ACKNOWLEDGMENTS:
This work is in collaboration with Richard Engelen and Graeme Stephens, also from the Department of Atmospheric Science at Colorado State University, and Peter Glecker and Mike Wehner from the Program for Climate Model Diagnosis and Intercomparison (PCMDI).

OBJECTIVES

The objectives of the three experiments described below are to quantitatively address the problem of temporal and spatial samplings of infrared (IR) and microwave (MW) brightness temperatures simulated by general circulation models (GCMs), for accurate comparison against brightness temperatures retrieved by the NOAA-11 and NOAA-12 satellites. A detailed description of the project objectives and proposed sampling procedures can be found in the AMIP SubProject No. 3 homepage.

In the Exact Orbit Simulation, the monthly-averaged IR and MW brightness temperatures are obtained by simulating the actual ascending and descending nodes of NOAA-11 and NOAA-12 in the Colorado State University General Circulation Model (CSU GCM).

In the Idealized Orbit Simulation, the ascending and descending nodes of both satellites are represented as north-south oriented stripes centered over the actual orbits.

Finally, in the GMT Sampling Simulation, the IR and MW temperatures are computed at each model grid-point, and monthly averages are computed by sampling the GCM every 6 hours at 0, 6, 12, and 18 GMT.

RESULTS

1. Exact Orbit Simulation
2. Idealized Orbit Simulation
3. GMT Sampling Simulation

When comparing the IR (Channel 8) and MW (Channel 1) brightness temperatures simulated in the Idealized Orbit and GMT sampling simulations against those obtained with the Exact Orbit simulation, we found absolute differences in excess of 1K over land, especially over desert regions.

These biases result because of the difference in the diurnal sampling of the surface temperature to which the IR Channel 8 and MW Channel 1 are strongly sensitive. The magnitude of the sampling bias is dependent upon the sensitivity the IR and MW channels to the surface temperature which diurnal cycle is greater over land than oceans. The sampling bias of brightness temperatures is strongly reduced for IR and MW channels that are more sensitive to upper-tropospheric temperatures and water vapor than surface temperatures.

Our results show that, in order to compare IR and MW brightness temperatures simulated by GCMs against satellite retrievals, it is best to simulate the actual orbit of the satellite in the GCM. Because of different spatial and temporal scales used in GCMs, the simulation of the actual satellite orbit is model dependent.

Simulating the true satellite orbit helps reduce the systematic bias in the diurnal sampling of the surface temperature, limits the computation of the brightness temperatures to grid-boxes over which the satellite is actually flying, and avoids the computation of brightness temperatures over cloudy regions for which the radiative transfer model does not work.