Intercomparison of the NCEP/NCAR and the NASA/DAO
5.1 Component Datasets
Most of the data collection for the NCEP/NCAR reanalysis has been performed at NCAR. The assimilated data includes surface and upper air observations that were available for the original operational NCEP analyses (available after March 1962) plus additional data that was unavailable operationally. The component datasets used in reanalysis are discussed briefly below. The component datasets are discussed in more detail in Kalnay et al. (1996).
The main source of upper-air observations is from the NCEP GTS (after March 1962); this dataset includes rawinsondes, pibals and aircraft data. The reanalysis has benefited greatly from a significant effort to improve the archive of upper air data. This includes the addition of raobs from the national archives of various countries which were made available to NCEP and/or NCAR for the reanalysis. Further details of the effort to improve the global archive of rawinsonde data is given in Kalnay et al. (1996). In addition, the U.S. Air Force prepared a global collection of data (TD54), mostly for the period 1948-1970, which is included in the reanalysis.
Satellite cloud drift winds were obtained from the original NCEP archives and the JMA.
The Comprehensive Ocean-Atmosphere Data Set (COADS), originally released in 1983, includes ships, fixed buoys, drifting buoys, pack-ice buoys, and near surface data from ocean station reports. An update for the period 1980-93 has been completed, and work is in progress for the period 1947-79.
Aircraft data are available from NCEP's GTS archive after March 1962. Additional sources of data that did not make it onto the GTS in time for operations are included (such as data from New Zealand for February 1984-June 1988). Aircraft data from experiments such as the Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE) and the First GARP Global Experiment (FGGE 1979) are included. In addition, selected USAF reconnaissance data (starting in 1947) and data from the Tropical Wind Energy Conversion Reference Level Equipment (TWERLE) constant-pressure balloons for the SH are included.
The source of satellite sounder data for the reanalysis after November 1978 is the original operational TOVS retrievals of NESDIS. Mo et al. (1995) examined the impact of satellite data on the reanalysis, which provides useful information regarding the uncertainties of the analysis without satellite data. This is particularly important for the period prior to 1979 when no TOVS soundings are available. Currently we are testing the use of VTPR data, which is available for the period Nov. 1972-Feb. 1979.
Surface land synoptic data are available from the Global GTS (usually every 3 h) starting in 1967. Earlier years are available from the Air Force and from NCDC; data coverage after 1949 is fairly good.
A text entitled "Data for Reanalysis: Inventories" (available from R. Jenne, NCAR, P.O. Box 3000, Boulder, CO, 80307-3000) has maps and tables of the typical spatial coverage of surface and upper air data used in reanalysis. Most of this information covers the period from about 1948 on. Many other reports have been prepared which give additional information about the attributes of various datasets and the status of projects to prepare the data; a list of these papers is found in Kalnay et al. (1996) in section 3.
5.2 Typical Spatial Coverage
This section summarizes the basic spatial characteristics of the observations that were used as input to the assimilation. Details about the analysis scheme including the quality control of the observations are described in Kalnay et al. (1996).
An example of the spatial coverage of the various data sources for July 15, 1986 (0000 UTC-1800 UTC) is given in Fig. 112. This figure shows the locations and counts (both wind and mass) of the radiosonde / pibal / dropsonde data (Fig. 112a), of the satellite cloud track winds (Fig. 112b), of the COADS marine data (Fig. 112c), of the surface bogus (PAOBS and SPLASH level dropsondes) data (Fig. 112d), of the commercial aircraft reports (Fig. 112e), of the TOVS retrievals of NESDIS (Fig. 112f), and of the surface land synoptic data (Fig. 112g).
This section shows the number of daily reports of each of the major data types discussed in the previous two sections for the period 1979-1995. For each major data type we show counts for the NH extratropics (30°N - 60°N), the SH extratropics (30° S - 60°S), the Tropics (30°S - 30°N) and the poles (60°N - 90°N; 60°S - 90°S). For the rawinsonde data (Fig. 113), the COADS data (Fig. 115), the commercial aircraft data (Fig. 117), and the land surface data (Fig. 119) we show counts for temperature and wind data separately. For the satellite cloud track winds (Fig. 114), we show low level and high level data. In Fig. 116 we show counts for the surface bogus and splash level dropsondes separately. In Fig. 118 we show TOVS retrievals of NESDIS over land and ocean separately.
Further examination of Fig. 113 shows the impact of recent major field experiments on the rawinsonde data. The number of reports during FGGE (1979) significantly increased in the Tropics and in the middle and high latitudes of the NH while the number during TOGA/COARE significantly increased in the Tropics. Figure 113 also shows that increases in the NH and SH extratropics during the 1980's have generally plateaued during the 1990's. There is a considerable decrease in data from the NH polar regions directly linked to the break up of the former Soviet Union. Figure 117 shows a tremendous increase in aircraft observations starting in 1991, when these reports became automated.