APPENDIX A

A POST-2000 IMPLEMENTATION STRATEGY FOR GCIP

The GEWEX Continental-scale International Project (GCIP) has made tremendous strides in several of its science objectives during the past five years while others, such as the water resources objective, have not yet started. It is clear that a more balanced effort is needed in the future if GCIP is to make progress toward its long-term goal of demonstrating skill in predicting changes in water resources on time scales up to seasonal, annual, and interannual as an integral part of the climate prediction system. GCIP also needs to develop its strategy for contributing to the overall program for the GEWEX Hydrometeorological Panel as well as contributing to the joint GCIP/PACS studies of the variability of warm season precipitation over North America.

A.1 GCIP Implementation and Progress

After the GCIP Project was conceived in 1990; its Science Plan was published in 1992; and its implementation Plan was completed in 1994. With primary funding from the NOAA's Office of Global Programs, early research activities were undertaken in 1993 in analysis of observational and model output data and in model development. Results of this research were published in a special issue of the Journal of Geophysical Research of the American Geophysical Union in March 1996.

A major phase of GCIP is the five-year Enhanced Observing Period that started on 1 October 1995 and is scheduled to be completed on 30 september 2000. This initiative is providing a comprehensive observational database needed for GCIP research and as a benchmark for future studies. GCIP research involves a systematic multiscale approach to accommodate physical process studies, model development, data assimilation, diagnostics, and validation topics. These research activities occur in a phased timetable and emphasize a particular region of the Mississippi River Basin with special characteristics for a period of about two years (see Figure A-1). Initial research emphasis has been on warm season processes using data from the Arkansas-Red River Basin in the southwestern part of the Mississippi River Basin. Cold season processes using data from the Upper Mississippi River Basin are being added to the ongoing research activities. These research activities, although initiated in limited regions, are leading toward an integrated continental-scale capability. Contributions to GCIP by the NASA Mission to Planet Earth Program are augmenting the level of GCIP research activities. Further details are available on the GCIP Home Page at the URL address: http://www.ogp.noaa.gov/gcip/

[LSAs]

Figure A-1 The Mississippi River basin with boundaries defining the Large Scale Areas (LSAs) for GCIP Focused Studies (top). Temporal emphasis for each LSA from 1994 through 2000 (bottom).

A rewriting of the GCIP Objectives by the NAS/NRC Gewex Panel in 1996 contributed to more focus of the GCIP research activities and near-term plans (IGPO 1996a):

The GEWEX Hydrometeorology Panel (GHP) was formed in 1995 and is the principal group within GEWEX for considering scientific issues associated with water cycle processes involved in the coupling of the atmosphere and the land surface, including the distribution of water and potential impacts on water resources. The main task of the GHP is to improve the collective contribution of the GEWEX Continental Scale Experiments (CSEs) and ensuring their regional results contribute to improvements in global scale prediction models. The CSEs , in addition to GCIP, consist of the Baltic Sea Experiment (BALTEX), the GEWEX Asian Monsoon Experiment (GAME), the Large-scale Biosphere-Atmosphere experiment in Amazonia (LBA), and the Mackenzie River Basin GEWEX Study (MAGS). The premise of the GHP is that the prediction of regional precipitation and runoff anomalies over period of several months is a possibility with improved understanding of water cycle processes. In this regard, the GHP will work toward the following scientific milestones:

GCIP and the Pan American Climate Studies(PACS) projects recently developed a prospectus as an initial step toward an integrated study of warm season predictability of precipitation and temperature over North America. It is predicated on the hypothesis that there is a deterministic element in the year-to-year variability of summertime precipitation and temperature over North America. The GCIP/PACS studies will address three major objectives:

The brief summary given above shows that the environment for GCIP research has changed significantly over the past five years since the GCIP Implementation Plan was written. Some changes were foreseen while others were not and the Preface to Volume I (IGPO 1993) states -- "These volumes of the implementation plan will evolve during the course of the project and each will be updated as required". It is apparent that updating should be done to accommodate the knowledge gained during the past five years. GCIP could benefit most from the updating of Volume II - RESEARCH- portion of the GCIP Implementation Plan. This proposal for a post-2000 implementation strategy is a first step in this process. It is also provides a framework for updating the GCIP Major Activities Plan for the period 1998, 1999 and Outlook for 2000.

A.2 Phased Objectives

GCIP is completing the first two research phases since the Science Plan was published in 1992. The Buildup Period from 1992 to 1994, which was largely devoted to implementation planning and the compilation of several initial data sets also included some early research studies which culminated in 25 papers published in a special issue of the Journal of Geophysical Research in March 1996. The Data Management Objective was emphasized in this early stage because it was recognized in the GCIP Science Plan (WMO 1992) that the success of the science objectives were heavily dependent on the availability of suitable data sets for the GCIP investigators.

The beginning of the five-year EOP in 1995 also initiated a three-year phase concentrated on budget studies and some early experience with coupled mesoscale NWP models (IGPO 1995). The early years of the Enhanced Observing Period through 1997 are emphasizing the water budgets of the Mississippi River basin and diagnostic studies of regional model output with emphasis on the variables needed to compute water budgets.

The results of the research during the past five years,especially the successes with the mesoscale NWP models show that GCIP can now increase the time scale for predictions and should focus on developing an initial version of a coupled hydrologic/atmospheric climate model. Also, GCIP needs to increase the priority of its efforts in water resources applications to provide a contribution to the strategic objective for the GEWEX Hydrometeorology Panel.

A composite assessment of the priorities for each of the five GCIP objectives (listed in Section A.1) is shown in Table A-1. The ranking for the objectives during phases 0 and I are based on the actual efforts expended during these six years. The rankings for Phases II, III and IV indicate the relative priorities that GCIP needs to give to each of the objectives assuming that the GHP strategic objective for 2005 is also a primary objective for GCIP. This assumption was used to design the implementation strategy described in the remainder of this paper. Consideration was also given to the GCIP/PACS prospectus described earlier.

[datasets]

Table A-1 GCIP Objectives Ranked in Project Phases.

The period from 1998 through the year 2000, identified as phase III in Table A-1 will focus on process and budget studies, some initial coupled modeling experiments and studies of precipitation predictability during the warm season. This phase of GCIP research will be entitled the Process, Early Coupled Modeling and Predictability studies phase or PRECOMP . The overall objective for the PRECOMP Phase is--- * By the end of the year 2000 demonstrate an initial version of a regional coupled hydrologic/atmospheric climate prediction model capable of carrying out prediction experiments up to annual time scales.

The period after the year 2000 through the year 2003 will focus on the applications of regional coupled models in climate prediction. Emphasis will be on Embedded Regional Modeling and Seasonal Prediction Experiments and will be given the shortened name of the ERMOSPE phase. Successful completion of the overall objective for the PRECOMP phase will enable GCIP to begin to implement a more complete climate prediction system focusing on water resources during the period 2001 through 2003. The overall objective during the ERMOSPE Phase is proposed as --- * By the end of the year 2003 demonstrate an initial climate prediction system capable of carrying out experiments to predict variabilities in water resources and soil moisture on time scales of seasonal to annual.

The last phase will include the years 2004 and 2005 and will concentrate on the applications of climate predictions in water resources management. Emphasis during this phase will be on Seasonal to Annual Water Resources Prediction EXperiments and will be labeled the SAWRPEX phase . It is envisioned that GCIP along with the other four CSEs (BALTEX, GAME, LBA and MAGS) will all be focused on demonstrating a capability and successful achievement of the GHP strategic objective --- * By the end of the year 2005 predict changes in water resources and soil moisture on time scales of seasonal to annual as an element of the World Climate Research Program's goals for the climate system.

A.3 Critical Activities in the PRECOMP Phase

An increasing emphasis is needed on regional climate modeling and predictability studies during the period 1998 through 2000 to provide the capability needed for the climate prediction experiments after the year 2000. Some of the critical activities during the PRECOMP Phase are summarized in terms of accomplishments needed to achieve the overall PRECOMP objective given in the previous section. A more specific description of the plans for all of the Principal Research Areas in GCIP is given in the GCIP Major Activities Plan for 1998, 1999 and Outlook for 2000.

A.3.1 Coupled Modeling

The function of the coupled modeling is to foster research which creates, calibrates and applies coupled models of the atmospheric and hydrologic system, with priority on research to improve prediction of weather and climate at time scales from days to seasons. The research focus is on determining , understanding and modeling those processes which are demonstrably important in coupling atmospheric and hydrological systems, rather than those processes which are separately important within these two systems.

A.3.1.1 Coupled Climate Model Research

The research activities in coupled climate modeling will focus on predictability studies for precipitation during different seasons in the annual cycle. These will be concerned with :

The exploratory work on seasonal to annual predictions needs to be emphasized and carried through to the extent that regional climate predictions can be produced in an on-line operational mode as well as off-line hindcast experiments.

A.3.1.2 Macroscale Land Surface/Hydrology Models

The research activities relevant to land surface schemes within the GEWEX Program have literally exploded during the past five years. The ISLSCP Workshop held at Columbia, MD in 1992 created the impetus for these later activities within the GCIP, ISLSCP and PILPS components of GEWEX. The PILPS Project is focusing on evaluating and improving land surface schemes for climate and weather prediction models and the results from the early phases have been published (Henderson-Sellers et al. 1995). ISLSCP has published an Initiative I global data set for land-atmosphere models formatted on a one by one degree grid and covering a one-year period in 1987-88 (Sellers et al. 1996). Support is now being sought for an ISLSCP Initiative II data set covering the period 1986 to 1995. GCIP is supporting research on land/hydrology models as well as compiling data sets for experiments with such models (IGPO 1996). Research is also underway in several of the other CSEs and is expected to become a major emphasis area in the next several years within the GHP.

The results from the GCIP, ISLSCP and PILPS activities during the past five years have enhanced the understanding of the performance of land/hydrology models and just as critically have made significant progress toward compiling more complete datasets for initialization/boundary conditions, forcing and validation of land/hydrology models.

GCIP needs to focus its efforts in this area on evaluating, selecting and implementing a macroscale land/hydrology model as a critical element of its PRECOMP objective of demonstrating an initial version of a regional coupled hydrologic/atmospheric climate prediction model . In particular, this emphasis needs to be placed on the land surface and hydrology components of a climate model which is focused on providing predictions at the seasonal to interannual time scale.

A.3.1.3 Regional Mesoscale Models

The implementation plan for GCIP(IGPO 1993) incorporated the operational numerical weather prediction models and associated 4DDA systems as fundamental systems for both the description and modeling of the energy and water cycles. GCIP is concentrating on three regional mesoscale models (IGPO 1995):

The regional mesoscale models are supporting GCIP research in the following manner:

The progress over the past four years in the demonstrated success by these three regional modeling efforts, especially with the initial land-physics components implemented in each of the models are enabling GCIP to accelerate its efforts in developing and testing an initial version of a coupled hydrologic-atmospheric climate model. It is anticipated that the three functions listed above can be essentially completed by the end of the PRECOMP phase. The requirements for model assimilated and forecast data products after the year 2000 have not yet been determined.

A.3.2 Water Resources

The overall objective in water resources is to improve the utility of hydrologic predictions for water resources management up to seasonal and interannual time scales. A recent coupled modeling workshop "... revealed a lack of understanding of how best to give seasonal-to-interannual predictions hydrological interpretation and a failure in communication between the atmospheric and hydrological communities on this issue." It was concluded that research is required to determine what type of hydrological prediction is possible from seasonal-to-interannual meteorological predictions and at what spatial and temporal scales hydrological interpretation can have worth while credibility and utility (IGPO 1996b).

The research activities in the PRECOMP phase will focus on the following:

It is especially critical that GCIP develop a consensus among interested investigators on an action plan for water resources in time for inclusion in the GCIP Major Activities Plan for 1998, 1999 and Outlook for 2000.

A.3.3 Data

A number of GCIP initial data sets (GIDS) were prepared to provide the data services support during the build-up period before the 5-year EOP. Preparation of the GIDS started in 1993, and the data sets were compiled for on-line access by GCIP investigators to the extent that is technically feasible. They were also packaged and distributed on CD-ROMs for wide distribution especially to international persons interested in performing initial diagnostic, evaluation and modeling studies on GCIP-related topics. The compiled and planned standard datasets for GCIP research are summarized in Figure A-2. Further details about each of these standard datasets are available (IGPO 1996a)

[datasets]

Figure A-2 Compiled and Planned Standard Data Sets for GCIP Research.

The major data collection and management activities during the PRECOMP phase are:

The current plans for data collection and management need to be reviewed and modified as needed in light of the developing plans for GCIP research after the year 2000.

A.4. Strategy and Guidelines for ERMOSPE Phase

The coupled modeling and prediction experiments for seasonal to annual time scales envisioned for the ERMOSPE phase from 2001 through 2003 will entail closely coordinated efforts among the five GEWEX CSEs and also closely coordinated/joint experiments with CLIVAR/GOALS and particularly with the PACS project.

The specific nature of these activities will depend on results achieved during the PRECOMP phase particularly with reference to the question - To what extent is meteorological prediction at daily to seasonal time scales sensitive to hydrologic- atmospheric coupling processes? They will also depend on the early result for the GCIP/PACS hypothesis that -- there is a deterministic element in the year-to-year variability of summertime precipitation and temperature over North America.

The activities during the ERMOSPE phase are described in the following section in terms of the GCIP objectives as was done earlier for the PRECOMP phase.

A.4.1 Coupled Hydrologic-Atmospheric Modeling

It is envisioned that the coupled modeling activities during the ERMOSPE phase will involve activities in the following areas:

A.4.1.1 General Approach for Climate Modeling and Prediction

The emphasis during the ERMOSPE phase will be on developing and carrying out regional modeling and prediction experiments with priority on the applications to water resources management. The costs for such experiments, especially for computer time and data sets needed, make it critical to lay out a GCIP strategy that is both affordable and provides sufficient opportunity for participation by interested investigators.

The implementation strategy will consist of a four-stage scenario for regional climate modeling and applications as summarized in Figure A-3:

[datasets]

Figure A-3 GCIP 4-Stage Scenario for Regional Climate Modeling and Applications.

The general approach will consist of on-line realtime predictions and off-line non-real time hindcasts and simulations. The former will be carried out primarily through NCEP by making use of both the operational regional climate model output and the parallel (non-operational) regional climate model output.

The off-line non-realtime hindcasts and simulations will consist of the following activity areas:

A.4.1.2 Infrastructure Guidelines

Figure A-4 provides a schematic of the infrastructure needed to support a regional NWP or Climate model. GCIP will make use of the following guidelines for infrastructure support during the ERMOSPE phase:

[datasets]

Figure A-4 Infrastructure for Regional Model (NWP or Climate).

A.4.1.3 Modular Land/Hydrology Model Infrastructure

A schematic depicting the role of land models coupled to atmospheric models is shown in Figure A-5 (Sellers et al. 1996). The current focus in GCIP is on the models used to calculate the exchanges of water and energy between the land surface and the atmosphere on different time scales from hours to seasons. These components represent unique portions of a completely coupled climate model which GCIP and the other CSEs in the Gewex Hydrometeorology Panel are being asked to contribute to the overall goal of the WCRP. The GCIP concentration in the Mississippi River basin, with its varied climatic regimes combined with a relatively data rich sets of observations and model assimilated data sets available provides an opportunity for GCIP to make a significant contribution to global climate modeling and prediction by developing the infrastructure needed to accelerate the research and development of a community land/hydrology model. The infrastructure needed for a potential GCIP - Modular Land/Hydrology Model Infrastructure is depicted schematically in Figure A-6. Setting up such an infrastructure in an appropriate institution will provide an opportunity for interested investigators from universities, national research laboratories and private research laboratories to readily contribute to such an accelerated research effort.

[datasets]

Figure A-5 Schematic for Land and Atmosphere Models.

[datasets]

Figure A-6 Infrastructure for Potential GCIP-Modular Land/Hydrology Models (GCIP-MLH).

The significant advantages for a GCIP- MLH will accrue for the activities outside the Mississippi River basin, especially in carrying out the GCIP role within the GEWEX Hydrometeorology Panel on transferability of results. In the context of GHP, transferability is defined as : "The demonstration that techniques (including models) developed in particular regions to account for critical water and energy cycles will adequately represent those of other regions without tuning when provided with appropriate initial conditions and background fields".

Setting up the infrastructure needs to consider this as a long-term effort that will extend beyond the life of the GCIP Project. The elements of Land Models shown in Figure A-5 (Carbon, Biogeochemistry and Terrestrial Ecology) are not yet quantified for inclusion in climate models in any meaningful way but are projected to be part of the next advances in coupled climate modeling.

A.4.2 Water Resources

Some initial indications point to the GCIP water resources activities such as the following:

It is anticipated that the activities will become better defined for water resources by the end of 1997 .

A.4.3 Data

It is envisioned that a few seasonal to annual prediction experiments can be conducted with the operational centers such as NCEP. However, it is assumed that the bulk of the coupled modeling research and prediction experiments will necessarily make use of historical data sets. A preliminary assessment indicates that most of the data requirements can be met by using the 25-year period from 1976 to 2000. This makes use of the model reanalysis from the period when the global models were first implemented for operations by the NOAA/NMC through the buildup period for The Global Weather Experiment (1979) and will provide the optimum global data set for the ERMOSPE phase.

A.5 Preparations for SAWRPEX Phase

It is envisioned that GCIP will be ready to conduct Seasonal to Annual Water Resources Prediction Experiments (SAWRPEX) by the year 2003. Achieving such a capability will entail some complex preparatory research and analysis during the period 1998 to 2003.

It was recognized during the implementation planning for GCIP that the task of predicting the consequences of climatic variability and change on regional hydrological and water resources is a formidable one (IGPO 1994). At some point the capability developed by GCIP to model water and energy cycle variability needs to be integrated into models that gauge societal impacts from climate variability and change. This testing must demonstrate consistency in local and regional as well as continental skill.

Presently, the uncertainty associated with estimates of how managed and uncontrolled watersheds may respond to a variety of climatic scenarios is enormous. What is called for, and what GCIP has the opportunity to provide, is an evaluative framework composed of methods and procedures for translating the output of climate models to a form appropriate and meaningful for use in water management models. Thus, GCIP water resource research activities during the next five years needs be organized around two distinct focuses:

A.5.1 Diagnostic Evaluations for Water Resources

Although the body of information dealing with the hydrological and water resource effects of climate variability and change is growing rapidly (especially case studies of hydrological impacts), few water resource policy and management insights have been produced in these efforts. This lack of intuitive knowledge is due largely to the reality that the analyses are based on GCM outputs that simply do not provide the information required for management and policy activities. However, a concerted effort at understanding the principal strengths as well as weaknesses of the different types of climate models vis-à-vis water resources assessment has never been undertaken. Clearly, GCIP research presents a unique opportunity for identifying ways to make climate model output, especially at the mesoscale, more useful as input to water resources simulations and to water resources management decision-making.

Initial studies will focus on the larger space and longer time scales and then work down to greater spatial and temporal detail. Thus, the first activity will be to assess the performance of operational models over the entire Mississippi River basin domain at an annual time step. Subsequent efforts will move to the LSAs and then to the smaller study areas, and down to seasonal and then monthly time steps. Questions of importance to water resources assessment and management to be addressed in these evaluations include:

A.5.2 Transfer of Research Results to Water Resource Managers

Another important consideration of the water management element relates to the presence of an incongruity between the research and modeling outputs of the GCIP effort and the practical problems (e.g., organizational constraints and liabilities) faced by water resources managers. This is not to say that GCIP information is not of potential significant benefit, but, rather, to note that such information cannot be directly transformed or transferred into water management activities without a lengthy process of institutional validation and without reconstituting it into a complex series of models, engineering regulations, manuals, and conventional practices. Thus, while the water management community may immediately see the long-range value of GCIP scientific information and products, that value may not be readily translated into operational payoffs. As a result, the practical benefits of the GCIP effort should be viewed in a long-term context.

To ensure that GCIP scientific research output is effectively evaluated and utilized in water management activities within the Mississippi River basin, a coordination function with the responsible regional and local water management organizations will be established and maintained.

As part of this liaison activity several distinct actions are planned. First, an implementation plan will be developed for incorporating GCIP products into each management organization's procedures. Second, the GCIP Water Resources PRA, or some working group thereof, will undertake the coordination of ongoing agency activities that complement GCIP objectives. Many organizations will not be funded specifically to conduct GCIP activities, but they do conduct operations that support GCIP objectives. A mechanism for incorporating these related activities is essential to the success of the water resource assessment component of the overall project. Third, most water management agencies also conduct research on the design of hydrological models. It is appropriate, therefore, to ensure that these modeling efforts, as well as the testing of existing models using GCIP data, are incorporated into the overall hydrological modeling component of the Mississippi River basin project.

There is also a need to describe clearly what operational improvements water resources managers should anticipate in what timeframe. In this regard, it is instructive to consider the water management perspective. Although a wide range of water models is available, two generic types may be considered: hydrological process models and water management models. The former type is designed to understand how changes in precipitation, the land surface, and soils affect runoff (discharge) and recharge. The latter type, which includes such schemes as simulation and optimization, tends to utilize statistical series of precipitation and streamflow and focus on resolving issues such as how to design, control, and distribute the water supply.

In general, process models are not directly used in water management operations, although certain features are incorporated for special problems. Rather, a number of specialized models have been developed for application to different decision needs, such as operations, planning, and design. Some examples of these models, along with the time scale at which they operate, appear in Figure A-7.

[LSAs]

Figure A-7 Models for Different Decision Needs in Water Resources Management.

Shorter term needs (days to several months), focused on operations, are depicted on the left side of the vertical dashed line in Figure A-7. Longer term needs (several months to more than 100 yr), aimed at planning and design, appear to the right of this line. The types of models that support operational issues include those designed to assist in decisions related to flood warning and evacuation, reservoir operation, water supply allocation, navigation, and the development of reservoir operating rules. At the longer time scales, planning and design-oriented models address issues such as interannual storage, safe yield, supply reliability, hydropower, drought, design flood, probable maximum flood, structural integrity, dam safety, and project lifespan. GCIP needs to focus on the establishment of an infrastructure for the improved management of water resources in the Mississippi River basin, across the range of decision needs, by ensuring that the gains in understanding hydrological processes and modeling techniques are linked with the appropriate water management organizations.