Small buildings that look like outhouses are disappearing from the banks of the nation's rivers, and concern is rising among some scientists that the loss will hamper their ability to monitor rising water levels.

The humble-looking structures house stream gauges, the main instrument for measuring water flows. Several years ago, 11,000 gauges dotted river banks across the nation, but that number has fallen to a little more than 9,000.

"Stream gauges are an endangered species," declared Emery Cleaves, director of the Maryland Geological Survey. In Maryland alone, the number has fallen from 95 in 1985 to 76 in 1996.

Knowing the flow of a particular river or stream may seem an abstract worry for many, but scientists say it has implications for everyone from bridge builders and flood forecasters to water quality managers and stream habitat restorers.

Engineers, for instance, rely on flow data to decide how large they need to construct bridges and culverts. "If the data arenot there," said Ed Doheny, of the U.S. Geological Survey's Baltimore office, "they have to overdesign it, play it real conservative, and maybe spend more money than they might have to if they had the data."

Over time, the gauges have fallen victim to tightening state, federal and local budgets. Sometimes, the results can be disastrous - a fatal flood in 1996 in Kentucky occurred downstream of where a gauge had been closed a few years earlier.

Gauges - which automatically monitor water levels every 15 minutes - typically cost $5,000 to $12,000 to build, and another $9,600 a year to operate in Maryland. Generally, the USGS picks up half of the cost of a gauging station, with a state or local agency picking up the rest.

But when budget decisions are made, out-of-sight sometimes translates to out-of-mind, and gauge stations have disappeared one after another.

Cleaves, who has been raising the alarm in both national and regional forums, is worried not only by the declining number of gauges, but by the way gauging stations are closed.

Decisions are made one-by-one, often by local or state governments, without regard to how a particular gauge fits into a regionwide monitoring effort.

As a result, Cleaves said, some closures threaten to leave whole areas in the dark about water. In Maryland, for example, more than half of the river gauges within the Coastal Plain have been closed.

Cleaves has advocated a comprehensive review of the gauge program to eliminate duplication, install gauges in areas where there are significant "data gaps," and secure adequate funding to maintain the system.

The oldest gauge in the Chesapeake watershed is along the Potomac River at Point of Rocks, MD. It dates to 1895. Most, though, are only a few decades old at most. But scientists say it is important to keep stations runningfor long periods of times to evaluate the effect of long-term climate patterns and - because of rapid development in the region - land use changes that can affect water flow.

"Streams are a natural expression of the landscape, and our landscape is changing," said Mitch Keiler, a watershed restoration project manager with the Alliance for the Chesapeake Bay.

As the amount of monitoring data decreases, certainty about all decisions that relate to water flow also decreases.

River gauges are used to determine things like 10-year and 100-year floods.

Cities, such as Baltimore, use stream gauge information to estimate water supplies. The data are even used for calculating the amount of pollution that enters the Chesapeake Bay.

As states come under pressure to develop Total Maximum Daily Loads - TMDLs - they will need the best available information about river flows. TMDLs are estimates of how much of a particular pollutant can be in a given stretch of river without violating water quality standards; knowing the volume of water is critical for the calculation.

The loss of data could also hamper the new emphasis placed on stream restoration efforts in the Bay watershed, which are aimed at improving habitat.

Long-term flow information is critical in determining how a reconstructed stream channel should be designed. If the channel is too small, sudden pulses of water will eat away the banks, causing erosion within the stream that results in an unstable, ever-changing channel unsuitable for many types of stream life.

If it's too big, the stream's energy - instead of being absorbed along the way - is simply transferred downstream where it will cause other problems, such as channel destabilization. "Stability is equal to habitat in the minds of most people in the stream restoration business," Keiler said.