NASA/GEWEX SRB Validation

Model results have been validated extensively (See reference list). Recent validations have included ground measurements obtained from the Baseline Surface Radiation Network (BSRN), the Swiss Federal Institute of Technology's Global Energy Balance Archive (GEBA) and NOAA's Climate Montitoring and Diagnostics Laboratory (CMDL). Results of monthly averaged SW and LW radiative fluxes show that generally flux errors are within 10 Wm^-2. Larger errors were found where there are larger uncertainties in the input data such as over snow/ice covered surfaces and where the site data did not represent the entire grid box. Larger errors in downward SW flux were also found over African and South American locations where aerosols from biomass burning are not accounted for in the SW model (Konzelman et al., 1995).

An assessment of the quality of the global fluxes was accomplished by comparisons with corresponding ground-measured fluxes from a number of sites of BSRN. Uncertainties associated with operational BSRN measurements during this period are believed to be about +/- 5 Wm^-2 (1.5%, Ellsworth Dutton, NOAA, BSRN Manager). Thus, mean bias results are within the uncertainty for BSRN measurements. Errors for individual 3-hourly values are subject to bias and random errors due to local meteorological conditions.

An assessment of the quality of the global fluxes was accomplished by comparisons with corresponding ground-measured fluxes from a number of sites of the Baseline Surface Radiation Network (BSRN). Uncertainties associated with operational BSRN measurements during this period are believed to be about +/- 5-20 Wm^-2 (Ellsworth Dutton, NOAA, BSRN Manager). This includes a possible thermal offset which would result in a systematic bias of up to 3% (personal communication, Rolf Philipona, World Radiation Center) depending on atmospheric humidity and cloudiness conditions. Errors for individual values are subject to bias and random errors due to local meteorological conditions. Thus, mean bias results are within the uncertainty for BSRN measurements.

There is a visible and common artifact in much of the data set period, due to a lack of coverage from geostationary satellites over an area centered on 70 degrees east longitude. This situation, commonly called the Indian Ocean gap, occurs for all of the July 1983 - June 1998 time period, except for April 1988 - March 1989, when data from the INSAT satellite is available to cover the gap. In July of 1998, Meteosat-5 was moved over the gap area, eliminating the gap. (See the full readme file for this data set for a more detailed discussion of the Indian Ocean Gap Artifact.)

The SRB algorithms rely heavily on radiances and cloud properties from ISCCP. Great care is taken by the ISCCP team to produce a well-calibrated, homogeneous dataset from dozens of satellites over the course of several decades. However, there are a few known issues and discontinuities in the ISCCP products which are likely to be reflected in SRB products.

ISCCP uses a reference afternoon polar orbiter as a calibration standard for the other satellites (geostationary and polar orbiting) in the constellation. The afternoon orbiters are subject to orbital drift and are replaced every few years. The dates of transition from one reference satellite to the next are particular dates where small discontinuities in the SRB products are possible. The transition dates are as follows:

February 1, 1985: NOAA-7 replaced by NOAA-9
November 1, 1988: NOAA-9 replaced by NOAA-11
September 30, 1994: NOAA-11 goes out of service
February 1, 1995: NOAA-14 goes into service

For the October 1994-January 1995 timeframe there was no reference orbiter available, so an interpolation between NOAA-11 and NOAA-14 is used. SRB results show noticeable anomalies in this period, some of which are likely artifacts of the calibration situation.

October 1, 2001: NOAA-14 replaced by NOAA-16; TOVS operational algorithm changed
January 1, 2006: NOAA-16 replaced by NOAA-18

From NOAA-16 onward, the visible calibration is bi-linear, which has led to some changes from the previous linear calibrations. The 2001 transition to NOAA-16 is accompanied by fairly strong radiance increases over ice, which has led to polar values of SRB surface albedo and cloud optical thickness which are probably anomalously high, and surface downward fluxes which may be too low. The NOAA-18 calibration appears to be raising overall reported visible radiances, especially reducing the frequency of very low radiance scenes, such as clear skies over ocean near the day-night terminator. The effect on SRB products from 2006 onward has been mainly to cause a jump in surface albedo over much of the planet. Shortwave surface downward fluxes are less affected.

Care should be taken when computing long term time series from SRB data, with special notice taken of known transition dates.