Assimilation of ground-based GPS observations in the Canadian Regional Analysis and Forecast System
Stephen R. Macpherson, MSC, Dorval, QC, Canada; and G. Deblonde and J. Aparicio
An accurate representation of the spatial and temporal variability of atmospheric humidity is critical to numerical weather prediction. Conventional humidity observations from radiosondes and satellites have limited usefulness in this respect. Observations of zenith tropospheric delay (ZTD) from networks of ground-based Global Positioning System (GPS) receivers are a relatively new source of atmospheric humidity information. Unlike radiometer or satellite measurements of water vapour, the accuracy of integrated water vapour derived from GPS ZTD is largely unaffected by the presence of precipitation.
The NOAA Forecast Systems Laboratory GPS network consists of over 300 sites equipped with dual-frequency GPS receivers and automated surface weather stations. Estimates of ZTD from the GPS data as well as surface pressure, temperature and relative humidity from the weather stations are received in near real time at the Canadian Meteorological Centre (CMC) since August 2004. The network covers the continental United States with relatively high spatial resolution, with several sites in Alaska, Hawaii, and Canada. To increase coverage over Canada, near real time ZTD estimates for a number of Canadian sites are produced by the Meteorological Research Branch since July 2005.
Monitoring of ZTD and surface meteorological observations is carried out in near real time through comparisons with 6-hour forecasts of the regional (15 km) version of the Canadian Global Environmental Multiscale (GEM) model. While overall agreement is good, significant biases in ZTD are observed at some locations. The causes of these biases are investigated, as well as methods to remove them. A data impact study is conducted with the CMC Regional Data Assimilation System (RDAS). The RDAS consists of three-dimensional variational (3D-Var) data assimilation with the regional GEM forecast model. The impact of adding GPS observations in the RDAS on 48-hour regional GEM forecasts is evaluated.
Extended Abstract (248K)
Session 2, Experiments Involving Observations, Real or Hypothetical: Data Impact Tests (Sensitivity of Forecasts to a Particular Source of Observations); Observing System Simulation Experiments (OSSEs) Part 2
Monday, 30 January 2006, 1:30 PM-5:30 PM, A405
Previous paper Next paper
Browse or search entire meeting
AMS Home Page