Introduction
Water vapor is a highly variable atmospheric constituent. It is fundamental to the transfer of energy in the atmosphere and in the formation and propagation of weather. Yet water vapor remains one of the most poorly characterized meteorological parameters. Improved knowledge of the water vapor field is needed for a variety of atmospheric research applications and for improved weather forecasting. Methods for sensing precipitable water vapor (PWV) with the Global Positioning System (GPS) promise improvement in short-term forecasting. To use PWV in forecasting it must be available close to real-time. Cosmic's Ground Based (COSMIC Ground Based GPS) GPS Research Group is analyzing GPS data from the SuomiNet Network and the NOAA Forecast Systems Laboratory GPS network and GPS data from CORS DOT NDGPS and USCG/USACE DGPS sites that have co-located surface met. sensors.

Processing Description
At present we process the GPS data in 1-hour segments. Once every hour our process transfers via ftp 1 hour of raw GPS data from the NOAA Forecast Systems Laboratory data hub. This one hour segment is translated from raw data to RINEX format using TEQC. Binex data from SoumiNet is sent to COSMIC via the IDD/LDM system developed by UNIDATA. Hourly RINEX files from the CORS network are ftp'd from the NGS CORS data centers.
     One-hour files are processed, with the Bernese software developed at the University of Berne Astronomical Institute. For the estimation of PWV we constrain the station coordinates tightly to positions obtained from processing daily solutions for the networks. The normal equations (NEQ) file from the 1-hour analysis is stored. Then we stack the NEQ files from the last 12 1-hour solutions to obtain the equivalent of a 12-hour GPS solution. The main reason to apply this stacking technique, rather than re-processing many hours of data every hour is the significant savings in CPU time. This results in shorter latency with modest requirements in computing power. Plots of the latest results are available about 1-hour after real time, currently the biggest factor in getting some of the GPS data
     TEC analysis is done at zero-difference level with 1-hour phase files from all stations in the NOAA/FSL network. We compute an independent solution each hour and do not require continuity between the hourly TEC solutions.
      The software for real-time processing is operated by the Bernese Processing Engine (BPE), developed jointly by the University of Bern and the UCAR GPS Research Group. Processing is currently done with ultra rapid predicted GPS orbits from the IGS. These are formed by combining 2 hour orbit solutions + 24 hour predictions from major IGS orbit centers. The BPE operation for real-time GPS PWV processing was implemented by Teresa Van Hove, who is also responsible for the ongoing automated process. We are using ocean loading correction models developed and provided by M.S. Bos and H.-G. Scherneck at the Chalmers Centre for Astrophysics and Space Science. Please go to http://www.oso.chalmers.se/~loading/index.html for explanation of ocean tidal loading.

GPS Data
We are presently, on an experimental basis, processing data from about 60 SuomiNet Network sites, and about 40 Sites operated by the NOAA Forecast Systems Laboratory. There are also about 60 USCG/ACE and DOT DGPS sites that are part of the Continuously Operating Reference System (CORS) managed by NOAA's NGS. Also 4 Scign sites, Several LSU-LSRC sites, Several Tx-Dot sites, 1 site operated by Univ of Utah, and 1 USNO site. All of these sites are using precision survey quality dual-frequency GPS receivers and antennas. Most of the Suominet, FSL network and many of the USCG/ACEsites have co-located surface met packs, the other sites are using surface met data from nearby ASOS sites. The Suomi network and FSL network use Trimble^TM Receivers and Antennae; the USCG/ACE and DoT sites are using Ashtech^TM Receivers and antennae.

SuomiNet Special Projects
Default Suominet operations are 30 second GPS tracking and 3 minute met records; but we provide higher-rate tracking up to 1hz for special projects. If the local suominet host university or organization can handle the increased bandwidth we can support requests for higher rate data from any of the suominet sites.

A recent example is SG06, the SUNY, Stony Brook suominet site, which was operated at 1 hz to allow it to serve as a base station for tracking an air-borne lidar survey over Sagamore Hills and Fire Island National Parks.

This survey was conducted in a research collaboration between USGS, NASA and the National Park Service (NPS), using NASA's Experimental Advanced Airborne Research Lidar (EAARL), developed by C. Wayne Wright at the NASA Wallops Flight Facility, Wallops Island, VA.