Quantifying the Magnitude of Anomalous Solar Absorption
Ackerman, T. P., University of Washington
General Circulation and Single Column Models/Parameterizations
Spurred by a series of articles published in 1995 claiming solar absorption in cloudy atmospheres far exceeded model predictions, Atmospheric Radiation Measurement (ARM) Program researchers at the Southern Great Plains (SGP) site in Oklahoma initiated a series of experiments to investigate the anomalies. Sponsored by the U.S. Department of Energy's Office of Biological and Environmental Research, ARM's Enhanced Shortwave Experiment (ARESE) began with a 6-week data collection period in late 1995 (which proved generally unfruitful due to prolonged clear weather), followed by a similar run (ARESE II) in spring 2000. This time, cooperative weather combined with the extensive instrumentation available at the SGP site provided a detailed, high-quality data set. Measurements and analysis of the comprehensive ARESE II experimental data were compared against simulations using two of the most sophisticated radiative transfer models available. Due to the overwhelming agreement between the ARESE II measurements and the newer models, ARM researchers concluded that while it is difficult to disprove the results of previous studies with the result of this current experiment, or to generalize from one cloud type to all cloud types, it is our opinion that the ARESE II results provide compelling evidence that it is time for the atmospheric science community to lay to rest these discussions of extreme solar absorption in cloudy atmospheric columns.
ARESE I ran from September 22 through November 1, 1995. Mostly clear weather for the duration of the experiment resulted in only 1 day of overcast skies to provide meaningful absorption data. To obtain useful data, ARM conducted ARESE II at the SGP during a 6-week span in February and March 2000. The experimental plan focused exclusively on extended stratus decks composed only of liquid water. The study used data from surface solar radiometers, cloud radar, the microwave radiometer, Raman lidar, micropulse lidar, and sonde releases. To measure fluxes above the clouds, a Twin Otter aircraft, carrying three different sets of upward and downward looking solar radiometers, flew a 2-hour daisy pattern over the Central Facility at a maximum altitude (about 7km). The Otter then typically flew a few albedo runs at low altitude (about 0.4km) on each day of the experiment (see figure 1). This daisy flight pattern provided a series of passes over the SGP Central Facility at different relative solar azimuths in order to minimize any directional bias that might be present in the mounting of the upward-looking radiometers. The experimental team treated each pass as an independent event.
The radiative transfer models used in the ARESE II experiment were: (1) RAPRAD (from Pacific Northwest National Laboratory), a δ2-stream discrete ordinate code with a correlated-k spectral integration, and (2) SBDART (from the University of California – Santa Barbara), which also uses a δ2-stream discrete ordinate approximation, but has a very different spectral integration scheme of considerably higher resolution than RAPRAD. The two models use similar numerical schemes, but treat gaseous absorption quite differently. ARM's experimental team analyzed measurements taken on 2 clear-sky days and 3 cloudy days and modeled the solar radiative transfer in each case with the RAPRAD and SBDART models.
On the 2 clear days, the calculated and measured column absorptions agree to better than 10 W m-2, which is about 10% of the total column absorption. Because both the model fluxes and the individual radiometer measurements were accurate to no better than 10 W m-2, the team concluded that the model and measurement were essentially in agreement. For the 3 cloudy days, the model calculations agreed very well with each other. On 2 of the 3 days, calculations agreed with the measurements to 20 W m-2 or less out of a total column absorption of more than 200 W m-2, which is again an agreement at better than 10%. Model measurements for the third day agree to either 8 or 14%, depending on the surface albedo value used. By including an aerosol with an optical depth similar to that found on clear days, the experimental team reduced the difference between model and measurement by 5% or more.
Results from ARESE II showed no disagreements larger than 14%, and the team concluded accuracy of measured absorption at 10% at best thanks to the unique dataset provided by redundant broadband measurements. This indicates any disagreement between model and measurement can be bracketed by some combination of measurement uncertainty analysis and uncertainty in the input of surface albedo and aerosol properties. The ARM experiment also emphasizes the importance of using demonstrated, high-accuracy models, especially when comparing theory with measurements.