What is a Dynamic Information Framework (DIF)?

From "Actionable Information" to High-Level Decision Support

"Actionable Information:" Developing a Dynamic Information Framework for Decision Support.

Key to better decision making in the face of complex challenges is "actionable information" - " the synthesis and "bringing to life" of the key information that integrates the end-to-end knowledge required to provide the high-level decision support to make the most informed decisions.  The objective of the Dynamic Information Framework (DIF) is to proived such information, as the foundation for a high-level decision support, providing (quantitative) analyses of complex, interdependent environmental problems. The strategy is via a "geospatial" gateway for dynamic understanding, management, and planning of the landscape. The functional components of DIF of include data modules (base and directed data layers) focused on synthetic objectives, provided by computation engines (geospatially-explicit, process-based, cross-sector simulation models) facilitated input/output (including visualizations), and decision support system and scenario testing capabilities. The objective is to provide a "time machine," for multiple stakeholders.

The DIF is based on how water and the landscape (topography, soils, vegetation, biodiversity) converge, in space and time. The central thread is that water provides spatial, time-based, and operational connectivity among the multitude of DIF layers, because everyone understands water (one has it or not, it is of adequate quality or not, it is available in the right place at the right time or not). Most important it is observable, measureable, subject to being modeled,  as a function of known drivers and spatial-temporal relationships.

To meet these challenging criteria, the modeling effort draws on the emergence of a new generation of Earth System Science, based on the rapidly evolving capabilities for addressing global change issues. Earth System Science involves use of satellites, new generations of dynamic computer "models," field measurements focused by model requirements covering wide areas, and, especially, a thinking and practice of "integrated systems." Fundamental to these is a new class of open and publically accessible hydrology models, which can be regarded not only as hydrology models, but also as overall landscape models, because of the processes (and data layers) they represent. The requirements of the model dictate what data modules must be assembled, and the structure of the model allows the production of the output variables, which ultimately provide the DSS with the information to make decisions.

Establishing the process to actually execute such models is not a trivial process, for several reasons. The information required comes from multiple sources, from individual rain gauges to statistics on hydropower and grain yields, to glacier melting to rock types. The information required comes from multiple disciplines, which presents problems with even communication between specialists. Existing data holdings are not always readily obtainable, sometimes for institutional reasons, or have to be purchased. New field measurements, especially holistic and cross-boundary, are challenging. Handling such diverse data and executing models is not straightforward. There are very real problems in converting data streams into useful information that go beyond a database. Perhaps most challenging is how to not only create such information, but how to get it into the hands of users of different levels, from the specialist to the local and regional decision makers to the local farmer or fisherman. Hence it is necessary to be clear and explicit about exactly what information is required by each stakeholder.

The Computation Engine, and Data Resources

The Earth System model, as the core of the computation engine, could be provided by different geospatial hydrology models. For example, the Variable Infiltration Capacity (VIC), is a so-called semi-distributed grid-based mesoscale to macroscale hydrologic model (Liang et al., 1994, 1996, Nijssen et al., 1997, 2001a, b), which represents explicitly the effects of vegetation, topography, and soils on the exchange of moisture and solar energy between land and atmosphere.  The core VIC model can then be coupled to other models, and compared to independent data sources, to ultimately provide the basis for management-focused applications in the DSS.

The information required to support modeling must, of necessity, be derived from multiple sources. Even in very remote, data-sparse regions, global coverages can provide at least first-order estimates (think Google Earth). It is convenient to first think in terms of input data; i.e., those data required for running the model. The first type of input data are essentially static (don't change over the course of the model run). This includes the basic structure of the river basin (topography, river networks), soil properties (how deep are the soils, what is their texture), vegetation properties (how deep are the roots, how tall is it, leaf area index).

The second type of input data is climate forcing, which includes the daily average precipitation, minimum and maximum temperature, and winds. These data are more dynamic, as they change over the course of a model run, and essentially "drive" the model. These data can be derived from meteorological observation networks, climate weather models, or directly from satellite observations. Changing the climate forcing data, while maintaining the more structural data constant, allows testing of different climate scenarios.

The third type of data required for the modeling effort is actual observed discharge data. These data are used for model calibration  (to adjust the model to observed values at several points, as input data are rarely good enough to specify exactly how the basin is structured) and validation (to test the calibrated model, against observed data from a different time period than used for calibration).

The results of complex, multi-layer, 4-dimensional (including space, time) analyses of landscapes and their resources is difficult enough for the specialist to understand, never mind the non-specialist. Visualizations as the medium are part of the message. Hence substantial effort has gone into how to present spatial and temporal information to a broader audience, from students to decision makers to the general public. Examples of this philosophy represent the underpining of this website.

The Information Environment

To keep track of all of these elements, it is necessary to establish a "Dynamic Information Framework" (DIF, Figure 4), with the objective of providing a consistent theoretical basis, overall capability of integrating across sectors, and providing information using recent advances in cyberinformatics (including delivery of advanced visualizations, to enable a viewer to more readily assimilate the message being delivered). The DIF then provides the core for a DSS. The process of creating the DIF provides

  • An integration of data from multiple sources (of interest to all stakeholders)
  • Provides a means for interpolation of sparse data
  • Provides quantitative baselines and an instrument for analysis of interdependent problems

 ·       Provides the basis for cross- scale/ upscaling analyses
  • Provides a foundation for "scenarios"
  • Perhaps most importantly, the construction of a DIF promotes cooperation and communication between individuals and sectors that rarely, if ever, communicate.  

Lessons Learned and Future Directions  

Developing a fundamental understanding of how all sectors of the landscape are interdependent, then creating a decision-evaluation framework and institutional support structure that can be accessed by all stakeholders is a critical step in the process of developing viable strategies for adaptation to global changes for specific regions and countries. The DIF concepts represent advances towards providing such a framework that has become practical in today's world. It is not a shrink-wrapped, commercial application; rather, it is a process and framework that must be jointly developed and evolved for specific applications. The "environmental cyber informatics" driving it is not trivial; there is a tremendous amount of "computer" detail that goes into making a DIF functional. The capability can't be represented not only as research at major universities, but operationally at Ministries and universities and field camps and in the media in the developing world. The expectation is that the "cost-benefit" ratio of executing a DIF pays off many times over, given what is at stake.

Developing a fundamental understanding of (1) how all sectors of the landscape (from water movement to species production and distribution), are interdependent, then (2) creating a decision-making framework and institutional support structure that can be accessed by all stakeholders reflecting that understanding, is a critical step in the process of developing viable strategies for adaptation to global changes. The DIF concepts represent advances towards providing such a framework that has become practical in today's world. The capabilities now being provided though "Earth System Sciences," with its utilization of geospatial information from satellites combined with ground measurements, internet-accessible databases, and dynamic process-based models provides tools that are simply of a new generation. The capabilities for advanced visualization not only make it easier for the advanced practitioner to understand his/her own results, but to convey them to a much broader audience, including decision makers. 

     

Highly promising prototypes have been developed. But they need to be taken to the next level, of sophistication and application. What is needed to take DIF to next level and how can this is done?

      The environmental cyber informatics" driving the DIF is pushing the limit.  There is a tremendous amount of "computer" detail that goes into making a DIF functional. A critical component of this endeavor is that the capability is represented not only as research at major universities of the U.S. and Europe, but operationally at Ministries and universities and field camps and in the media in the developing world. This presents important challenges.

     

To address the issues on the table, the component models are being pushed to the limits of what they were designed for. Streaming information from multiple information sources (satellites, weather records and operational climate models, soil profiles, stream gauges, species lists and attributes), then ingesting that information into the models is hard work. Producing compelling visualizations and interactive scenario generation capabilities to the non-specialist, and to make that available through a web-portal, is challenging.   

The ability to convey the DIF distributed data, simulation platforms, and products through the "Cloud" could greatly enable the application.

The most sophisticated, yet usable, models must be deployed. Multiple earth-system models being developed around the world today, each with its own attributes (and liabilities). The objective of the DIF isn't to a priori develop component models, but to provide a framework for and to harness the power of the most suitable models for the tasks at hand. Of particular relevance is being able to improve and couple models of:

·        

Daily to seasonal weather (especially for operational models). If Hydromet and Druk Green Power Corporation of Bhutan can do a better of job of predicting the weather of the next few days, they can do a better job of predicting how much power they can commit to sell to India, while keeping track of flooding risks. At planting time, a farmer along the Zambezi would love to have an idea of what the weather might be in three weeks, so he would know whether it is safe to plant the higher risk/higher value crop. But today's ability to do that is limited.

Climate, and climate impact, scenarios. This is obviously an area of major focus, throughout the world, for the decades out time horizons (which is on the time scale for major infrastructure investments). But to be especially relevant to a particular region, how to downscale, and to ingest the results of model ensembles for those regions is necessary. On shorter time scales, if the tea farmer of Rwanda has an idea of climate shifts three to four years out, she can produce cultivars responsive to those changes.

Regional water balances and hydropower. The degree to which the water resources of many regions are not well-defined is stunning. The partitioning between surface and groundwater, and groundwater depletion through irrigation, is not even a blackbox; it is a virtual unknown. The Aral Sea has shrunk so much because the original  "planners" had little idea of how much water (aka precipitation) was available (the region still has little idea). Appropriation of surface waters between domestic use, commercial use, agriculture, and hydropower is one of the most profound of immediate environmental challenges. Current regional-scale hydropower analysis models (capable of computing the cumulative effects of, say, the upper and lower Mekong dams) are surprisingly primitive. The current era of "Hydropower Renaissance" has taken on new meaning, in the post-Japan tsunami world. In an even broader context, the ability to predict water and temperature distributions impacts human health concerns.

          

Terrestrial carbon and agricultural productivity. A fundamental issue is, of course, how much carbon is sequestered and released across the landscape, as function of what activities. Agriculture in much of the developing world is still quite primitive. Models that can relate productivity of specific crops to soil fertility and soil infiltration, the relative merits of different fertilizers, and to (changing) water availability and temperature would enable adaptation including in the short term).

The (physical) structure of the landscape and biodiversity. An important issue underlying biodiversity is, why are the species where they are? History? Current conditions? An interesting aspect brought by the DIF construct is to be able to "map" the biophysical world, from topography to soils to water,  at a higher level than is traditional done, in many regions. It might be provocative to explore the "biophysical basis of biodiversity."

  The application of the DIF construct to a range of data-rich environments would advance the development of how to analyze complex regions. The range of pilot applications to date has been quite limited, but has certainly developed the capability to move more rapidly to new regions. Potential target areas might include the Amazon basin (including linking the Andes-Amazon and the tropical river-to-ocean continuum projects) and the Pacific Northwest, building from the Puget Sound and salmon initiatives). Southeast Asia, mainland and insular, is ripe with possibilities - and needs.

     

Essentially, a DIF is a numeric and quantitative "Commons," or meeting place, which builds on the legacy of knowledge from experience, with the goal of "harmonizing" region function for multiple users. The best way to develop something is to actually do it.