FSL Workshop Investigates a Next Generation Weather Observing System

Leaders in the field of Global Positioning System (GPS) measurement science met in Boulder, Colorado on June 21, 2001 to discuss a new and potentially significant way to use GPS for atmospheric remote sensing. The all-day workshop was convened by the FSL GPS-Met Observing Systems Branch to assess the feasibility of making operational line-of-sight (or slant path) GPS signal delay measurements for us in weather forecasting and climate monitoring. Recent computer simulations by A.E. (Sandy) MacDonald and Yuanfu Xie at FSL indicate that it may be possible to retrieve the vertical distribution of moisture in the atmosphere using 3-dimensional variational analysis techniques. Information about the vertical distribution of moisture is highly desirable, but current GPS data acquisition and processing methods only provide measurements of vertically integrated (total column) water vapor.

Radio signals from the GPS satellites in Earth Orbit are refracted (slowed and bent) by changes in temperature, pressure, and water vapor in the atmosphere. This slowing and bending results in an apparent increase in the distance between the GPS satellites and a fixed receiver on the ground. The excess path length manifests itself as an error in the observed position of the GPS antenna that changes in time, mostly due to the temporal and spatial variability of water vapor along the paths of the signals. Of course other factors affect the accuracy of a GPS measurement including satellite and receiver clock errors, orbit errors, and multipath reflections. In practice, these other factors must be separated from the atmospherically induced errors in order to make a useful meteorological observation. The most common way to do this is through a process called "double-differencing." In double differencing, a new GPS observable is formed by calculating the differences between the carrier phase observations made by to two receivers simultaneously tracking the same pair of GPS satellites. In the process of doing this, all information about the signal delays along an individual line of sight are apparently lost. The question presented to the workshop participants was "Can slant-path signal delays be objectively and unambiguously measured or retrieved with GPS, and can the measurement be verified and quality controlled to make it useful in numerical weather prediction?"

The workshop began with a presentation by FSL Director Sandy MacDonald describing the experiments that led to an understanding of the potential importance of slant-path observations to NOAA. Next, FSL's Chief Scientist, Thomas Schlatter, discussed the need for information about the accuracy and error characteristics of any observation used in numerical weather prediction. This was followed by presentations about measurement techniques and results of experiments GPS experts from the University Corporation for Atmospheric Research; NASA Jet Propulsion Laboratory; Scripps Institution of Oceanography; Harvard Smithsonian Center for Astrophysics; and Massachusetts Institute of Technology. At the conclusion of the day, Seth Gutman of FSL led a discussion of possible ways to determine, verify, and monitor the quality of slant-path signal delay measurements under operational conditions.

There was agreement among the workshop participants that most of the unique information available from a GPS observation is contained in the zenith delay, the gradient in the zenith delay, and in the formed double-differences. Techniques using multiple receivers and satellites permit the unambiguous (and ultimately verifiable) retrieval of zenith-scaled signal delays and the gradients in the zenith delays. There may also be some higher-order (residual) information, possibly coming from moisture variability, but it is of unknown content and structure. Still more information may be available if observations are made with networks of closely spaced receivers.

Most of the participants agreed that it is not currently possible to make an unambiguous single slant-path signal delay measurement using only the information derived from a single GPS receiver and satellite. There was a difference of opinion, however, as to whether the methods currently used to estimate slant-path delay, slant-path water vapor, or vertical refractivity structure from formed double differences can provide a definitive solution with known (or determinable) error characteristics. Most participants believe that these estimates require additional information such as an assumed atmospheric structure to form a solution; information that must, in turn, be independently verified or quality controlled for operational use. The exception, as pointed out by researches from the Harvard Smithsonian Center for Astrophysics, is when observations are made at various elevations in mountainous terrain they contain direct information about the vertical refractivity structure. The ability to retrieve is information in flat terrain with the same level of confidence is uncertain.

The workshop produced three recommendations. First, one or more observing system simulation experiments should be conducted using the direct assimilation of simulated zenith delays, zenith delay gradients, and/or double-differences, rather than simulated slant-path delays or slant water vapor. The purpose of these experiments would be to see if direct assimilation of the former can resolve apparent ambiguities in existing slant-path GPS techniques, and produce similar results to those achieved by MacDonald and Xie using simulated slant-path measurements. Second, use double blind and triple-blind experiments with simulated data to determine if the vertical refractivity structure of the atmosphere can be retrieved using GPS techniques alone, with what accuracy, what under what circumstances. Finally, use the comprehensive data sets expected to be generated during the IHOP campaign in 2002 for testing and evaluation of the various techniques under real-world conditions.

Contact information
Name: Seth I Gutman
Tel: 303-497-7031