CCSM3 surface upward LW (rlus) seems low

[tony_beesley@noaa_gov]

Recently I have been looking at surface energy fluxes in the CCSM3 as part of a comparative analysis of the polar energy budget in the IPCC AR4 models. For the region poleward of 80N between 90E and 270E, the CCSM3 has a longterm annual average net surface energy balance of about -45 W/m2 (i.e. 45W/m2 from atmosphere to sea ice). The first figure below shows a single year. The actual surface energy balance for this region is believed to be around +1 to +5 W/m2 (upwards).

The primary cause the discrepency appears to be a large negative bias in the upward LW flux (rlus). The CCSM3's longterm annual mean of about 193 W/m2 implies an effective surface blackbody temperature (Tb) of about 241K, assuming a surface emissivity of unity. For the same region, the GFDL CM2.1 has a mean rlus of about 248 W/m2, implying a more realistic effective surface Tb of about 257 K.

Below are some plots of the annual cycle of sig*Ts4 (black) and rlus (red) at 0N,180E and 80N,180E. The two plots at 80N are for different years. The difference between sig*T4 and rlus is greater at the Equator than at 80N, as is the overall magnitude of these terms. However, sig*T4 and rlus do not appear to be linearly proportional. (I used the surface air temperature (tas) for sig*T4 in these plots, but the result is essentially the same using the skin temperature (ts).)

My analysis thus far is limited to the pre-industrial simulation (picntrl).

Please let me know whether anyone can confirm these results. Does the definiition of lrus in the CCSM3 differ from others? Or is CCSM3 AR4 data incompatible with the Ferret data display tool that I'm using?

Tony Beesley

Surface Energy Budget 80-90N, 90E-270E year 380

Upward Surface LW (rlus) [red] and sigma*Ts^4 [black] at Equator & Dateline (Year 380)

Surface LW (rlus) [red] and sigma*Ts^4 [black] at 80N &180E (Year 280)

Surface LW (rlus) [red] and sigma*Ts^4 [black] at 80N &180E (Year 380)

 

[tony_beesley@noaa_gov]

I looked at the diagnostic plots for the CCSM preindustrial run b30.009 on the CCSM website; and the downward LW (FLDS) and net upward LW (FLNS) values seem realistic. The implied upward LW flux FLNS + FLDS also seems to agree sigma*Ts^4. (There is no diagnostic plot of upward LW flux).

In my previous post I noted that the surface upward LW (FLUS) in the AR4 data (from IPCC site) was well below sigma*TS^4. From looking at the CCSM diagnostic plots, it appears that

sigma*TS^4 - FLUS ~ 2*FLNS.

For example, near the equator sigma*TS^4 - FLUS is around 100 W/m2 (in the AR4 data). This is about twice the value of FLNS (from the diagnostic plots).
And near the north pole (in winter), sigma*TS^4 - FLUS is around 50W/m2. This is about twice FLNS from the diagnostic plots.

In other words, the appears that FLUS in the AR4 data was computed as

FLUS = FLDS - FLNS

instead of FLUS = FLDS + FLNS .

I would appreciate it if someone from the CCSM crew would verify this.

Tony