Hi;
I've been using ANA_SRFLUX and ALBEDO with NCEP clouds, air temp and relative humidity to calculate shortwave radiation. When I first started using it, I found it made the ocean too warm, so I reduced the solar constant from 1385 to 700 W/m2 in mod_scalars.F. That was a hack justified at the time by the fact that NCEP underestimates clouds because it doesn't have stratus clouds. This hack has propagated into the distribution version of ROMS, so you might be doing it too!
Lately, Enrique has questioned the validity of this hack, especially down near the equator. He suggested instead increasing the NCEP cloudiness by some amount in northern areas. I wanted to see what this did, so I compared the shortwave radiation produced by ROMS (via Parkinson and Washington's 1979 algorithyms) with varying amounts of clouds to NCEP's shortwave radiation in the Bering Sea. NCEP overestimates shortwave radiation up there by 70-80 W/m2 in the summer (Ladd and Bond, 2002), so NCEP could be considered an upper limit for reasonable values.
In almost every case I tried, using solar constant=1385, the ROMS formulation overestimated the daily average shortwave radiation by a factor of 1.5 (ie. 200 vs 300 W/m2). The only way I could bring it down to resonable values is to ditch NCEP's clouds and use constant cloudiness of .9, which according to ISCCP is a good climatological value. I also plotted NCEP's clouds, and they looked noisy-bad; there wasn't even a seasonal signal.
So here are the questions I submit for discussion:
1) Has anyone else evaluated the Parkinson and Washington shortwave radiation formulation? Is the solar constant reduction needed everwhere? If not, check your code cause you might be using it.
2) Is there a more "modern" formulation available?
3) If there is no other formula, NCEP clouds are clearly unacceptable, at least up north. Is there another cloud data set? We found ISCCP, but it seems to end in 2001.
4) We are considering ditching the ALBEDO option, and instead adjusting NCEP shortwave radiation where ISCCP shows a high percentage of stratus clouds. Anybody have a strong feeling about this?