poster html version

Satellite measured radiances from a variety of instruments are regularly archived for climatological purposes. In any comparison between a climate model simulation at satellite observations it is important to compute radiances in a manner that is a reasonable approximation to the way the original satellite sampling is performed. BMRC has been a participant in studies to find a computationally efficient, but accurate, method of sampling calculated radiances that can be recommended to all participants in Phase II of the Atmospheric Model Intercomparison Project (AMIP-2).

Various sampling schemes have been proposed in AMIP-2 to produce TOVS simulations:

Figure 1. This diagram displays an LCT sampling orbit (pink) in comparison with an LECT orbits (blue) at the 0 GMT model time step for NOAA-11 with local equatorial crossing time of 13:30. Note that all the orbits and the local time frame will move westward by 15^o lon. each hour.

Sampling Method	Acronym	Description
Local Crossing Time	LCT	Sample grid boxes at the same local solar time as Local Crossing Time of the satellite
Local Equatorial Crossing Time	LECT	Sample grid boxes at the same local solar time as the satellite Local Crossing Time at the Equator.
Exact Model	EXACTMOD	Sample the whole model grid at every time step.
Greenwich Mean Time	GMT	Sample the whole model grid 4 times per day at 0, 6, 12 and 18 GMT.
Exact satellites	EXACTSAT	Sample only where the TOVS observations occurred according to the historical archives.

SIMULATED BRIGHTNESS TEMPERATURE

Figure 2a. January average of the LCT simulated brightness temperatures for 7 HIRS and 4 MSU channels for the ascending nodes corresponding to a single satellite (NOAA-11).

Figure 2b. January average of the LCT simulated brightness temperatures for 7 HIRS and 4 MSU channels for the descending nodes corresponding to a single satellite (NOAA-11).

DIFFERENCES DUE TO SAMPLING

The results in Figures 3-4 (below) reveal differences between the LCT sampling and the LECT sampling are due to diurnal temperature change over land.

Figure 3. Differences in the HIRS-8 and MSU-1 brightness temperatures between LECT and LCT sampling for NOAA-11.

Figure 4. The differences when an additional satellite (NOAA-12) is added.

Figure 5. Differences in HIRS-8 and MSU-1 brightness temperatures between the 6-hourly GMT sampling and the EXACTMOD simulations for NOAA-11 and NOAA-12. The figure reveals an east-west wave of wave number 4 over global land region. This is a result of the aliasing of the diurnally varying surface temperature by the 6hr sampling frequency.

Figure 6. Differences in HIRS-8 and MSU-1 brightness temperatures between the 6-hourly GMT sampling and the LCT sampling for NOAA-11 and NOAA-12. This result confirms that the east-west wave problem is due to the LECT sampling.

Our recommendation to the AMIP Science Panel was that sampling based on a simple calculation of the satellite local crossing time be used by all AMIP II modelling groups. However even in the case when modelling groups do not comply with this recommendation we have a readily applicable aliasing method for determining the magnitude of the aliasing errors.

Satellite Sampling

Simulation of TOVS Radiances With Local-Crossing-Time Sampling

For more information, please contact:

BMRC, Bureau of Meteorology Research Centre
Department of Environment and Heritage
Commonwealth of Australia

Satellite Sampling

Simulation of TOVS Radiances With Local-Crossing-Time Sampling

For more information, please contact:

BMRC, Bureau of Meteorology Research Centre Department of Environment and Heritage Commonwealth of Australia

BMRC, Bureau of Meteorology Research Centre
Department of Environment and Heritage
Commonwealth of Australia