Amid the car lots, freeways and fast food joints of Marlowe Heights in Prince George's County rises the augustly named World Weather Building - the center of a giant electronic web designed to trap the information that will tell us if it's going to snow next Thursday.

By satellite, computer, radar, teletype, facsimile machine, telephone and occasionally by foot. National Weather Service technicians amass a constant jumble of meteorological data night and day. They examine it, massage it, sculpt it, give it a human evaluation.

They take pages of printed numbers, arrows, squiggles and dots, translate them into humidity, temperature and wind direction and - presto - you get your friendly forecast for the next couple of days.

Granted, the forecast may not always be right, but weather service veterans claim they're right much more often than they're wrong.

Most recently, forecasters acknowledge they blew it when they predicted sleet and freezing rain for the Washington area on jan. 20, and half a foot of snow fell instead.

"The computer said we had only a 4 percent chance of snow," said Jerry LaRue, meteorologist in charge of the local forecast office in the World Weather Building. ". . . It was 40 degrees at 5,000 feet and even warmer further south, and with southerly winds, we called for only freezing rain and sleet."

What happened? "That storm coming up from the south had tremendous upward vertical motions, much more than we would normally expect," explained LaRue. "The storm raised the warm southerly air to a higher altitude which cooled it and caused the unexpected snow."

Despite the weather service's array of electronic feelers, probes, antennas and other gadgets constantly sweeping the skies, "you're never going to have a perfect forecast," says LaRue, ". . . It's a no-win situation."

Predicting snow for Washington is especially difficult, he said. "D.C. and other eastern cities are often right on the rain-snow line, and it's hard to know exactly which side of the line we'll be on."

Even so, LaRue said, the accuracy of local forecasts has generally improved over the years as the wheather service's monitoring instruments have been refined.

According to forecast office figures, the weather service increased its percentage of correct predictions of precipitation or nonprecipitation from 75-30 percent, in the 1945-1950 period, to 83-86 percent in the 1970s. Similarly, according to the figures, the weather service narrowed the margin of error in its daily maximum and minimum temperature predictions from 3.5 to 4.0 degrees in the 1945-1950 period to 3.1 to 3.4 degrees in more recent years.

Central to the forecast operation is an enormous computer at the U.S. Census Bureau complex in Suitland into which meteorological data from ground observation stations and airborne balloons is pumped continuously.

The computer juggles the almost countless combinations of temperature, humidity and wind values at both ground level and several elevations up to 60,000 feet across the nation and comes up with a set of probabilities for each local area, 40 percent chance of precipitation, a 50 percent chance of freezing temperatures.

A crew of skilled forecasters at the World Weather Building punches up the probability figures on computer terminals and further analyzes them in conjunction with satellite maps and local historical data.

The forcasters then draft a variety of forecasts - local, regional, two-day, 10-day, marine, aviation, river and agricultural - which are disseminated by recorded telephone message, weather radio broadcast and teletype transmission to news media and other outlets.

Supplementing the forecast effort is a constant stream of hemispheric photographs transmitted by a National Oceanic and Atmospheric Administration satellite suspended 22,000 miles in space over the equator.

A series photographs showing cloud formations and densities across eastern United States, the Gulf of Mexico and parts of the Atlantic Ocean, is taken at 30-minute intervals and then projected on a television screen in the forecast office in an accelerated time sequence so that forecasters can see within a matter of seconds the cloud movement over the area during the preceding several hours.

In addition, the office constantly monitors its weather radar installation at Patuxent Naval Air Station 40 miles south of Washington. The radar picture is transmitted to the forecast office by facsimile copier and shows patterns and intensities of rain and snow throughout the area.

LaRue says the radar is of limited use, however, shows little detail beyound 125 miles of Patuxent and sometimes fails to pick up low-lying snow or rain clouds.

The 29-employe forecast office, which operates 24 hours a day, has a number of other operations including:

Emergency communications center - gives updates on snow accumulations and other extreme conditions to local school and highway maintenance officials.

River district office - monitors levels of the Potomac, Shenandoah and Rappahannock rivers and issues flood alerts. Chief river forecaster Leo Harrison relies on a network of about 100 local observers - famers, retired teachers, assorted weather buffs - plus 40 river gauges that automatically teletype flow levels to his office in Marlowe Heights.

Air pollution forecast operation - disseminates data on potential air pollution conditions for the D.C.-Maryland-Virginia-Delaware area.

Feeding the forecast center's omnivorous computer with its basic diet of numbers and squiggles are National Weather observers at all large commercial airports.

At Washington National Airport, weather office chief Ed Sherry and his five-member staff work round the clock in the main terminal building just under the air control tower.

Monitoring a console of clikcing dials and meters with names like hygrothermometer, ceilometer and microbarograph aneroid barometer, the airport weather office keeps both the forecast office office and commercial airline pilots alerted to cloud conditions, visibility, wind speed and direction, temperature, dew point, humidity and air pressure.

Most meteorological conditions are recorded electronically, but occasionally the electronics fail or need verification, and Sherry and his crew resort to more traditional methods.

These include reading a simple old-fashioned mercury thermometer to get the current temperature and sending up a helium balloon and counting how long it takes to disappear to determine the cloud ceiling.

"The observer keeps a continuous watch," Sherry says. "When there's turbulence, he goes out on the roof frequently . . . He's checking sky conditions, measuring precipitation, checking for visibility, range, fog and the onset of any kind of (adverse) weather.

"When we start getting bad weather, the phones start ringing, and they just won't stop."