From Research to Operations in Weather Satellites and Numerical Weather Prediction: Crossing the Valley of Death (2000)

Environmental satellites serve as a major component of the forecasting system, both from the standpoint of capabilities and of costs. The launch of the first meteorological satellite in April 1960 opened up an important new vantage point for observing the atmosphere and eventually the entire earth system. Even with the new potential of satellite observations, it took over a decade before the benefits of satellite meteorology were being routinely used in the operations of the National Weather Service. Another half decade passed before the satellite data were used in operational numerical weather prediction.

In the 1970s, in an effort to assist in the transition of satellite observational data from research to operations, NASA conducted the Operational Satellite Improvement Program (OSIP) that was responsible for developing new instruments for environmental satellites. In this program, NASA developed the prototype instruments, flew them frequently on high altitude aircraft missions for a preliminary checkout, and then deployed them on research spacecraft for a complete evaluation. Successful instruments were then available for transition to the operational NOAA satellite series (Fleming, 1996).

For budgetary and mission clarity reasons, NASA canceled the program in 1982. NOAA never developed a replacement for OSIP; instead, it continued with a procurement practice of specifying the instrument performance and having the contractor deliver the instrument for flight

on the operational satellites. This procedure did not allow for the iterative development process that was so successful in the OSIP program (Fleming, 1996).

Because environmental satellites currently provide the only effective means of covering the entire globe in a short period of time, the climate research community depends on satellite coverage for many of the studies of climate processes and overall climate monitoring. The NRC Committee on Global Change Research suggested that NASA' s Earth Science Enterprise (ESE) may not be able to fulfill the needs of the climate community, particularly with respect to the long-term monitoring needed for studies of detection and attribution of climate change and recommended that the task of monitoring climate variables be assigned to the operational satellite systems of NOAA (NRC, 1999a). A recent review of the ESE mission plans confirmed that concern and stressed that the planning for the transition from research to operational satellites must be carefully accomplished to ensure the successful support of all users of satellite data (NRC, 1999b). BASC learned that both the Administrator of NASA and the Administrator of NOAA have recognized these problems in mission definition, requirements process, and related issues in separate letters to the President 's Science Advisor and are presently working on policy papers to clarify these issues.

In response to the growing concern for the long term monitoring mission for climate, NASA and NOAA have proposed a National Polar-Orbiting Operational Environmental Satellite System (NPOESS) Preparatory Project (NPP) which will serve as a bridging mission, enabling the near-term evaluation of some of the EOS instruments to determine their suitability for implementation on NPOESS operational satellites. In some respects, the NPP fulfills some of the important characteristics of the former OSIP program. However, it is presently planned as a single mission to bridge the time between the present NOAA and DOD polar orbiters and the initiation of the converged NPOESS system and does not serve as an ongoing mechanism for the development and testing of new instruments.

BASC has reviewed the NPP program plans. Several of the BASC members participated in earlier NRC studies (NRC, 1998c, 1999b) and were able to compare and evaluate the progress that had been made in the planning process for the NPP. The NPOESS program has a clearly articulated objective to meet the operational satellite requirements of NOAA and the Department of Defense (DOD) and to couple this program with the efforts to achieve NASA's mission. However, the process

for promoting the transition from research to operations has not been clearly articulated.

Research satellites are often justified in terms of benefits to society through potentially improved products or services. At other times, they are justified in terms of obtaining better understanding of environmental processes through a single experiment, with little intent for the satellite system to become operational. In many cases data that were originally intended only for research purposes also prove to be extremely valuable in the preparation of useful operational products. For example, the TOPEX mission has produced new radar scatterometer data that has demonstrated its usefulness for determining sea state, from which one can infer the surface wind velocity, a measurement that had been extremely difficult to obtain. These measurements are now being considered by ECMWF for assimilation into operational prediction models. A similar situation is now developing with several of the projected EOS suite of measurements having the potential to become an operational set of measurements on NPOESS.

However, little evidence was presented to BASC illustrating dedicated efforts to examine new uses of satellite data. Further, instrument development programs for operational satellite programs typically do not include funding for development of the necessary assimilation algorithms. Lack of such funding often impedes the timely use of the new data products by the operational forecasting community.

After a new observing system is developed and deployed operationally, the taxpayer does not benefit by receiving improved forecasts until the new observations are included in the forecast preparation process. This inclusion is, in general, very complex, and requires close collaboration between the instrument designers and the data assimilation scientist. It requires the development of observation models that transform atmospheric model variables into model observations (e.g., satellite radiances) and quality control procedures that take advantage of information from the data and the model, etc. Experiences at ECMWF, EMC, the United

Kingdom Met. Office (UKMO), and other centers indicate that this preparation for implementation requires two or more scientists for two to five years before launch if the new instrument is to be used efficiently shortly after launch. Observing System Simulation Experiments (OSSE) can help to create all the necessary software and can begin using the new data soon after launch. When this does not happen, there is a gap of several years when the data is collected, but not used efficiently in the EMC models. During that period, the cost of the observing system, which may be on the order of $100 million per year, does not produce a direct forecast benefit to the user. With advanced planning and a full science team employed early in the new sensor development, the users of the new data can begin operations immediately after launch and sensor checkout, greatly extending the effective lifetime of the instrument. The science team should include experts in instrumentation, algorithm development, operational forecasting, and the end user requirements. BASC believes that proper planning for transition, with full stakeholder participation, is likely to be very cost effective in the long run.

The return on the nation's investment in remote sensing systems depends on data utilization for the public good. Utilization depends on public and private access to, and exploitation of, these new sources of weather, climate, and other environmental data. NASA's EOSDIS program, although facing many technical and budgetary difficulties, was conceived with this purpose in mind. Now that the conversion of some of the long-term measurements from NASA satellites to NOAA satellites seems inevitable, the accessibility of both NOAA and NASA data should be addressed. Current plans do not appear adequate to ensure reasonable accessibility of archived NPOESS data. Several NRC reports have also expressed concern regarding a national commitment for ensuring that long-term monitoring of earth from space takes place (NRC 1998c, 1999b). Current plans include a transfer of key earth observations from the research-based EOS to the operational-based NPOESS. While plans are being developed for the sensors and ground processing, it appears that only limited consideration has been given to the requirements for data archiving (Withee, 1999).

An efficient, robust data archiving system is at the core of the transition from the research to operations processes and the generation of advanced forecast products. It provides the case studies necessary to develop and evaluate new forecasting techniques. A major challenge for the data system is the huge size of the projected data archives. It is necessary, but difficult, to ensure that data are quality controlled, accessible, and reprocessed as necessary for a long enough period of time (years to decades) to serve as climate monitoring data. One way to handle this challenge is to use and, if necessary, add to EOSDIS. In the future it is reasonable to anticipate that a considerable number of weather and climate related forecast products will need to be captured and archived into a data system. These products will be used by EMC to make operational forecasts, but the business sector and the research community will also use them. Thus the model data archive is a resource for research and application that should be readily available to these diverse communities. Given the advances in the private sector with Internet data webs, NASA and NOAA need to be more forward looking with respect to incorporating advances in data base systems for the delivery of data to users.

The transition to the NPOESS satellite observational system has focused on the difficulty of combining the mission and research requirements of NOAA, DOD, and NASA. The coordination across observational systems, including both in situ and satellite observations, has not been a part of the satellite convergence charge (NRC, 2000a). Better coordination of the data availability from satellite and in-situ observational systems has considerable potential. Such coordination is critical to evaluate the NPOESS sensors through calibration/validation activities, to improve the efficiency of data assimilation through better utilization of all data sources, and to enable research in weather and climate and development of user-specific advanced forecast products through one-stop data shopping.

Based on the discussions and presentations at the summer study, BASC assessed the transition process for the environmental satellites using the template developed in Chapter 2.

The research programs in support of the environmental satellites are strong and offer the operational community opportunities for improvement.

The NASA Operational Satellite Improvement Program (OSIP) was a very effective mechanism for developing new sensors, proving their capability, and handing them over to the operational agencies for implementation. The present NASA efforts, including the Instrument Incubator program and the New Millennium missions, address part of this process and are to be commended (NRC, 1999b). The NPP program offers a method for transferring some of the sensors from NASA to NPOESS. BASC also recognizes that programs such as NPOESS are not a solution to all climate and weather observational needs.

A strong research and exploratory program in advanced technologies and sensor development, recognizing modernization as a continuing requirement, has not been formulated.

There are deficiencies in the interface with the user community, particularly with respect to the data archival needs of the various users of the satellite data and products (NRC, 1998c, 1999b).

BASC believes that an efficient, robust data archiving system is at the core of effectively linking research to operations, the generation of advanced forecast products, and continual data utilization for the public good. The size of the observational and forecast data archives presents a considerable challenge that has not yet been met. The recent track record

of the U.S. environmental community has not been encouraging. For instance, the Earth Observing System Data and Information System (EOSDIS) has attempted to meet the archival needs of NASA's Mission to Planet Earth, but, despite large expenditures, has yet to prove its capability to fulfill that need. The NPOESS planning process has not yet addressed the need for archiving the resulting data, and there is little effort being devoted to the process of integrating in situ and satellite observations to create a more efficient observational system. NASA's Tropical Rainfall Measuring Mission (TRMM) satellite program, while a relatively modest effort, did an excellent job in planning for data archival and access and could serve as a model for future activities.

There is a limited capability within the present program to examine the impacts of the various sensors. Part of this is tied to the deficiency identified in Chapter 3 dealing with the lack of system infrastructure to run parallel operations to evaluate the impact of new sensors and forecasting techniques.

Several NRC reports expressed considerable concern about the U.S. plans for NPOESS (NRC, 1998c, 1999b, 2000a). BASC members involved in these earlier assessments concluded that, based on the agency plans for the NPP presented at the study, the program is moving in the right direction to address many of these earlier concerns, including the acceptance of the climate monitoring mission by NOAA as evidenced in the NPP plans. BASC encourages the transition of research achievements into long-term environmental monitoring and applications. There is a growing need for an operational global climate and environmental monitoring system to meet the requirements for public safety and economic development. BASC encourages NOAA, NASA, and DOD to continue the strong cooperation required to ensure success in this effort. At the same time, the transition from research to operations cannot be viewed as simply a sensor or instrument transition.

1. Replacement for the Operational Satellite Improvement Program (OSIP).

NPOESS will not satisfy the full spectrum of research or operational needs of the nation, particularly for the study of longer-term issues such as climate change, nor is it likely to generate an impetus for technologic innovation. Robust programs of technologic development, exploratory sensor development, and research missions at NASA are needed to enable a continuous push toward improved capability and innovative product development. In turn, a continuing process similar to the former OSIP can assess the state of research and operational technology and update NPOESS as needed to accomplish its national mission. This requires a more coordinated plan for research and technologic innovation that includes coherent budget and timing links for a transition to operations. The modernization of operational sensors should be a continuous process.

2. Enable EMC to utilize new satellite observations in improving forecast capability.

The objective is to develop a timely feedback and advanced planning component of the NPOESS process that ensures a more rapid transition of new observations to operational use. Within NOAA, these efforts could be closely linked to the development, testing, and integration facility at NOAA's EMC that was recommended earlier in this report (see Chapter 3).

3. Plan for archiving NPOESS data.

Data are central to both the research and operation enterprise, and its availability is critical for advances in research and improvements in operations. For major studies of the atmosphere, upper air data are crucial. While radiosonde data are archived at a number of centers around the world in accordance with international agreements through

the World Meteorological Organization, no such mechanism is in place for satellite data. There is little incentive for operational forecasters to keep satellite data either for climate studies or for advanced operational research. During the summer study, the Assistant Administrator for NESDIS acknowledged that archiving plans for NPOESS are insufficient. Early planning and assessment is essential and requires a very skilled working group whose members are collectively able to foresee how technology will advance over the next decade and how these advances might make this huge data archiving and accessibility task more tractable with the human and dollar resources available. The NPOESS program has set up a variety of review teams related to science, algorithm performance, etc. The team in Recommendation Satellite-2 can assist the NPOESS program in developing the archiving requirements for the NPOESS user communities.

NASA's EOSDIS has demonstrated the complexities and costs associated with the development of an accessible data archive. Given that EOSDIS is currently the only data system with the potential to address the large-scale problems described here, careful assessment of EOSDIS suitability for NPOESS archival is appropriate. Additionally, NASA 's Tropical Rainfall Measuring Mission (TRMM) satellite program, although much smaller than NPOESS, established an excellent archival program and could be studied for best practices.

4. Integrating across observation networks.

NPOESS has begun the process of convergence in national satellite systems, but the issues being addressed are equally applicable to the entire observational network. Similar integration is appropriate for in situ networks and for combined satellite-in situ measurement systems. This is a key issue that is likely to expand as the forecasting family expands to include atmospheric chemistry, health, and other variables. To facilitate the data assimilation process and therefore accelerate the transition of the use of disparate and complex data into operational forecasting, the first steps might focus on national and international sounding capabilities and the importance of developing a consistent, high-quality, long-term observational capability (NRC, 1998a).

5. Continuity of measurements for climate studies.

Several recent NRC reports have recommended that careful attention be paid to the climate observing system. Global Environmental Change: Research Pathways for the Next Decade (NRC, 1998c) specifically recommended that the spaced-based operational systems be developed to also meet the climate observing needs. Adequacy of Climate Observing Systems (NRC, 1999a) identified needs with the entire range of observing systems and identified several principles that should be implemented to ensure viable climate data for the future. BASC reiterates the recommendations of these and other NRC reports stressing the need for more effective planning and execution of the observing systems for climate and weather.