subsequent spacecraft and for the instruments that are not currently under development are preliminary, based on derivation from the current EOS-AM definition and development estimates and analogy with previous flight project experience. These estimates will be updated before beginning development of these spacecraft and instruments.

Most of the space hardware costs in FY 1993 and FY 1994 are associated with development of the EOS-AM spacecraft and instruments, and the beginning of the development of EOS-PM. The growth in the budget in FY 1995 is essential to maintain the schedule of EOS-AM and EOS-PM and to begin the full development of AERO, CHEM, and ALT. It should be noted that the small COLOR mission is not shown in this figure. NASA assumes that it can acquire the data from COLOR as a data purchase only, in the same manner as it is acquiring the data from the SeaStar mission that will fly during Phase 1. Accordingly, the costs for COLOR are bookkept in EOSDIS. The category called “Other” in the figure includes the costs of a copy of the CERES instrument which is to fly on the US/Japan TRMM mission in 1997 and begin measurements of the Earth's radiation budget prior to EOS. The "Other" category also includes a wedge to fund additional instruments for flights of opportunity, such as to provide measurements of solar luminosity, or to study other aspects of global change that may become important in the coming years. [18] $1.1 billion, Japan $1.8 billion, Canada - $0.35, and NOAA/DOD/DOE - $2.5 billion. (Costs for European, Japanese and Canadian missions are estimates only).

The costs for the science portion of EOS are shown in Figure 7. Included here are the costs for the Interdisciplinary Teams, the costs for scientists to develop the algorithms to process EOS data, and the costs for both of their computer facilities. In the NASA budget, the costs for personnel and computer facilities to do algorithm development are bookkept with the costs for the spacecraft for which the algorithms will be developed. NASA is committed to having the algorithms for each EOS spacecraft fully developed prior to its launch, so that there will be no delay in the use of the data. The costs for algorithm development thus increase beginning in FY 1995, in preparation for the EOSAM launch.

Figure 8 illustrates the costs associated with EOSDIS. Funds for Version 0 support NASA's efforts to take existing Earth science data sets and make them more available and useful to researchers. This effort will both improve our understanding of global change and provide an experience base on which to develop the full EOSDIS. Each of the DAAC's will participate in Version 0 and is provided a small level of institutional support to do so. Funds for Version 1, etc. support the establishment of the EOSDIS capability – its basic architecture and associated networking. As in the case of algorithm development, NASA is committed to have the appropriate level of EOSDIS capability on line prior to the launch of EOS-AM. Funding is also provided for the Independent Verification and Validation activities and the in-house support for managing the data system at the Goddard Space Flight Center. Finally, funding is provided to establish the data processing capability at selected DAAC's, as is needed. The first DAAC's to be brought on line to process EOS-AM data are the Goddard Space Flight Center, the EROS Data Center, and the Langley Research Center. The remaining DAAC's will be activated as they are needed to process data from subsequent EOS spacecraft.

Cost comparisons with the original EOS program are difficult since the restructuring has been such a fundamental change. It is interesting to note, however, that to fly most of the instruments on the original large platform now requires three spacecraft - EOS-AM, EOS-PM, and CHEM - in the restructured program. The total cost of these three intermediate satellites is of course larger than the cost of the original large platform, in part because the program is spread over more years and paid for now in inflated dollars, and in part because the economy of scale of the large platform is lost. However, as can be seen in Figure 9, if we compare the cost profile required to launch the large platform and its instruments in late 1998 with the cost profile available for EOS-AM, EOS-PM, and CHEM, we see that the large platform required substantially more up-front funding. Equivalently, with limited funds available in FY 1993 and FY 1994, the large platform could not be launched until the year 2000, or later. The multiple platform approach has the clear advantage that it can begin the crucial EOS measurements early, with EOS-AM in June 1998, with only a modest growth in funds required in the early years.