In the far eastern tropical Pacific, a stratus cloud deck forms over the cool surface water near Peru and Chile and deep convection associated with the ITCZ forms over the warm surface waters north of the equator. Clouds have both a cooling effect on the ocean surface due to the reduction in solar radiation and a warming effect due to the emission of longwave infrared radiation. The net effect of the clouds on the surface radiation depends upon many factors including cloud base height, cloud thickness, coverage, background moisture, and aerosol loading. While cloud types defined by low, middle, and high values of cloud optical thickness versus cloud-top pressure provide a useful tool for relating cloud type to cloud forcing properties at the top of the atmosphere (e.g., Chen et al. 2000), the cloud forcing properties at the surface are less understood and more difficult to observe.
As part of the Eastern Pacific Investigation of Climate Processes (EPIC), from 2000 through 2003, 11 buoys in the far eastern Pacific were equipped with a suite of sensors from which surface solar and longwave cloud forcing could be computed. In this study, we use this buoy data to analyze the meridional structure and seasonal cycle of the cloud forcing in the stratus deck region at 85°W, 20°S and across the stratocumulus, cold tongue, ITCZ complex along 95°W from 8°S to 12°N. Buoy measurements in this region show both large reduction of solar radiation by the stratus cloud deck and a relatively large enhancement of longwave radiation. The net effect of the stratus clouds, however, is a reduction in the incoming radiation. Indeed, at all sites in the EPIC region solar cloud forcing was larger in magnitude than the longwave cloud forcing. In contrast to the stratus site, in the ITCZ region the longwave cloud forcing was weak, while the solar cloud forcing was for some months more than -165 W m. Consequently, in the ITCZ region clouds had a very large impact on the net radiation. Over the cold tongue region, particularly during the warm season, there were many periods of clear sky when the cloud forcing was near zero.
The buoy cloud forcing fields were compared to three of the most commonly used global atmospheric products: satellite-based ISCCP fields, NCEP2 reanalysis, and 40-yr ECMWF Re-Analysis (ERA-40) fields. The latter two products combine NWP models and observations and therefore tend to have good realism on weather time scales. While reanalysis products are often treated as data, the radiation fields therein are not well constrained by data. If realistic surface radiation fields are sought, satellite fields should be used in lieu of NWP fields. However, in this case, one must question the validity of the other modeled fields. Incorrect cloud forcing is indicative of incorrect global energy balance. This is particularly true in the ITCZ region that acts as the rising branch of the Hadley cell. If, as was found in this study, the cloud forcing biases are not offset by biases in the clear-sky radiation, then these biases in the cloud forcing also represent biases in the net surface heat flux on the ocean module of the coupled NWP models. Thus, biases in the cloud forcing fields may have large impact on other aspects of the numerical weather prediction products. For this reason, a quantitative assessment of the NWP cloud forcing is justified.
Both NWP fields show too much solar cloud forcing (implying a cold bias) in the ITCZ region and near the equator during the warm season from December through July (Fig. 6). Because NCEP2 forms precipitating clouds with ITCZ-like radiative properties on the equator during the warm season (February–April), Figs. 10–11, the NCEP2 solar cloud forcing bias is particularly large (more than -100 W m) there when the cold tongue normally weakens. If used as boundary conditions for an OGCM, the NCEP2 solar cloud forcing would result in a cold SST bias. For example, the NCEP2 solar forcing error of over -100 W m would lead to a 7°C cold bias within 3 months, assuming a 30-m-thick slab ocean mixed layer. Because this is the warm season when the cold tongue is weak, this solar cloud forcing bias would cause an unrealistically prominent and persistent cold tongue. Persistently cold SSTs along the equator (i.e., a permanent cold tongue) would be expected to have important ramifications on the seasonal cycle in the entire eastern tropical Pacific.
FIG. 6. (upper left) Mean annual cycle of solar cloud forcing field along 95°W from 8°S to 12°N and at 20°S, 85°W from buoy measurements and mean annual cycle of difference between surface solar cloud forcing and buoy field along 95°W from 8°S to 12°N and at 20°S, 85°W for (upper right) ISCCP, (lower left) NCEP2 reanalysis, and (lower right) ERA40.
On the other hand, in the stratus cloud deck region at 20°S solar cloud forcing bias is of the opposite sign, 18 W m for NCEP2 and 19 W m for ERA-40, which would tend to produce a warm bias in SST. Furthermore, the positive (warm) solar cloud bias associated with the stratus cloud deck extends northward to the equator during the cold season. Because the reanalyses have solar clear-sky radiation larger than that modeled for the 20°S buoy, their incident solar radiation is even larger than these cloud forcing deviations would suggest.
Longwave cloud forcing comparisons were better than for solar cloud forcing; deviations were typically less than 20 W m. Longwave cloud forcing appears to be too strong in the ITCZ and in the Southern Hemisphere during the warm season (February–June) for all three products. Likewise, longwave cloud forcing appears to be too weak during the cold season when a stratus cloud deck typically develops. Overall, discrepancies between the ISCCP and buoy cloud forcing are smaller than for the NWP fields. If ISCCP were used as the benchmark, similar patterns for the NWP biases would be found.
The results described here are consistent with the results found by Weare (1997) for NCEP1, by Jakob (1999) for the ECMWF 15-yr Re-Analysis and by the European Cloud Systems Study (EUROCS) GCM intercomparison (Siebesma et al. 2004): almost all models that were tested underpredicted the cloud cover in the stratocumulus regions [in the case of Siebesma et al. (2004), Californian stratus], while in the trade-wind region including the ITCZ, cloud cover was overpredicted. As a consequence, Siebesma et al. (2004) found that downwelling surface shortwave radiation was overpredicted by typically 60 W m in the stratocumulus regimes and underpredicted by 60 W m in the ITCZ and trade wind regions. Curiously, these cloud forcing discrepancies appear to be larger than those found by Weare (1997) for NCEP1. As pointed out by Roads (2003), although the changes made between the NCEP1 and NCEP2 appeared to be minor, the resulting tropical hydrologic cycles were surprisingly different. In particular, NCEP2 tended to produce a more noticeable double ITCZ and too much precipitation closer to the equator.
It is interesting to note that these biases in surface cloud forcing are of the correct sign to account for the warm SST bias in the stratus deck region west of Peru (Mechoso et al. 1995) and the cold SST bias found at the equator in most coupled GCMs (Davey et al. 2002). These SST biases, however, are not just due to errors in radiative effects but also involve complicated interactions with wind stress fields, turbulent air-sea heat exchanges, and mixing processes.
The in situ measurements from buoys on which the present study was based can be used to validate and assess NWP and satellite products. Linear fit polynomials for relating surface solar and longwave cloud forcing are shown in Tables 1–3. These empirical relations can be used as a ground truth test for atmospheric models that generate their own cloud and surface radiation fields.
The authors thank the NCEP, ECMWF, ISCCP, TRMM, and TAO project offices for providing the data used in this analysis. NCEP2 data were accessed from http://nomad2.ncep.noaa.gov:9090/dods/reanalyses/reanalysis-2/month/dg3/dg3.info through the Distributed Oceanographic Data System (DODS). ISCCP data were obtained from http:// isccp.giss.nasa.gov/projects/browse_fc.html. ECMWF ERA-40 data were downloaded from the ECMWF data server at http://www.ecmwf.int/. TRMM data were obtained from http://daac.gsfc.nasa.gov/data. The authors also wish to thank K. Huang, K. Colbo, and Y. Serra for their help in processing and accessing data, and J. Hare for providing the clear sky models used here. Helpful comments were provided by three anonymous reviewers. This research was supported by grants from the NOAA Office of Global Programs, Pan American Climate Studies.
Return to previous section or go to next section
PMEL Outstanding Papers
PMEL Publications Search