An Assessment of Global fields of Sensible and Latent Heat Fluxes from

AIR-SEA FLUXES FROM THE NCEP OPERATIONAL ANALYSIS/FORECAST SYSTEM

Glenn H. White¹, K. Campana¹ and J. Janowiak²

¹Environmental Modelling Center/²Climate Prediction Center

National Centers for Environmental Prediction/National Weather Service

National Oceanic and Atmospheric Administration/U. S. Dept. of Commerce

Washington, D.C., USA

Glenn.White@noaa.gov

Analysis/forecast systems used in numerical weather prediction (NWP) have matured to the point that further development requires the direct verification of model physics. Changes in operational NWP systems are designed to improve their synoptic forecast performance; their effect on surface fluxes is often not evaluated as fully, in part because reliable independent estimates of surface fluxes are more difficult to obtain than measurements of atmospheric fields.

NCEP conducted two major reanalyses, one from 1948 to the present (NCEP-1; Kalnay et al., 1996; Kistler et al., 2001) and a second from 1979 to the present (Kanamitsu et al., 1999). The long, consistent data sets from reanalyses provided a golden opportunity to evaluate surface fluxes against independent estimates and many studies did so (Taylor, 2000, section 11.4). However, the operational NCEP global system has evolved considerably from the versions used in the reanalyses and it is not immediately obvious how applicable the lessons learned from reanalysis are to the operational model. This paper examines air-sea fluxes from the current NCEP operational system and compares them to fluxes from NCEP’s Climate Data Assimilation System, the continuation of the NCEP/NCAR (NCEP-1) reanalysis. It seeks to determine whether the system's physics has improved over the last 6 years (the NCEP-1 reanalysis is based on the operational system of 1995) and to gain insight into problems in the operational model physics.

Global Mean Balances

May 2000-April 2001

	CDAS	Operational	Kiehl & Trenberth	Range
Precipitation	2.85 mm/d	3.00	2.69	2.69-3.1 (2.83,2.83)
Evaporation	2.90	3.02
Precipitation minus Evaporation	-.05	-.02
*Sensible heat*	*16.53 W/m²*	*11.6*	24	16-27(21,21)
Latent heat	84.06	87.46	78	78-90(82,82)
*Surface downward short wave*	*206.13*	*197.58*	*198*
*Upward short wave*	*45.31*	*28.31*	30
*Net short wave*	*160.82*	*169.27*	*168*	142-174 (163, 168)
Downward long wave	334.66	336.7	324
Upward long wave	395.89	397.5	390
Net long wave	61.23	60.8	66	40-72(59, 63)
*Net radiation*	*99.58*	*108.48*	*102*	99-119 (104,103)
*Surface Net Heat Flux*	-1	*9.42*	0
TOA downward short wave	341.87		342
*Upward short wave*	*116.81*	*102.38*	*107*
Outgoing long wave	237.78	243.46	235
*TOA net radiation*	*-12.71*	*-3.97*
Atmospheric heating	-11.72	-13.39

Table 1 Global mean surface and top of the atmosphere fluxes averaged over May 2000-April 2001 from 0-6 hr forecasts from the NCEP-1 reanalysis or CDAS, and the operational NCEP global analysis/forecast system compared to a climatological estimate from Kiehl and Trenberth (1997). The last column gives the range of the estimates reviewed by Kiehl and Trenberth and in parentheses the mean and median value. Highlighted values indicate substantial differences between CDAS and the operational analyses.

The operational model considered here has higher resolution (T170, 42 levels) than CDAS (T62, 28 levels) and uses satellite radiances in its assimilation rather than NESDIS temperature soundings. It has updated short wave radiation and boundary layer parameterisations and a corrected surface albedo. It also has changes to horizontal and vertical diffusion, gravity wave drag and a retuned cloudiness parameterisation. To correct a problem in the air-sea flux parameterisation (Zeng et al., 1998), a new thermal roughness based on TOGA COARE observations is used in the operational model. Since the period compared here, a new parameterisation of cloudiness based on prognostic cloud liquid water and a parameterisation of cumulus momentum mixing has been introduced in the operational system on May 15, 2001. The new cloudiness appears to give more realistic surface net short wave near the equator and has more low-level stratus clouds in the eastern sub-tropical oceans. Further tuning of the new cloudiness is planned.

As shown in Tables 1 and 2, the operational system has more realistic surface short wave radiation than CDAS. However, the surface energy balance is more out of balance in the operational system and sensible heat flux in the operational system is lower than other estimates. Oceanic evaporation in NCEP-1 was higher than ship-based estimates and is higher still in the operational model; decreasing the evaporation would increase the surface energy imbalance.

Global Mean Balances

May 2000-Apr. 2001

Ocean

	CDAS	Operational	COADS	SRB
Precipitation	3.1 mm/day	3.29
Evaporation	3.4	3.52
Precipitation minus Evaporation	-.29	-.23
*Sensible heat*	*12.71 W/m²*	*6.27*	*10.1*
Latent heat	98.47	101.97	88
Surface downward short wave	199.85	195.04
*Upward short wave*	*35.15*	*18.25*
*Net short wave*	*164.7*	*176.79*	*170.4*	*173.4*
Downward long wave	352.06	350.89
Upward long wave	408.11	408.19
Net long wave	56.04	57.29	49.2	41.9
*Net radiation*	*108.66*	*119.50*
*Surface Net heat Flux*	*-2.52*	*11.25*	*23.3*
TOA downward short wave	349.53
*Upward short wave*	*117.95*	*98.71*
Outgoing long wave	239.46	246.60
*TOA net radiation*	*-7.87*	*4.22*
Net atmospheric heating	-5.35	-7.03

Table 2 Global mean fluxes over the ocean for May 2000-April 2001. COADS refers to climatological estimates by da Silva et al. (1994) for 1981-92, SRB to satellite based radiation estimates by Darnell et al. (1992) and Gupta et al. (1992).

Figure 1 compares precipitation during June-August 2000 from 0-6 hr forecasts from the NCEP operational system and from CDAS to OPI, an independent estimate based on rain gauges and infrared satellite estimates. The NCEP1 reanalysis did not concentrate tropical precipitation enough and that is clearly evident. Precipitation from the operational analysis forecast system is clearly more like the OPI estimate in the tropics than CDAS is.

Figure 2 displays total cloudiness for March-May 2000 from U.S. Air Force nephanalyses, CDAS and the NCEP operational system. The operational system clearly resembles the nephanalyses more in the tropics, except in the eastern subtropical oceans. The operational system clearly has too few low-level stratus clouds in these regions, a common problem in NWP data assimilation systems (Taylor, 2001).

Differences in surface fluxes for March-May 2000 between CDAS and the NCEP operational system are shown in Fig. 3. Large differences in surface net heat flux can be seen and are clearly dominated by differences in surface net short wave flux. Large differences in net short wave can be seen over the Indonesian region, where CDAS has too little cloudiness, and in the eastern subtropical oceans where the operational system has too little.

Air-sea fluxes from the currently operational NCEP global analysis/forecast system are distinctly different from CDAS. The precipitation pattern is substantially improved and short wave radiation is more realistic. Surface winds and stress in the Pacific are stronger and in better agreement with other estimates. However, oceanic sensible heat flux is now lower than other estimates and the ocean surface energy balance is more out of balance. Radiation and clouds still require substantial improvement. Low-level stratus clouds in particular are a problem; the lack of them in the operational system distorts the surface radiation budget. Our knowledge of the surface fluxes also requires substantial improvement for model development: for example, the magnitude of evaporation and the hydrological cycle is not well known. Estimates from ship-based observations imply that both CDAS and the operational model have too much evaporation; however, it is not clear that surface flux estimates from such observations and parameterisation schemes yield reasonable global surface energy balance.