IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 39, NO. 2, FEBRUARY 2001 439 Tomography of the Lower Troposphere Using a Small Dense Network of GPS Receivers Alejandro Flores, Student Member, IEEE, Jordi Vilà-Guerau de Arellano, Lubomir P. Gradinarsky, and Antonio Rius Abstract—The application of tomographic techniques to the tro- posphere with GPS signals was demonstrated in previous work using data from the Kilauea permanent network, Hawaii. Local orography of the network considered there, however, played a key role in the resolution capabilities of the technique. Here we ex- plore the possibilities of tomographic reconstruction of the four-di- mensional (4–D) structure of water vapor using a very small net- work of global positioning satellite (GPS) receivers with virtually no height differences between the stations. The analyzed campaign consisted of seven GPS receivers located at the Onsala Space Ob- servatory, Onsala, Sweden, and was carried out in August 1998. Traditional meteorological data sources and tools such as the nu- merical weather model NCAR Mesoscale Model (MM5), satellite data from the National Oceanic and Atmospheric Administration (NOAA), Washington, DC, and data and analysis from the Euro- pean Center for Medium-Range Weather Forecasting (ECMWF), Reading, U.K., have been used to evaluate our results. A good agreement is found between GPS tomography and clas- sical methods, even in meteorological situations with complex ver- tical structure of water vapor. Index Terms—Global positioning satellite (GPS), meteorology, tomography, troposphere. I. INTRODUCTION T HE EFFECT of the neutral atmosphere in the global po- sitioning satellite (GPS) signal is an extra delay related to the neutral atmosphere refractivity as (1) (2) where 10 can be expressed in mm/km; slant path through the atmosphere; path length along the slant path; straight-line path length; Manuscript received October 18, 1999; revised May 2, 2000. This work was supported by EC Grant WAVEFRONT PL-952007 and the Comissionat per a Universitats i Recerca de la Generalitat de Catalunya, Barcelona, Spain. A. Flores and A. Ruis is with the Institut d’Estudis Espacials de Catalunya, Barcelona, Spain (e-mail: flores@ieec.fcr.es; rius@ieec.fcr.es). J. V.-G. de Arellano is with the Wageningen University, Wageningen, The Netherlands (e-mail: jvila@hp1.met.wau.nl). L. P. Gradinarsky is with the Onsala Space Observatory, Onsala, Sweden (e-mail: lbg@oso.chalmers.se). Publisher Item Identifier S 0196-2892(01)01169-X. total atmospheric pressure in mbar (1 mbar 1 hPa); water vapor pressure in mbar; atmospheric temperature in K [1], [2]; liquid water vapor in the atmosphere in g/m [3]. The latter term is generally neglected. The term is the bending term that is about 1 cm or less for elevations greater than 15 and in general will not be considered [4]. The total atmospheric delay can be split into two components: the hydrostatic delay, due to the dry gases in the troposphere and the nondipole component of water vapor, and the wet delay, due to the dipole component of water vapor [2]. The refractivity is also divided into hydrostatic and wet components, as shown in (3). The slant tropospheric delay ( ) may be modeled as a zenith component plus horizontal gradients and discriminated between hydrostatic and wet components [5] (3) where and hydrostatic and wet zenith delays; and satellite elevation and azimuth as seen from the station, respectively; horizontal unit vector defining the direction of the projection of the ray on the horizontal plane; and hydrostatic and wet delay gradients; and hydrostatic and wet mapping functions. Mapping functions currently in use in GPS data processing take into account different factors such as Earth’s curvature as well as seasonal changes. In particular, we used Niell mapping func- tions (see [6]). The delay gradient in (3) has units of excess path length and is defined as [5] (4) where horizontal gradient of the refractivity; horizontal displacement vector; height above the surface. 0196–2892/01$10.00 © 2001 IEEE