SURFACE FLUXES
255
feedback to retard the accumulation of flux error, as there is in the case
of the heat flux dependence on SST. Therefore, large systematic errors
are particularly problematic. The comparison of zonal averages from
different precipitation data sets in Fig. 7 illustrates the problem, even
though large regional differences are lost, because of compensation along
latitude circles. The best agreement is found around 30 ◦ latitude in both
hemispheres. Ideally, one data set would be applicable globally, but it
appears that each precipitation product is demonstrably deficient in one
or more of the following five zones; the Antarctic (poleward of 61 ◦ S), the
Southern (65 ◦ S to 30 ◦ S), the Equatorial (30 ◦ S to 30 ◦ N ), the Northern
(30 ◦ N to 62 ◦ N ), and the Arctic (poleward of 62 ◦ N ).
Comparing the satellite products in the tropics, the CMAP values of
Xie and Arkin (1996) are about 30% greater than GPCP (Global Precipitation Climatology Project) (Huffman et al., 1997), and more than
an ocean climate model can deal with. The MSU (Microwave Sounding
Unit; Spencer, 1993) falls between. In contrast, the subtropical ocean
simulations are much better when forced with CMAP than either GPCP,
or MSU. Polar latitudes are not sampled by some satellites, including
the MSU, so CMAP becomes reliant on NWP model output, complete
with spectral ringing signals in the precipitation. Therefore, the only
viable Antarctic product is GPCP. It is also available in the Arctic, but
doesn’t compare all that well to the observed compilation of Serreze and
Hurst (2000). Therefore, as denoted GXGXS in Fig. 7, a possible, but
far from ideal, compromise, is to use GPCP(G), CMAP(X), GPCP(G),
CMAP(X) and Serreze/Hurst(S), respectively, in the five zones noted
above. Another satellite based product is HOAPS (Hamburg Ocean Atmosphere Precipitation System), but it appears to be an outlier, giving
only 15 mg/m 2 /s at 44 ◦ S.
6.
A merged flux climatology
According to WGASF (2000), ”there is presently no one flux climatology which does not exhibit significant errors in one region or another
in each of the various flux components.” In particular, there are serious issues with the flux products from NWP reanalysis, which are also
expected in the operational fluxes that would be very convenient for
GODAE. Most problematic are fields like radiation and precipitation
which strongly depend on the cloud field. For example, tropical radiation from NCEP (Fig. 6) is significantly lower than all satellite products,
with ERA-15 producing even less solar heating. Also, NCEP precipitation (Fig. 7) around 5 ◦ S, is much greater than other datasets, with
the better agreement with CMAP due to CMAP’s blending of station
255
feedback to retard the accumulation of flux error, as there is in the case
of the heat flux dependence on SST. Therefore, large systematic errors
are particularly problematic. The comparison of zonal averages from
different precipitation data sets in Fig. 7 illustrates the problem, even
though large regional differences are lost, because of compensation along
latitude circles. The best agreement is found around 30 ◦ latitude in both
hemispheres. Ideally, one data set would be applicable globally, but it
appears that each precipitation product is demonstrably deficient in one
or more of the following five zones; the Antarctic (poleward of 61 ◦ S), the
Southern (65 ◦ S to 30 ◦ S), the Equatorial (30 ◦ S to 30 ◦ N ), the Northern
(30 ◦ N to 62 ◦ N ), and the Arctic (poleward of 62 ◦ N ).
Comparing the satellite products in the tropics, the CMAP values of
Xie and Arkin (1996) are about 30% greater than GPCP (Global Precipitation Climatology Project) (Huffman et al., 1997), and more than
an ocean climate model can deal with. The MSU (Microwave Sounding
Unit; Spencer, 1993) falls between. In contrast, the subtropical ocean
simulations are much better when forced with CMAP than either GPCP,
or MSU. Polar latitudes are not sampled by some satellites, including
the MSU, so CMAP becomes reliant on NWP model output, complete
with spectral ringing signals in the precipitation. Therefore, the only
viable Antarctic product is GPCP. It is also available in the Arctic, but
doesn’t compare all that well to the observed compilation of Serreze and
Hurst (2000). Therefore, as denoted GXGXS in Fig. 7, a possible, but
far from ideal, compromise, is to use GPCP(G), CMAP(X), GPCP(G),
CMAP(X) and Serreze/Hurst(S), respectively, in the five zones noted
above. Another satellite based product is HOAPS (Hamburg Ocean Atmosphere Precipitation System), but it appears to be an outlier, giving
only 15 mg/m 2 /s at 44 ◦ S.
6.
A merged flux climatology
According to WGASF (2000), ”there is presently no one flux climatology which does not exhibit significant errors in one region or another
in each of the various flux components.” In particular, there are serious issues with the flux products from NWP reanalysis, which are also
expected in the operational fluxes that would be very convenient for
GODAE. Most problematic are fields like radiation and precipitation
which strongly depend on the cloud field. For example, tropical radiation from NCEP (Fig. 6) is significantly lower than all satellite products,
with ERA-15 producing even less solar heating. Also, NCEP precipitation (Fig. 7) around 5 ◦ S, is much greater than other datasets, with
the better agreement with CMAP due to CMAP’s blending of station