On May 9, the Occoquan Reservoir was filled with 9.8 billion gallons of water, and surplus was spilling over the thick concrete dam just above the little fishing village of Occoquan.

Today the reservoir is not even a third full, despite a month and a half of voluntary conservation among the 600,000 Northern Virginians who depend on it. Each day there is new, lower shoreline. The top water intake valve, no longer usable, juts above the surface -a stark, emphatic example of just how serious Northern Virginia's water emergency has become and why Fairfax and Alexandria imposed severe restrictions on water use.

What happened to the Occoquan? Why is the reservoir, which the experts thought would be adequate at least for two more years and maybe three, so low that in the worst possible scenario of the forecasters, it could run out of water in December?

The answer most commonly given is the long, hot dry summer. The summer was long and hot, but it wasn't significantly drier than last year when there was no emergency at the reservoir.

A composite of five readings at stations in the middle of the Occoquan watershed shows that rainfall from April through August totaled 16.35 inches. For the same period in 1976, the total was 14.71 inches.At the Warrenton station, at the edge of the watershed. April-August rainfall totaled just under 14 inches. The total for the same period in 1976 was just under 15 inches.

So the answer is not as simple as summer drought. To find the answer the experts have reached further back -to the long, cold and -yes -dry winter.

But if the reservoir is full in May, why does it matter how much precipitation fell in the preceding winter?

It matters because of two factors that generally only concern hydrologists -soil moisture and water table. Last winter because of below-normal precipitation and the hard freeze that delayed normal snow melt, soil lost its moisture and the water table went down, in some cases falling to new lows.

In the spring, these factors were not crucial because there was enough heavy rainfall to run off into the tributaries of the Occoquan and then flow into the reservoir itself. That's why the water was spilling over the dam in early May.

But this summer, it appears, the factors were a crucial element in why the reservoir went down to record low levels. According to Floyd Eumpu, the Fairfax County Water Authority's director of construction and engineering, when soil moisture and the water table are low, even normal amounts of rainfall may, in effect, disappear.

Instead of becoming runoff that can make its way into the reservoir, the rainfall is absorbed into the ground, replacing lost soil moisture and raising the depleted water table.

That is what happened, Eumpu thinks, this summer. The comparative figures on stream flow -which is high when there is a lot of runoff and low when runoff isless -offer powerful supporting evidence.

For example, the flow in the Occoquan Creek, one of the reservoir's tributaries, averaged only about 11 cubic feet per second this June. The average for June, 1976, was 25 cubic feet per second. The Occoquan Creek's flow rate is an important indicator beaause the land it drains is largely undeveloped.

Eumpu said that when the ground is extremely dry as it was after the cold winter, one inch of rain throughout the watershed puts only about one tenth of a billion gallons of water in the reservoir. At the other extreme, when the ground is extremely wet, he said, the same amount of rain puts about two and half billion gallons of water in the reservoir. When the ground has a normal amount of moisture, between a half and one billion gallons of water can end up in the reservoir.

Eumpu's figures would explain why this year's spring and summer rainfall, even though it was about the same as last year's, resulted in far less water going into the reservoir than in 1976, when the ground wasn't nearly as dry.

When the water table is low, it causes another problem for the reservoir - it reduces something hydrologists call "base flow". That is the flow from the tributaries to the Occoquan that continues even when there is no rainfall. Base flow comes from the ground, but it can slow to a trickle, or even stop like a faucet tap shut tight - when the water table falls low enough.

That, too, has happened this year.

Thomas J. Grizzard, director of the Occoquan monitoring laboratory, has also pointed out that while this season's rainfall, taken as a total, is not much different from last year's, it fell in a scattered, rather than concentrated, pattern. This also served to reduce runoff, along with the extremely dry soil.

Though the Occoquan, like any reservoir, was designed to handle unusual as well as normal circumstances, the unusual phenomena of this year - a cold winter with little precipitation and an extremely hot summer with more scattered rainfall - simply over whelmed the fail-safe mechcnism.

The Occoquan was designed to meet present demand, with some cushion, all year round. Hydrologists refer to a reservoir's "safe yield". The Occoquan's safe yield is 65 million gallons daily. But as Grizzard says, "If you enter a cycle where the hydrology changes, then all that (estimate safe yield) can go out the windows."

While the reservoir continues to go down, the water authority's Eumpu sees the crisis easing by the end of October. But he cautions, "Things will get worse before they get better.

Eumpu bases his projection on WTOP weatherman (and Authority consultant) Gordon Barnes' forecast for a wetter than normal October. Eumpu figures that, based on his "runoff curve", one inch of rain, instead of producing only a fraction of a billion of gallons of reservoir water will start producing a billion.

"The situation is serious," Eumpu says, "but it's not a panic."

Fairfax County Executive Leonard Whorton and his water advisers are not as sanguine as Eumpu, though neither are they saying there is a panic situation. They point out that Eumpu's projections go only to the end of October. What about November, they ask, and what if Barnes' rainfall preditions prove too generous.

But even if the Occoquan survives this year's crisis, what if there is another cold winter, and this year's scenario is repeated?

Three billion gallons of capacity could be added to the Ocoquan if the dam wall was raised five feet. But such a project would probably take a couple of years to go through the planning, financing and construction cycle. And the proposed supplementary supply on the Potomac River is all several years away -even assuming it gets approval from the Army Corps of Engineers.

So the only option, if there is a repeat climatalogical performance next year, will be more conservation - dirtier cars and browner lawns.