UCAR's University Navstar Consortium is now attacking the PWV problem by way of the Global Positioning System (GPS) satellite network. In the five years since it joined UCAR, UNAVCO has wrung more and more atmospheric data out of the GPS signals used worldwide for nonmeteorological purposes. This time, they are intercepting signals with ground-based receivers across the southern Plains and using them to extract water-vapor information.
Chris Rocken, head of UNAVCO's research group, is leading the PWV project. He's been working with Russell Chadwick, chief of the demonstration division of NOAA's Forecast Systems Laboratory. Early in 1995, NOAA began deploying GPS receivers at sites across Colorado, Kansas, Oklahoma, and New Mexico where wind profilers were in place. These receivers have served as the test bed for the PWV technique, which performed well in extensive tests during the past year (see graph on page 2). It is soon to be studied further, perhaps through an expanded network of receivers.
The root of the new technique is the venerable receiver used for years by UNAVCO to detect tiny seismic shifts by collecting signals from GPS transmitters fixed in space. "To measure these tectonic signals accurately," notes Chris, "you need to measure the signal delay caused by the atmosphere." It is this delay that holds information on water vapor and other atmospheric qualities such as pressure, temperature, and density.
Under typical conditions, the atmosphere brakes a GPS signal coming to earth from a satellite overhead so that it falls about 250 centimeters short at the time when it would normally end up reaching the surface. All but an average of 30 cm of this is what researchers call a dry delay; it's related only to temperature and pressure and is correctible by knowing these quantities at ground level. The rest is the wet delay. Wildly variable, it can range from 0 to 60 cm depending on the amount of moisture in the atmosphere along the signal's path.
For his doctoral thesis at the University of Colorado at Boulder (CU), Chris examined how the GPS signals could be used to calculate wet delay, with the goal of removing it to improve GPS accuracy. At the same time, he recalls, "Mike Bevis and Steve Businger at North Carolina State University were saying 'Let's use this wet-delay information directly.' " The delay indicates how much water vapor is above the receiver, and that vapor is directly linked to cloud development and rainfall.
The next step was to compare the wet-delay technique with data from radiometers--upward-pointing instruments that measure the microwave emissions from water vapor overhead. Although radiometers are a pioneering tool in PWV measurement, they are limited for routine monitoring because they are hindered by thick clouds and virtually useless in rain. In contrast, GPS sensors can function equally well in rain or shine.
Businger and Chris joined other scientists at universities and UNAVCO in 1993 for a one-month comparison of data from six ground-based receivers and from radiometers stationed at three of the sites. The results were encouraging. Agreement in PWV estimates between the GPS and radiometer was on the order of 1 to 2 mm, compared to typical PWV totals in the southern Plains of 1 to 3 cm.
The network of receivers was bolstered to a total of nine in 1996 for more extensive testing. Meanwhile, Chris and his colleagues worked on some of the trickier problems that remained. "One of our challenges for collecting this data in real time is to know the real-time positions of the GPS transmitters in space," he says. Each transmitter is deployed about 20,000 km above earth, moving in a 12-hour orbit at more than 14,000 km/hr. To pin down PWV, Chris needs to know the precise location of the transmitter at any moment to within 50 cm.
That's even more challenging than it sounds, he says. "To get the orbits, your models are anything but simple." For instance, programmers must account for the color and shape of the satellite, because these influence absorption of solar radiation and the resulting thermal effects on the satellite's path. Also, variable solar radiation pressure, caused by the stream of photons and particles emitted by the sun, must be estimated for each of the GPS satellites. Even the impact of relativity has to be accounted for (see sidebar below).
The task of calculating after-the-fact satellite orbits is made easier by the International GPS Service, an international partnership based at the Jet Propulsion Laboratory. IGS publishes GPS orbits to within 5 cm. However, these calcuations aren't available until several days after a given orbit is complete.
Using the best orbit predictions and supplementary data available, UNAVCO's Theresa Van Hove has been poring over the PWV measurements taken across the plains during 1995-96. Of particular interest is the springtime tango of moist and dry air masses across the core of the observing network in northern Oklahoma and southern Kansas. John Braun has compiled animations that show Oklahoma's perennial dry line oscillating back and forth, triggering severe storms during its sojourns to the east. "We're seeing some really steep moisture gradients right in the heart of the network," John says.
The new GPS technique has several pluses in its favor. Aside from being impervious to rain, the ground-based receivers are unlikely to trigger the public nervousness that Doppler radars and other large, bulky weather instruments sometimes induce. The receiving element of each antenna is only a few centimeters across, mounted on a half-meter-square plate that helps to limit signal reflection problems. "It's a passive instrument," says Chris. "There are no concerns with noise, radiation, or the like. You could put it on your roof."
With more than two dozen GPS transmitters beaming signals from space, the raw material for PWV measurement is plentiful. Every half hour, each receiver in the test network collects over 400 measurements, which are then pooled into a single average to reduce random errors. Improved receiver antennas may soon reduce the need for averaging through a technique called slant water vapor measurement, using the signals that arrive at various angles.
Horizontal resolution is limited only by the number of ground-based receivers, each of which costs UNAVCO about $15,000 and uses about 10 watts of power. The potential of a superdense receiver network is being explored in Japan, where 650 receivers are now operating for earthquake research and 250 more are scheduled to arrive in the next year. UNAVCO is working on analysis of these data and is collaborating with Japanese scientists to explore the use of the network for PWV estimation.
More work on data processing is needed for real-time application of the new technique, says Chris. As that progresses, "We'll be waiting for meteorologists to start using the data." He notes that MMM senior scientist Bill Kuo and colleagues already have found that GPS-derived meteorological data, when assimilated into mesoscale computer models, can lead to significant improvements in short-term precipitation forecasting.
The new PWV technique may also support the GPS/Meteorology sensing program. Launched last year, GPS/MET uses a space-borne receiver to intercept and interpet signals slicing into and out of the atmosphere. (See the April 1995 Staff Notes.) Though it's proven successful at higher-altitude measurements, GPS/MET still has trouble directly separating the effects of water vapor and air density in the moisture-laden atmosphere below 15 km. According to Chris, the new PWV technique could prove handy as a complement to GPS/MET, with the combined systems filling in each other's gaps.
It could be that UNAVCO and collaborating scientists have gathered only the first nuggets from a rich vein of meteorological gold hidden in the bedrock of existing GPS technology. As Chris puts it, "We're leveraging a $10 billion investment in GPS by the U.S. taxpayer." BH
Other issues of Staff Notes