Overview

Copyright: Hans-Peter Plag.

A proposed IERS Reference Surface Pressure Field for ECMWF Operational and Reanalysis Data

Here, I give access to data sets and tools related to the air pressure reference surface, which can be used to computed the air pressure anomaly. However, I caution strongly to read this page in detail before using any of the products provided here. The reference pressure fields provided here are specific to the way they are calculated and each field refers to a specific meteorological data set and Earth topography. Any field should only be used with air pressure computed based on the same meteorological data set and Earth's topography using an algorithm consistent with the one used for the computation of the reference field. Due to significant differences in meteorological datasets and topopgraphy, the errors in the air pressure anomaly due to inconsistencies can be considerable and deviations of 20 HPa cannot excluded.

All reference pressure fields provided here are for air pressure at topographic height, not mean sea level. All fields are currently provided with a spatial resolution of 5 minutes, using the ETOPO5 topography as a basis.

Reference for ECMWF data

For the ECMWF data, surface pressure is computed from geopotential heights of isobars (see below in next section for more details). We use two different data set provided by ECMWF, namely the reanalysis data (ERA40) and operational data provided in near-real time. There is an offset between these two data set, and therefore, two different reference surfaces need to be used.

The reference pressure field that has to be used if the pressure is computed using the operational data is available with 5 minute spatial resolution as gzip file (42 MB). For a plot, see here. The reference is the average average pressure of the ECMWF operational data for the years 1999 and 2000. I have chosen this interval to get a reasonable average around the reference epoch of ITRF2000 and ITRF2005, which is in both cases 2000.0.

This choice will induce some deformation of the reference coordinates and I am currently computing this deformation. Alternatively, we could have used the pressure of 2000.0 as reference, which would have turned the pressure anomaly on 2000.0 to zero all over the globe. This would have introduced minimal deformation of the reference polyhedron on 2000.0. However, the air pressure loading effect is already filtered in the cause of the determination of ITRF with a very complex filter characteristic that is time-dependent. Thus, aiming for zero anomaly at the reference epoch may not be the best choice. It may be better to try to minimize the average anomaly over a time window close to the reference epoch, hence my choice.

The offset between reanalysis and operational observations is computed on the basis of the full year 2001. The differences are of the order of +-3 HPa (see plot). The reference surface for the reanalysis data which accounts for this offset is available with 5 minutes spatial resolution as gzip file (42 MB).

Interdecadal Variability of the Average Surface Pressure

• ERA40 Data Set:

  The average pressure field is available for the interval 1992.0 to 2002.0 as gzip compressed xyz file here (size approx. 40 MB). This surface is the ten-year average for the interval 1992.0-2002.0 (see also plot).

  The f95 routine era40_etopo5_geopot_ref.f90 is avaiable as a gzip compressed file. This routine returns the reference pressure for any geographical location. The routine can be used to reduce the spatial resolution of the pressure field. The program arp_grid.f90 creates a global grid of the pressure field with a predefined spatial resolution.

  The surface pressure field is the average of ten years with the surface pressure computed with a sampling interval of six hours. Topographic heights are taken from the ETOPO5 topography for the land surface and are set to zero for the ocean surface. For each six hour sample, the pressure at topographic height (determined from ETOPO5) is computed by linear interpolation of the logarithm of the geopotential heights given in the ERA 40 data set (see Plag et al., 2007). The ERA 40 grids have a spatial resolution of 2.5 degrees, and the geopotential heights are interpolated linearly to 5 minute resolution.

  Plots of the 1992-2001 surface and the differences between this surface and those for earlier 10-year windows are available here. The temporal variation between different ten-year intervals are of the order of -6 HPa to +3 HPa, with the largest air pressure decrease occuring around Antarctica.