Water managers are increasingly inclined to fully answer this question: where does all the water go when snow melts?

The melting of last winter's huge snowpack has delivered a runoff in the Gunnison Basin of around 1 million acre-feet, which is about 35 percent above the long-time average.

But that was about 10 percent less than expected by the forecasters at the Colorado Basin River Forecast Center (CBRFC) in Salt Lake City and by the Bureau of Reclamation (BOR), managers of the Taylor and Aspinall Unit Reservoirs (largest of which is Blue Mesa).

That forecast led the BOR to release extra water from all the Gunnison Basin reservoirs in the early spring, to "make a hole" for the anticipated water -- to prevent flooding down river and the spilling of water that might otherwise go through turbines and create electricity. But now, there is not enough water coming off the mountains to fill those "holes."

Blue Mesa Reservoir will be about seven feet short of filling. Taylor was four feet short of filling and releases for the rafting industry below the dam have been cut to maintain both storage in the reservoir and a balance with flatwater recreation above the dam.

The forecasting challenge can also go the other way. Steve Glazer, water analyst for High Country Citizens' Alliance, noted that three years ago the runoff came down larger than forecast, which set the BOR scrambling to avoid spilling from the Aspinall reservoirs.

Other basins in the Southern Rockies have had similar experiences in recent years, and water managers and scientists are trying to figure out why, and what might be done to make the forecasting more accurate.

Dan Crabtree, who oversees operation of the Aspinall Unit and Taylor dams for the BOR, ran down the set of variables that might affect the runoff, and suggested that the forecasters are doing well to "come within 10 percent."

But as the cities of the Intermountain West continue to grow around a finite (and possibly shrinking) water supply, and as intensively used and reused as water is today, a 10 percent shortfall from annual predictions can create major problems for water managers.

The recent proposed settlement for a Black Canyon National Park water right is a case in point. There is no unappropriated water in the Gunnison River to allocate for the spring flows needed to sustain the mandated natural condition of the park. Its needs can only be met through complicated changes in the way water is released from the Aspinall Unit reservoirs.

But the amount of water the Black Canyon will get will be determined by the April 1 runoff forecast. Inaccuracies in that forecast, one way or the other, will have implications for other Upper Colorado users that will resonate through the rest of the water year.

The forecasting for the entire Upper Colorado River Basin is done from CBRFC headquarters in Salt Lake City, using an array of remote SNOTEL (Snowpack Telemetry) data-collection and transmission sites throughout the mountain region. The sites operate unattended, powered by solar-charged batteries, broadcasting data daily -- and sometimes more frequently. The federal Natural Resource Conservation Service installs and maintains the sites.

Colorado has 102 active SNOTEL sites, with 13 in the mountains around the Gunnison River Basin. Each SNOTEL site costs in the range of $20,000.

Standard SNOTEL sites are equipped to measure snowpack water content, cumulative precipitation and air temperature (daily minimums, maximums and averages). This would be sufficient information for forecasting if all the snow simply turned into water, and all that water ran down into the nearest stream.

Of course, it is not as simple as that.

How fast the snow melts, when it melts, the amount of wind during melting and the amount of moisture in the soil before the snow even falls are factors -- mostly unmeasured -- now recognized as important in determining how much of the water in a snowpack will actually make it into the usable water supply.

The abnormal dryness of the past decade has led snow forecasters and water managers to recognize that the amount of moisture in the soil under the snowpack is probably a significant factor in obtaining most of that elusive last 10 percent of accuracy.

The first two months of the current water year (last October and November), for example, were very dry in the Gunnison Basin. It is now assumed that a significant amount of the snowpack's water went to recharging groundwater -- especially lower in the valley, which had a snowpack far above average on abnormally dry soil.

Instrumentation can be installed on SNOTEL sites to measure soil moisture content, with probes planted at two, eight and 20 inches. Three of the SNOTEL sites in the Upper Gunnison -- in Crested Butte, Taylor Park and above Lake City -- are so equipped. But the added equipment is expensive, and as John Scott, Gunnison's NRCS agent, noted, "There is so much ground to cover."

Scott pointed out that most of the SNOTEL sites are relatively high in the watersheds, while most of the driest soil is lower down. Summer rain, he observed, will green up the plants with roots near the surface, but almost a decade of dry years has left a "deep soil moisture deficit" that is more quickly impacted by a dry autumn.

Other factors influence runoff. A spring punctuated with spells of cooler than average weather slows the melt, and might leave the snowpack more vulnerable to sublimation -- which is snow being changed directly to water vapor, rather than water. Wind moving over a snowpack also increases sublimation; wind speed indicators can be added to SNOTEL sites -- at more expense.

Some factors influencing snowpack and runoff are totally beyond early predictability -- such as the wet spring of 1984, for example, when major late-spring storms contributed enough to an already substantial runoff to almost take out Glen Canyon Dam.

Adding more SNOTEL sites seems like the most logical route to help the forecasters. But how many more, at $20,000 apiece? And with what (also expensive) additional instrumentation so the CBRFC forecasters are not flying blind on some critical factors?

This Thursday, Aug. 7, the Upper Gunnison River Water Conservancy District is assembling a meeting of water managers and forecasters from the CBRFC, the Bureau of Reclamation, the NRCS, the Rocky Mountain Biological Laboratory (RMBL) and other entities interested in the runoff predicting challenge. According to UGRWCD Manager Frank Kugel, this meeting was initiated when RMBL announced that it has received a grant to install five new weather stations in the Upper Gunnison watershed.

The meeting is open to anyone interested, and will be at 2 p.m. in the County Commissioner meeting room in the Courthouse.