Data source:
 
   Atmospheric pressure at the Earth's surface from the model GEOS-IT
   developed at the Global Modelling Assimilation Office at NASA.

   The model has resolution over longitude 0.625 deg, over latitude 
   0.5 deg, has 72 levels from the surface to 1 Pa, at has time
   resolution 3 hours. Three parameters were extracted from the
   output of GEOS-IT model: DELP -- atmospheric pressure increment
   between to adjacent layers, Q -- specific humidity at middle of 
   the layer, and T -- air temperature at the middle of the layer
   are used for analysis. The vertical grid of the numerical weather
   model GEOS-IT is irregular, i.e. the height of a layer depends on
   longitude, latitude and time. Dependence of the atmospheric pressure 
   on geometric height for a given latitude, longitude, and time is 
   computed by solving hypsometric equation that follows from the 
   condition of hydrostatic equilibrium of the atmosphere and the gas 
   law of moist air that takes into account of air compressibility.

   The first step in data processing chain is to compute the geometric
   height of atmospheric pressure at middle of a level. Then atmospheric
   pressure is to re-gridded into the regular grid that starts from 
   -1000 meters below the sea level and ends at height 90 km. The grid
   points of the new grid that lay within the range of level heights 
   of the numeric weather models are computed by interpolation. The grid
   points of the new grid that lay beyond the range of level heights 
   of the numeric weather models are computed by extrapolation.
  
   The second step is to expand the 3D field of atmospheric pressure
   at a given time on a new regular grid into a tensor product of B-splines.

   Then the atmospheric pressure is computed for the Earth's surface
   defined by the digital elevation model GTOPO30 by computing the 
   coefficients of the B-spline. The 2D surface atmospheric pressure 
   is computed at a equi-area grid with resolution 10800 over latitude
   and 5400 over longitude (~3.7 km). The height of the grid point
   is computed from GTOPO30 model that have resolution 30" (~0.9km)
   by averaging among the GTOPO30 points that are land within the area
   of the cell. Thus, the surface atmospheric pressure is related to 
   the average height of a land area 3.7 x 3.7*cos(latitude) km.

   Then the model that includes the mean pressure and harmonic variations 
   at each grid cell at the following frequencies is computed using 
   least squares: 

   SA    1.990968752920D-07  ! Long period  annual
   SSA   3.982127698995D-07  ! Long period  semi-annual
   PI1   7.232384890619D-05  ! Diurnal
   P1    7.252294578148D-05  ! Diurnal
   S1    7.272206167609D-05  ! Diurnal 
   K1    7.292115855138D-05  ! Diurnal
   PSI1  7.312025542667D-05  ! Semi-diurnal
   2T2   1.450459105823D-04  ! Semi-diurnal
   T2    1.452450074576D-04  ! Semi-diurnal
   S2    1.454441043329D-04  ! Semi-diurnal
   R2    1.456432012082D-04  ! Semi-diurnal
   K2    1.458423171028D-04  ! Semi-diurnal
   U3    2.177679627494D-04  ! Ter-diurnal
   T3    2.179670596247D-04  ! Ter-diurnal
   S3    2.181661850283D-04  ! Ter-diurnal
   R3    2.183652533752D-04  ! Ter-diurnal
   K3    2.185643502505D-04  ! Ter-diurnal
   U4    2.904900529562D-04  ! Four-times-a-day
   T4    2.906891498303D-04  ! Four-times-a-day
   S4    2.908882467044D-04  ! Four-times-a-day

   The mean, sine, and cosine amplitudes of harmonic variations were
   computed for the time range [1998.0 -- 2024.95], i.e. 27 years.

   Finally, the contribution of the mean, diurnal, semi-diurnal,
   ter-diurnal, and four-times-a-day surface variations were removed
   from the surface atmospheric pressure field. Annual and semi-annual
   variations are kept with the data.

References:

   Petrov, L., The International Mass Loading Service, 
      Proceedings of the Reference Frames for Applications in 
      Geosciences Symposium, held in Luxembourg in October 2014, 2015. 
      DOI: 10.1007/1345_2015_218, http://arxiv.org/abs/1503.00191

   Petrov L., J.-P. Boy, "Study of the atmospheric
      pressure loading signal in VLBI observations", Journal of
      Geophysical Research, 10.1029/2003JB002500, vol. 109,
      No. B03405, 2004.

   Rienecker, M.M., Suarez M.J., Todling R., Bacmeister J., Takacs L., 
      Liu H.-C., Sienkiewicz W. M., Koster R.D., Gelaro R., Stajner I., 
      and Nielsen E., "The GEOS Data Assimilation System -- Documentation 
      of Versions 5.0.1, 5.1.0, and 5.2.0.", NASA/TM--2008--104606, 2008. 
      http://gmao.gsfc.nasa.gov/pubs/docs/tm27.pdf
