible, and if it's cost-effective at some stage, then no matter how lunatic it seems, it often comes to pass."

That sequence has occurred so often in science that it is dogma, offering continual hope and encouragement to the fertile imaginations of the far-out thinkers.

"Yes, I guess I have always been on the fringe," allowed Robert M. Salter, father of the underground bullet train concept and senior physical scientist for advanced programs at the Rand Corp. think tank in Santa Monica, Calif.

The idea for the tunnel came to him, he said, during his work on the space program about 20 years ago.

"We were thinking about satellites and ballistic systems and how they could be used for transportation. It became apparent that if you could just replicate outer space in a tunnel underground, you could have your cake and eat it too."

The technical difficulties of creating and then maintaining a coast-to-coast vacuum tube, given earthquakes and rivers and a thousand other things, are mere matters of competent engineering, Salter said. He is more worried about the political problems of hooking up existing urban subway systems and raising the $250 billion he thinks the project will cost.

He has that much in common with most of the fringe thinkers: Their projects involve no scientific breakthroughs but only sophisticated engineering. The obstacles, they say, are political or personal or, in some cases, simple shortsightedness on the part of the public.

"Only a minority of projects make no sense for technical or scientific reasons, things like perpetual motion machines," observed Jerome Wiesner, president of Masschusetts Institute of Technology and science adviser to several presidents.

"Projects that violate no [scientific] laws are often rejected for cost and feasibility reasons that then change over time," Wiesner continued. Technology improves, costs decline and popular issues change, "and suddenly a project that once looked absurd now looks feasible."

A previously cockamamie idea now in the throes of becoming respectable is the notion of building massive solar energy collectors in outer space and beaming electricity back to earth. Peter Glaser, a consultant with Arthur D. Little Inc., proposed the solar satellite a decade ago, but that was before the oil embargo, when energy was still cheap. The idea, Glaser recalled recently, "elicited a polite smile and total disbelief."

Then the moon landing exhausted public interest in new blastoffs and the aerospace industry began to get hungry. Energy prices skyrocketed and the long-planned space shuttle needed something further to justify its existence. As is often the case in this process, other scientists joined the fray.

Princeton physicist Gerard K. O'Neill came up with the refinement of having colonies of workers live in space and build the satellites from moon and asteroid material instead of hauling the parts up from earth. A riveting speaker, O'Neill played the campuses and helped to popularize the basic idea. When last heard from, "SunSat" had a board of directors that read like a who's who of the aerospace industry and had won a $25 million study from Congress. Glaser estimates the first satellite would cost $60 billion, and many physicists expect to have it in about 15 years. What Kinds of Research?

A SURE SIGN that a project has merit, more than one scientist suggested, is that it wins the Golden Fleece award from Sen. William Proxmire (D-Wis.). The senator's boondoggle-of-the-month trophy recently went to NASA's Jet Propulsion Laboratory in Pasadena, Calif., for its plan to spend up to $15 million over seven years to try to find intelligent life in outer space.

The senator's guideline in these matters, an aide said, is "common sense . . . consider the average taxpayer who looks and says, 'My $2,000 a year goes for that?'" Ordinary citizens, he added, "understand the need for basic research but don't want things that clearly have a very small chance of any return at all."

Proxmire's idea of return on investment is not the same as the scientists'. "Look at nuclear fusion," said Robert Salkeld, a Californiaaerospace industry consultant. He referred to the effort to produce energy the way the sun does: forcing the nuclei of hydrogen atoms together under terrific heat and temperature until they fuse.

"It may be another 50-100 years before we solve that one, and we're spending $400 million a year. It's not nutty-fringy in one sense, because many capable scientists are working on it; but it's fringy in terms of what we're getting out of it now," Salkeld said.

The experts draw a careful line between far-out projects and far-out research. "Going to the moon was a trivial step from a scientific point of view, although it was significant technically," said National Science Foundation Director Richard C. Atkinson. "Telling scientists to forget basic research and work on things of practical value is the great danger of our time . . . if there had been a war on polio in the '20s and '30s, we'd have ended up engineering the most elegant and beautifully running iron lungs one could imagine."

Further out than Salkeld, further out than Glaser, even further out than the Russian river flowing uphill is William L. Gale, physicist with Bell Telephone Laboratories in Murray Hill, N.J.

Gale has left mere mortals behind and is alone out there in the stars, theorizing about the processes that will be needed after the solar system is colonized. He is working on ways to transform the molecules of distant galaxies into industrial raw materials and is concerned about the ultimate limits to growth.

"The physical resources of the solar system could support about a sextillion humans," he said. "At a 1 percent growth rate, we will achieve that in only 3,000 years . . . assuming massive space colonization, of course."

Of course. Asked whether no one ever called him a nut for worrying about that while people are already starving here on earth, he seemed to have difficulty grasping the question. "Well, people tend to worry about whether they'll feel claustrophobic on the colonies, things like that," he said. "It will be a major adjustment."